Merge branch 'master' into worksplit-multigpu

comfy_extras/nodes_audio.py

@@ -142,9 +142,10 @@ def save_audio(self, audio, filename_prefix="ComfyUI", format="flac", prompt=Non
         for key, value in metadata.items():
             output_container.metadata[key] = value
 
+        layout = 'mono' if waveform.shape[0] == 1 else 'stereo'
         # Set up the output stream with appropriate properties
         if format == "opus":
-            out_stream = output_container.add_stream("libopus", rate=sample_rate)
+            out_stream = output_container.add_stream("libopus", rate=sample_rate, layout=layout)
             if quality == "64k":
                 out_stream.bit_rate = 64000
             elif quality == "96k":
@@ -156,7 +157,7 @@ def save_audio(self, audio, filename_prefix="ComfyUI", format="flac", prompt=Non
             elif quality == "320k":
                 out_stream.bit_rate = 320000
         elif format == "mp3":
-            out_stream = output_container.add_stream("libmp3lame", rate=sample_rate)
+            out_stream = output_container.add_stream("libmp3lame", rate=sample_rate, layout=layout)
             if quality == "V0":
                 #TODO i would really love to support V3 and V5 but there doesn't seem to be a way to set the qscale level, the property below is a bool
                 out_stream.codec_context.qscale = 1
@@ -165,9 +166,9 @@ def save_audio(self, audio, filename_prefix="ComfyUI", format="flac", prompt=Non
             elif quality == "320k":
                 out_stream.bit_rate = 320000
         else: #format == "flac":
-            out_stream = output_container.add_stream("flac", rate=sample_rate)
+            out_stream = output_container.add_stream("flac", rate=sample_rate, layout=layout)
 
-        frame = av.AudioFrame.from_ndarray(waveform.movedim(0, 1).reshape(1, -1).float().numpy(), format='flt', layout='mono' if waveform.shape[0] == 1 else 'stereo')
+        frame = av.AudioFrame.from_ndarray(waveform.movedim(0, 1).reshape(1, -1).float().numpy(), format='flt', layout=layout)
         frame.sample_rate = sample_rate
         frame.pts = 0
         output_container.mux(out_stream.encode(frame))
@@ -360,7 +361,7 @@ class RecordAudio:
     def load(self, audio):
         audio_path = folder_paths.get_annotated_filepath(audio)
 
-        waveform, sample_rate = torchaudio.load(audio_path)
+        waveform, sample_rate = load(audio_path)
         audio = {"waveform": waveform.unsqueeze(0), "sample_rate": sample_rate}
         return (audio, )
 

comfy_extras/nodes_audio_encoder.py

@@ -1,44 +1,62 @@
 import folder_paths
 import comfy.audio_encoders.audio_encoders
 import comfy.utils
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 
-class AudioEncoderLoader:
+class AudioEncoderLoader(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "audio_encoder_name": (folder_paths.get_filename_list("audio_encoders"), ),
-                             }}
-    RETURN_TYPES = ("AUDIO_ENCODER",)
-    FUNCTION = "load_model"
+    def define_schema(cls) -> io.Schema:
+        return io.Schema(
+            node_id="AudioEncoderLoader",
+            category="loaders",
+            inputs=[
+                io.Combo.Input(
+                    "audio_encoder_name",
+                    options=folder_paths.get_filename_list("audio_encoders"),
+                ),
+            ],
+            outputs=[io.AudioEncoder.Output()],
+        )
 
-    CATEGORY = "loaders"
-
-    def load_model(self, audio_encoder_name):
+    @classmethod
+    def execute(cls, audio_encoder_name) -> io.NodeOutput:
         audio_encoder_name = folder_paths.get_full_path_or_raise("audio_encoders", audio_encoder_name)
         sd = comfy.utils.load_torch_file(audio_encoder_name, safe_load=True)
         audio_encoder = comfy.audio_encoders.audio_encoders.load_audio_encoder_from_sd(sd)
         if audio_encoder is None:
             raise RuntimeError("ERROR: audio encoder file is invalid and does not contain a valid model.")
-        return (audio_encoder,)
+        return io.NodeOutput(audio_encoder)
 
 
-class AudioEncoderEncode:
+class AudioEncoderEncode(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "audio_encoder": ("AUDIO_ENCODER",),
-                              "audio": ("AUDIO",),
-                             }}
-    RETURN_TYPES = ("AUDIO_ENCODER_OUTPUT",)
-    FUNCTION = "encode"
+    def define_schema(cls) -> io.Schema:
+        return io.Schema(
+            node_id="AudioEncoderEncode",
+            category="conditioning",
+            inputs=[
+                io.AudioEncoder.Input("audio_encoder"),
+                io.Audio.Input("audio"),
+            ],
+            outputs=[io.AudioEncoderOutput.Output()],
+        )
 
-    CATEGORY = "conditioning"
-
-    def encode(self, audio_encoder, audio):
+    @classmethod
+    def execute(cls, audio_encoder, audio) -> io.NodeOutput:
         output = audio_encoder.encode_audio(audio["waveform"], audio["sample_rate"])
-        return (output,)
+        return io.NodeOutput(output)
 
 
-NODE_CLASS_MAPPINGS = {
-    "AudioEncoderLoader": AudioEncoderLoader,
-    "AudioEncoderEncode": AudioEncoderEncode,
-}
+class AudioEncoder(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            AudioEncoderLoader,
+            AudioEncoderEncode,
+        ]
+
+
+async def comfy_entrypoint() -> AudioEncoder:
+    return AudioEncoder()

comfy_extras/nodes_clip_sdxl.py

@@ -1,43 +1,52 @@
-from nodes import MAX_RESOLUTION
+from typing_extensions import override
 
-class CLIPTextEncodeSDXLRefiner:
+import nodes
+from comfy_api.latest import ComfyExtension, io
+
+
+class CLIPTextEncodeSDXLRefiner(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "ascore": ("FLOAT", {"default": 6.0, "min": 0.0, "max": 1000.0, "step": 0.01}),
-            "width": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "height": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "text": ("STRING", {"multiline": True, "dynamicPrompts": True}), "clip": ("CLIP", ),
-            }}
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodeSDXLRefiner",
+            category="advanced/conditioning",
+            inputs=[
+                io.Float.Input("ascore", default=6.0, min=0.0, max=1000.0, step=0.01),
+                io.Int.Input("width", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                io.Int.Input("height", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                io.String.Input("text", multiline=True, dynamic_prompts=True),
+                io.Clip.Input("clip"),
+            ],
+            outputs=[io.Conditioning.Output()],
+        )
 
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, ascore, width, height, text):
+    @classmethod
+    def execute(cls, clip, ascore, width, height, text) -> io.NodeOutput:
         tokens = clip.tokenize(text)
-        return (clip.encode_from_tokens_scheduled(tokens, add_dict={"aesthetic_score": ascore, "width": width, "height": height}), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens, add_dict={"aesthetic_score": ascore, "width": width, "height": height}))
 
-class CLIPTextEncodeSDXL:
+class CLIPTextEncodeSDXL(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "width": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "height": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "crop_w": ("INT", {"default": 0, "min": 0, "max": MAX_RESOLUTION}),
-            "crop_h": ("INT", {"default": 0, "min": 0, "max": MAX_RESOLUTION}),
-            "target_width": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "target_height": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "text_g": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "text_l": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            }}
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodeSDXL",
+            category="advanced/conditioning",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.Int.Input("width", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                io.Int.Input("height", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                io.Int.Input("crop_w", default=0, min=0, max=nodes.MAX_RESOLUTION),
+                io.Int.Input("crop_h", default=0, min=0, max=nodes.MAX_RESOLUTION),
+                io.Int.Input("target_width", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                io.Int.Input("target_height", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                io.String.Input("text_g", multiline=True, dynamic_prompts=True),
+                io.String.Input("text_l", multiline=True, dynamic_prompts=True),
+            ],
+            outputs=[io.Conditioning.Output()],
+        )
 
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, width, height, crop_w, crop_h, target_width, target_height, text_g, text_l):
+    @classmethod
+    def execute(cls, clip, width, height, crop_w, crop_h, target_width, target_height, text_g, text_l) -> io.NodeOutput:
         tokens = clip.tokenize(text_g)
         tokens["l"] = clip.tokenize(text_l)["l"]
         if len(tokens["l"]) != len(tokens["g"]):
@@ -46,9 +55,17 @@ class CLIPTextEncodeSDXL:
                 tokens["l"] += empty["l"]
             while len(tokens["l"]) > len(tokens["g"]):
                 tokens["g"] += empty["g"]
-        return (clip.encode_from_tokens_scheduled(tokens, add_dict={"width": width, "height": height, "crop_w": crop_w, "crop_h": crop_h, "target_width": target_width, "target_height": target_height}), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens, add_dict={"width": width, "height": height, "crop_w": crop_w, "crop_h": crop_h, "target_width": target_width, "target_height": target_height}))
 
-NODE_CLASS_MAPPINGS = {
-    "CLIPTextEncodeSDXLRefiner": CLIPTextEncodeSDXLRefiner,
-    "CLIPTextEncodeSDXL": CLIPTextEncodeSDXL,
-}
+
+class ClipSdxlExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            CLIPTextEncodeSDXLRefiner,
+            CLIPTextEncodeSDXL,
+        ]
+
+
+async def comfy_entrypoint() -> ClipSdxlExtension:
+    return ClipSdxlExtension()

comfy_extras/nodes_compositing.py

@@ -1,6 +1,9 @@
 import torch
 import comfy.utils
 from enum import Enum
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
+
 
 def resize_mask(mask, shape):
     return torch.nn.functional.interpolate(mask.reshape((-1, 1, mask.shape[-2], mask.shape[-1])), size=(shape[0], shape[1]), mode="bilinear").squeeze(1)
@@ -101,24 +104,28 @@ def porter_duff_composite(src_image: torch.Tensor, src_alpha: torch.Tensor, dst_
     return out_image, out_alpha
 
 
-class PorterDuffImageComposite:
+class PorterDuffImageComposite(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "source": ("IMAGE",),
-                "source_alpha": ("MASK",),
-                "destination": ("IMAGE",),
-                "destination_alpha": ("MASK",),
-                "mode": ([mode.name for mode in PorterDuffMode], {"default": PorterDuffMode.DST.name}),
-            },
-        }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="PorterDuffImageComposite",
+            display_name="Porter-Duff Image Composite",
+            category="mask/compositing",
+            inputs=[
+                io.Image.Input("source"),
+                io.Mask.Input("source_alpha"),
+                io.Image.Input("destination"),
+                io.Mask.Input("destination_alpha"),
+                io.Combo.Input("mode", options=[mode.name for mode in PorterDuffMode], default=PorterDuffMode.DST.name),
+            ],
+            outputs=[
+                io.Image.Output(),
+                io.Mask.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("IMAGE", "MASK")
-    FUNCTION = "composite"
-    CATEGORY = "mask/compositing"
-
-    def composite(self, source: torch.Tensor, source_alpha: torch.Tensor, destination: torch.Tensor, destination_alpha: torch.Tensor, mode):
+    @classmethod
+    def execute(cls, source: torch.Tensor, source_alpha: torch.Tensor, destination: torch.Tensor, destination_alpha: torch.Tensor, mode) -> io.NodeOutput:
         batch_size = min(len(source), len(source_alpha), len(destination), len(destination_alpha))
         out_images = []
         out_alphas = []
@@ -150,45 +157,48 @@ class PorterDuffImageComposite:
             out_images.append(out_image)
             out_alphas.append(out_alpha.squeeze(2))
 
-        result = (torch.stack(out_images), torch.stack(out_alphas))
-        return result
+        return io.NodeOutput(torch.stack(out_images), torch.stack(out_alphas))
 
 
-class SplitImageWithAlpha:
+class SplitImageWithAlpha(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-                "required": {
-                    "image": ("IMAGE",),
-                }
-        }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="SplitImageWithAlpha",
+            display_name="Split Image with Alpha",
+            category="mask/compositing",
+            inputs=[
+                io.Image.Input("image"),
+            ],
+            outputs=[
+                io.Image.Output(),
+                io.Mask.Output(),
+            ],
+        )
 
-    CATEGORY = "mask/compositing"
-    RETURN_TYPES = ("IMAGE", "MASK")
-    FUNCTION = "split_image_with_alpha"
-
-    def split_image_with_alpha(self, image: torch.Tensor):
+    @classmethod
+    def execute(cls, image: torch.Tensor) -> io.NodeOutput:
         out_images = [i[:,:,:3] for i in image]
         out_alphas = [i[:,:,3] if i.shape[2] > 3 else torch.ones_like(i[:,:,0]) for i in image]
-        result = (torch.stack(out_images), 1.0 - torch.stack(out_alphas))
-        return result
+        return io.NodeOutput(torch.stack(out_images), 1.0 - torch.stack(out_alphas))
 
 
-class JoinImageWithAlpha:
+class JoinImageWithAlpha(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-                "required": {
-                    "image": ("IMAGE",),
-                    "alpha": ("MASK",),
-                }
-        }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="JoinImageWithAlpha",
+            display_name="Join Image with Alpha",
+            category="mask/compositing",
+            inputs=[
+                io.Image.Input("image"),
+                io.Mask.Input("alpha"),
+            ],
+            outputs=[io.Image.Output()],
+        )
 
-    CATEGORY = "mask/compositing"
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "join_image_with_alpha"
-
-    def join_image_with_alpha(self, image: torch.Tensor, alpha: torch.Tensor):
+    @classmethod
+    def execute(cls, image: torch.Tensor, alpha: torch.Tensor) -> io.NodeOutput:
         batch_size = min(len(image), len(alpha))
         out_images = []
 
@@ -196,19 +206,18 @@ class JoinImageWithAlpha:
         for i in range(batch_size):
            out_images.append(torch.cat((image[i][:,:,:3], alpha[i].unsqueeze(2)), dim=2))
 
-        result = (torch.stack(out_images),)
-        return result
+        return io.NodeOutput(torch.stack(out_images))
 
 
-NODE_CLASS_MAPPINGS = {
-    "PorterDuffImageComposite": PorterDuffImageComposite,
-    "SplitImageWithAlpha": SplitImageWithAlpha,
-    "JoinImageWithAlpha": JoinImageWithAlpha,
-}
+class CompositingExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            PorterDuffImageComposite,
+            SplitImageWithAlpha,
+            JoinImageWithAlpha,
+        ]
 
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "PorterDuffImageComposite": "Porter-Duff Image Composite",
-    "SplitImageWithAlpha": "Split Image with Alpha",
-    "JoinImageWithAlpha": "Join Image with Alpha",
-}
+async def comfy_entrypoint() -> CompositingExtension:
+    return CompositingExtension()

comfy_extras/nodes_differential_diffusion.py

@@ -1,34 +1,41 @@
 # code adapted from https://github.com/exx8/differential-diffusion
 
+from typing_extensions import override
+
 import torch
+from comfy_api.latest import ComfyExtension, io
 
-class DifferentialDiffusion():
+
+class DifferentialDiffusion(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "model": ("MODEL", ),
-            },
-            "optional": {
-                "strength": ("FLOAT", {
-                    "default": 1.0,
-                    "min": 0.0,
-                    "max": 1.0,
-                    "step": 0.01,
-                }),
-            }
-        }
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "apply"
-    CATEGORY = "_for_testing"
-    INIT = False
+    def define_schema(cls):
+        return io.Schema(
+            node_id="DifferentialDiffusion",
+            display_name="Differential Diffusion",
+            category="_for_testing",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input(
+                    "strength",
+                    default=1.0,
+                    min=0.0,
+                    max=1.0,
+                    step=0.01,
+                    optional=True,
+                ),
+            ],
+            outputs=[io.Model.Output()],
+            is_experimental=True,
+        )
 
-    def apply(self, model, strength=1.0):
+    @classmethod
+    def execute(cls, model, strength=1.0) -> io.NodeOutput:
         model = model.clone()
-        model.set_model_denoise_mask_function(lambda *args, **kwargs: self.forward(*args, **kwargs, strength=strength))
-        return (model, )
+        model.set_model_denoise_mask_function(lambda *args, **kwargs: cls.forward(*args, **kwargs, strength=strength))
+        return io.NodeOutput(model)
 
-    def forward(self, sigma: torch.Tensor, denoise_mask: torch.Tensor, extra_options: dict, strength: float):
+    @classmethod
+    def forward(cls, sigma: torch.Tensor, denoise_mask: torch.Tensor, extra_options: dict, strength: float):
         model = extra_options["model"]
         step_sigmas = extra_options["sigmas"]
         sigma_to = model.inner_model.model_sampling.sigma_min
@@ -53,9 +60,13 @@ class DifferentialDiffusion():
             return binary_mask
 
 
-NODE_CLASS_MAPPINGS = {
-    "DifferentialDiffusion": DifferentialDiffusion,
-}
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "DifferentialDiffusion": "Differential Diffusion",
-}
+class DifferentialDiffusionExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            DifferentialDiffusion,
+        ]
+
+
+async def comfy_entrypoint() -> DifferentialDiffusionExtension:
+    return DifferentialDiffusionExtension()

comfy_extras/nodes_edit_model.py

@@ -1,26 +1,38 @@
 import node_helpers
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 
-class ReferenceLatent:
+class ReferenceLatent(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"conditioning": ("CONDITIONING", ),
-                            },
-                "optional": {"latent": ("LATENT", ),}
-               }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ReferenceLatent",
+            category="advanced/conditioning/edit_models",
+            description="This node sets the guiding latent for an edit model. If the model supports it you can chain multiple to set multiple reference images.",
+            inputs=[
+                io.Conditioning.Input("conditioning"),
+                io.Latent.Input("latent", optional=True),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ]
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "append"
-
-    CATEGORY = "advanced/conditioning/edit_models"
-    DESCRIPTION = "This node sets the guiding latent for an edit model. If the model supports it you can chain multiple to set multiple reference images."
-
-    def append(self, conditioning, latent=None):
+    @classmethod
+    def execute(cls, conditioning, latent=None) -> io.NodeOutput:
         if latent is not None:
             conditioning = node_helpers.conditioning_set_values(conditioning, {"reference_latents": [latent["samples"]]}, append=True)
-        return (conditioning, )
+        return io.NodeOutput(conditioning)
 
 
-NODE_CLASS_MAPPINGS = {
-    "ReferenceLatent": ReferenceLatent,
-}
+class EditModelExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            ReferenceLatent,
+        ]
+
+
+def comfy_entrypoint() -> EditModelExtension:
+    return EditModelExtension()

comfy_extras/nodes_eps.py

@@ -0,0 +1,74 @@
+from typing_extensions import override
+
+from comfy_api.latest import ComfyExtension, io
+
+
+class EpsilonScaling(io.ComfyNode):
+    """
+    Implements the Epsilon Scaling method from 'Elucidating the Exposure Bias in Diffusion Models'
+    (https://arxiv.org/abs/2308.15321v6).
+
+    This method mitigates exposure bias by scaling the predicted noise during sampling,
+    which can significantly improve sample quality. This implementation uses the "uniform schedule"
+    recommended by the paper for its practicality and effectiveness.
+    """
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="Epsilon Scaling",
+            category="model_patches/unet",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input(
+                    "scaling_factor",
+                    default=1.005,
+                    min=0.5,
+                    max=1.5,
+                    step=0.001,
+                    display_mode=io.NumberDisplay.number,
+                ),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
+
+    @classmethod
+    def execute(cls, model, scaling_factor) -> io.NodeOutput:
+        # Prevent division by zero, though the UI's min value should prevent this.
+        if scaling_factor == 0:
+            scaling_factor = 1e-9
+
+        def epsilon_scaling_function(args):
+            """
+            This function is applied after the CFG guidance has been calculated.
+            It recalculates the denoised latent by scaling the predicted noise.
+            """
+            denoised = args["denoised"]
+            x = args["input"]
+
+            noise_pred = x - denoised
+
+            scaled_noise_pred = noise_pred / scaling_factor
+
+            new_denoised = x - scaled_noise_pred
+
+            return new_denoised
+
+        # Clone the model patcher to avoid modifying the original model in place
+        model_clone = model.clone()
+
+        model_clone.set_model_sampler_post_cfg_function(epsilon_scaling_function)
+
+        return io.NodeOutput(model_clone)
+
+
+class EpsilonScalingExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            EpsilonScaling,
+        ]
+
+async def comfy_entrypoint() -> EpsilonScalingExtension:
+    return EpsilonScalingExtension()

comfy_extras/nodes_flux.py

@@ -1,60 +1,80 @@
 import node_helpers
 import comfy.utils
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
-class CLIPTextEncodeFlux:
+
+class CLIPTextEncodeFlux(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "clip_l": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "t5xxl": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "guidance": ("FLOAT", {"default": 3.5, "min": 0.0, "max": 100.0, "step": 0.1}),
-            }}
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodeFlux",
+            category="advanced/conditioning/flux",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.String.Input("clip_l", multiline=True, dynamic_prompts=True),
+                io.String.Input("t5xxl", multiline=True, dynamic_prompts=True),
+                io.Float.Input("guidance", default=3.5, min=0.0, max=100.0, step=0.1),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    CATEGORY = "advanced/conditioning/flux"
-
-    def encode(self, clip, clip_l, t5xxl, guidance):
+    @classmethod
+    def execute(cls, clip, clip_l, t5xxl, guidance) -> io.NodeOutput:
         tokens = clip.tokenize(clip_l)
         tokens["t5xxl"] = clip.tokenize(t5xxl)["t5xxl"]
 
-        return (clip.encode_from_tokens_scheduled(tokens, add_dict={"guidance": guidance}), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens, add_dict={"guidance": guidance}))
 
-class FluxGuidance:
+    encode = execute  # TODO: remove
+
+
+class FluxGuidance(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "conditioning": ("CONDITIONING", ),
-            "guidance": ("FLOAT", {"default": 3.5, "min": 0.0, "max": 100.0, "step": 0.1}),
-            }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="FluxGuidance",
+            category="advanced/conditioning/flux",
+            inputs=[
+                io.Conditioning.Input("conditioning"),
+                io.Float.Input("guidance", default=3.5, min=0.0, max=100.0, step=0.1),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "append"
-
-    CATEGORY = "advanced/conditioning/flux"
-
-    def append(self, conditioning, guidance):
+    @classmethod
+    def execute(cls, conditioning, guidance) -> io.NodeOutput:
         c = node_helpers.conditioning_set_values(conditioning, {"guidance": guidance})
-        return (c, )
+        return io.NodeOutput(c)
+
+    append = execute  # TODO: remove
 
 
-class FluxDisableGuidance:
+class FluxDisableGuidance(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "conditioning": ("CONDITIONING", ),
-            }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="FluxDisableGuidance",
+            category="advanced/conditioning/flux",
+            description="This node completely disables the guidance embed on Flux and Flux like models",
+            inputs=[
+                io.Conditioning.Input("conditioning"),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "append"
-
-    CATEGORY = "advanced/conditioning/flux"
-    DESCRIPTION = "This node completely disables the guidance embed on Flux and Flux like models"
-
-    def append(self, conditioning):
+    @classmethod
+    def execute(cls, conditioning) -> io.NodeOutput:
         c = node_helpers.conditioning_set_values(conditioning, {"guidance": None})
-        return (c, )
+        return io.NodeOutput(c)
+
+    append = execute  # TODO: remove
 
 
 PREFERED_KONTEXT_RESOLUTIONS = [
@@ -78,52 +98,73 @@ PREFERED_KONTEXT_RESOLUTIONS = [
 ]
 
 
-class FluxKontextImageScale:
+class FluxKontextImageScale(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"image": ("IMAGE", ),
-                            },
-               }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="FluxKontextImageScale",
+            category="advanced/conditioning/flux",
+            description="This node resizes the image to one that is more optimal for flux kontext.",
+            inputs=[
+                io.Image.Input("image"),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "scale"
-
-    CATEGORY = "advanced/conditioning/flux"
-    DESCRIPTION = "This node resizes the image to one that is more optimal for flux kontext."
-
-    def scale(self, image):
+    @classmethod
+    def execute(cls, image) -> io.NodeOutput:
         width = image.shape[2]
         height = image.shape[1]
         aspect_ratio = width / height
         _, width, height = min((abs(aspect_ratio - w / h), w, h) for w, h in PREFERED_KONTEXT_RESOLUTIONS)
         image = comfy.utils.common_upscale(image.movedim(-1, 1), width, height, "lanczos", "center").movedim(1, -1)
-        return (image, )
+        return io.NodeOutput(image)
+
+    scale = execute  # TODO: remove
 
 
-class FluxKontextMultiReferenceLatentMethod:
+class FluxKontextMultiReferenceLatentMethod(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "conditioning": ("CONDITIONING", ),
-            "reference_latents_method": (("offset", "index", "uxo/uno"), ),
-            }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="FluxKontextMultiReferenceLatentMethod",
+            category="advanced/conditioning/flux",
+            inputs=[
+                io.Conditioning.Input("conditioning"),
+                io.Combo.Input(
+                    "reference_latents_method",
+                    options=["offset", "index", "uxo/uno"],
+                ),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+            is_experimental=True,
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "append"
-    EXPERIMENTAL = True
-
-    CATEGORY = "advanced/conditioning/flux"
-
-    def append(self, conditioning, reference_latents_method):
+    @classmethod
+    def execute(cls, conditioning, reference_latents_method) -> io.NodeOutput:
         if "uxo" in reference_latents_method or "uso" in reference_latents_method:
             reference_latents_method = "uxo"
         c = node_helpers.conditioning_set_values(conditioning, {"reference_latents_method": reference_latents_method})
-        return (c, )
+        return io.NodeOutput(c)
 
-NODE_CLASS_MAPPINGS = {
-    "CLIPTextEncodeFlux": CLIPTextEncodeFlux,
-    "FluxGuidance": FluxGuidance,
-    "FluxDisableGuidance": FluxDisableGuidance,
-    "FluxKontextImageScale": FluxKontextImageScale,
-    "FluxKontextMultiReferenceLatentMethod": FluxKontextMultiReferenceLatentMethod,
-}
+    append = execute  # TODO: remove
+
+
+class FluxExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            CLIPTextEncodeFlux,
+            FluxGuidance,
+            FluxDisableGuidance,
+            FluxKontextImageScale,
+            FluxKontextMultiReferenceLatentMethod,
+        ]
+
+
+async def comfy_entrypoint() -> FluxExtension:
+    return FluxExtension()

comfy_extras/nodes_fresca.py

@@ -1,6 +1,8 @@
 # Code based on https://github.com/WikiChao/FreSca (MIT License)
 import torch
 import torch.fft as fft
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 
 def Fourier_filter(x, scale_low=1.0, scale_high=1.5, freq_cutoff=20):
@@ -51,25 +53,31 @@ def Fourier_filter(x, scale_low=1.0, scale_high=1.5, freq_cutoff=20):
     return x_filtered
 
 
-class FreSca:
+class FreSca(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "model": ("MODEL",),
-                "scale_low": ("FLOAT", {"default": 1.0, "min": 0, "max": 10, "step": 0.01,
-                                        "tooltip": "Scaling factor for low-frequency components"}),
-                "scale_high": ("FLOAT", {"default": 1.25, "min": 0, "max": 10, "step": 0.01,
-                                        "tooltip": "Scaling factor for high-frequency components"}),
-                "freq_cutoff": ("INT", {"default": 20, "min": 1, "max": 10000, "step": 1,
-                                        "tooltip": "Number of frequency indices around center to consider as low-frequency"}),
-            }
-        }
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
-    CATEGORY = "_for_testing"
-    DESCRIPTION = "Applies frequency-dependent scaling to the guidance"
-    def patch(self, model, scale_low, scale_high, freq_cutoff):
+    def define_schema(cls):
+        return io.Schema(
+            node_id="FreSca",
+            display_name="FreSca",
+            category="_for_testing",
+            description="Applies frequency-dependent scaling to the guidance",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input("scale_low", default=1.0, min=0, max=10, step=0.01,
+                               tooltip="Scaling factor for low-frequency components"),
+                io.Float.Input("scale_high", default=1.25, min=0, max=10, step=0.01,
+                               tooltip="Scaling factor for high-frequency components"),
+                io.Int.Input("freq_cutoff", default=20, min=1, max=10000, step=1,
+                             tooltip="Number of frequency indices around center to consider as low-frequency"),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+            is_experimental=True,
+        )
+
+    @classmethod
+    def execute(cls, model, scale_low, scale_high, freq_cutoff):
         def custom_cfg_function(args):
             conds_out = args["conds_out"]
             if len(conds_out) <= 1 or None in args["conds"][:2]:
@@ -91,13 +99,16 @@ class FreSca:
         m = model.clone()
         m.set_model_sampler_pre_cfg_function(custom_cfg_function)
 
-        return (m,)
+        return io.NodeOutput(m)
 
 
-NODE_CLASS_MAPPINGS = {
-    "FreSca": FreSca,
-}
+class FreScaExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            FreSca,
+        ]
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "FreSca": "FreSca",
-}
+
+async def comfy_entrypoint() -> FreScaExtension:
+    return FreScaExtension()

comfy_extras/nodes_gits.py

@@ -1,6 +1,8 @@
 # from https://github.com/zju-pi/diff-sampler/tree/main/gits-main
 import numpy as np
 import torch
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 def loglinear_interp(t_steps, num_steps):
     """
@@ -333,25 +335,28 @@ NOISE_LEVELS = {
     ],
 }
 
-class GITSScheduler:
+class GITSScheduler(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required":
-                    {"coeff": ("FLOAT", {"default": 1.20, "min": 0.80, "max": 1.50, "step": 0.05}),
-                     "steps": ("INT", {"default": 10, "min": 2, "max": 1000}),
-                     "denoise": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.01}),
-                      }
-               }
-    RETURN_TYPES = ("SIGMAS",)
-    CATEGORY = "sampling/custom_sampling/schedulers"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="GITSScheduler",
+            category="sampling/custom_sampling/schedulers",
+            inputs=[
+                io.Float.Input("coeff", default=1.20, min=0.80, max=1.50, step=0.05),
+                io.Int.Input("steps", default=10, min=2, max=1000),
+                io.Float.Input("denoise", default=1.0, min=0.0, max=1.0, step=0.01),
+            ],
+            outputs=[
+                io.Sigmas.Output(),
+            ],
+        )
 
-    FUNCTION = "get_sigmas"
-
-    def get_sigmas(self, coeff, steps, denoise):
+    @classmethod
+    def execute(cls, coeff, steps, denoise):
         total_steps = steps
         if denoise < 1.0:
             if denoise <= 0.0:
-                return (torch.FloatTensor([]),)
+                return io.NodeOutput(torch.FloatTensor([]))
             total_steps = round(steps * denoise)
 
         if steps <= 20:
@@ -362,8 +367,16 @@ class GITSScheduler:
 
         sigmas = sigmas[-(total_steps + 1):]
         sigmas[-1] = 0
-        return (torch.FloatTensor(sigmas), )
+        return io.NodeOutput(torch.FloatTensor(sigmas))
 
-NODE_CLASS_MAPPINGS = {
-    "GITSScheduler": GITSScheduler,
-}
+
+class GITSSchedulerExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            GITSScheduler,
+        ]
+
+
+async def comfy_entrypoint() -> GITSSchedulerExtension:
+    return GITSSchedulerExtension()

comfy_extras/nodes_hidream.py

@@ -1,55 +1,73 @@
+from typing_extensions import override
+
 import folder_paths
 import comfy.sd
 import comfy.model_management
+from comfy_api.latest import ComfyExtension, io
 
 
-class QuadrupleCLIPLoader:
+class QuadrupleCLIPLoader(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "clip_name1": (folder_paths.get_filename_list("text_encoders"), ),
-                              "clip_name2": (folder_paths.get_filename_list("text_encoders"), ),
-                              "clip_name3": (folder_paths.get_filename_list("text_encoders"), ),
-                              "clip_name4": (folder_paths.get_filename_list("text_encoders"), )
-                             }}
-    RETURN_TYPES = ("CLIP",)
-    FUNCTION = "load_clip"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="QuadrupleCLIPLoader",
+            category="advanced/loaders",
+            description="[Recipes]\n\nhidream: long clip-l, long clip-g, t5xxl, llama_8b_3.1_instruct",
+            inputs=[
+                io.Combo.Input("clip_name1", options=folder_paths.get_filename_list("text_encoders")),
+                io.Combo.Input("clip_name2", options=folder_paths.get_filename_list("text_encoders")),
+                io.Combo.Input("clip_name3", options=folder_paths.get_filename_list("text_encoders")),
+                io.Combo.Input("clip_name4", options=folder_paths.get_filename_list("text_encoders")),
+            ],
+            outputs=[
+                io.Clip.Output(),
+            ]
+        )
 
-    CATEGORY = "advanced/loaders"
-
-    DESCRIPTION = "[Recipes]\n\nhidream: long clip-l, long clip-g, t5xxl, llama_8b_3.1_instruct"
-
-    def load_clip(self, clip_name1, clip_name2, clip_name3, clip_name4):
+    @classmethod
+    def execute(cls, clip_name1, clip_name2, clip_name3, clip_name4):
         clip_path1 = folder_paths.get_full_path_or_raise("text_encoders", clip_name1)
         clip_path2 = folder_paths.get_full_path_or_raise("text_encoders", clip_name2)
         clip_path3 = folder_paths.get_full_path_or_raise("text_encoders", clip_name3)
         clip_path4 = folder_paths.get_full_path_or_raise("text_encoders", clip_name4)
         clip = comfy.sd.load_clip(ckpt_paths=[clip_path1, clip_path2, clip_path3, clip_path4], embedding_directory=folder_paths.get_folder_paths("embeddings"))
-        return (clip,)
+        return io.NodeOutput(clip)
 
-class CLIPTextEncodeHiDream:
+class CLIPTextEncodeHiDream(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "clip_l": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "clip_g": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "t5xxl": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "llama": ("STRING", {"multiline": True, "dynamicPrompts": True})
-            }}
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
-
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, clip_l, clip_g, t5xxl, llama):
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodeHiDream",
+            category="advanced/conditioning",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.String.Input("clip_l", multiline=True, dynamic_prompts=True),
+                io.String.Input("clip_g", multiline=True, dynamic_prompts=True),
+                io.String.Input("t5xxl", multiline=True, dynamic_prompts=True),
+                io.String.Input("llama", multiline=True, dynamic_prompts=True),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ]
+        )
 
+    @classmethod
+    def execute(cls, clip, clip_l, clip_g, t5xxl, llama):
         tokens = clip.tokenize(clip_g)
         tokens["l"] = clip.tokenize(clip_l)["l"]
         tokens["t5xxl"] = clip.tokenize(t5xxl)["t5xxl"]
         tokens["llama"] = clip.tokenize(llama)["llama"]
-        return (clip.encode_from_tokens_scheduled(tokens), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens))
 
-NODE_CLASS_MAPPINGS = {
-    "QuadrupleCLIPLoader": QuadrupleCLIPLoader,
-    "CLIPTextEncodeHiDream": CLIPTextEncodeHiDream,
-}
+
+class HiDreamExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            QuadrupleCLIPLoader,
+            CLIPTextEncodeHiDream,
+        ]
+
+
+async def comfy_entrypoint() -> HiDreamExtension:
+    return HiDreamExtension()

comfy_extras/nodes_hunyuan.py

@@ -2,42 +2,60 @@ import nodes
 import node_helpers
 import torch
 import comfy.model_management
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 
-class CLIPTextEncodeHunyuanDiT:
+class CLIPTextEncodeHunyuanDiT(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "bert": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "mt5xl": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            }}
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodeHunyuanDiT",
+            category="advanced/conditioning",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.String.Input("bert", multiline=True, dynamic_prompts=True),
+                io.String.Input("mt5xl", multiline=True, dynamic_prompts=True),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, bert, mt5xl):
+    @classmethod
+    def execute(cls, clip, bert, mt5xl) -> io.NodeOutput:
         tokens = clip.tokenize(bert)
         tokens["mt5xl"] = clip.tokenize(mt5xl)["mt5xl"]
 
-        return (clip.encode_from_tokens_scheduled(tokens), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens))
 
-class EmptyHunyuanLatentVideo:
+    encode = execute  # TODO: remove
+
+
+class EmptyHunyuanLatentVideo(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "width": ("INT", {"default": 848, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                              "height": ("INT", {"default": 480, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                              "length": ("INT", {"default": 25, "min": 1, "max": nodes.MAX_RESOLUTION, "step": 4}),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096})}}
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "generate"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="EmptyHunyuanLatentVideo",
+            category="latent/video",
+            inputs=[
+                io.Int.Input("width", default=848, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("height", default=480, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("length", default=25, min=1, max=nodes.MAX_RESOLUTION, step=4),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    CATEGORY = "latent/video"
-
-    def generate(self, width, height, length, batch_size=1):
+    @classmethod
+    def execute(cls, width, height, length, batch_size=1) -> io.NodeOutput:
         latent = torch.zeros([batch_size, 16, ((length - 1) // 4) + 1, height // 8, width // 8], device=comfy.model_management.intermediate_device())
-        return ({"samples":latent}, )
+        return io.NodeOutput({"samples":latent})
+
+    generate = execute  # TODO: remove
+
 
 PROMPT_TEMPLATE_ENCODE_VIDEO_I2V = (
     "<|start_header_id|>system<|end_header_id|>\n\n<image>\nDescribe the video by detailing the following aspects according to the reference image: "
@@ -50,45 +68,61 @@ PROMPT_TEMPLATE_ENCODE_VIDEO_I2V = (
     "<|start_header_id|>assistant<|end_header_id|>\n\n"
 )
 
-class TextEncodeHunyuanVideo_ImageToVideo:
+class TextEncodeHunyuanVideo_ImageToVideo(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "clip_vision_output": ("CLIP_VISION_OUTPUT", ),
-            "prompt": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "image_interleave": ("INT", {"default": 2, "min": 1, "max": 512, "tooltip": "How much the image influences things vs the text prompt. Higher number means more influence from the text prompt."}),
-            }}
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="TextEncodeHunyuanVideo_ImageToVideo",
+            category="advanced/conditioning",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.ClipVisionOutput.Input("clip_vision_output"),
+                io.String.Input("prompt", multiline=True, dynamic_prompts=True),
+                io.Int.Input(
+                    "image_interleave",
+                    default=2,
+                    min=1,
+                    max=512,
+                    tooltip="How much the image influences things vs the text prompt. Higher number means more influence from the text prompt.",
+                ),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, clip_vision_output, prompt, image_interleave):
+    @classmethod
+    def execute(cls, clip, clip_vision_output, prompt, image_interleave) -> io.NodeOutput:
         tokens = clip.tokenize(prompt, llama_template=PROMPT_TEMPLATE_ENCODE_VIDEO_I2V, image_embeds=clip_vision_output.mm_projected, image_interleave=image_interleave)
-        return (clip.encode_from_tokens_scheduled(tokens), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens))
 
-class HunyuanImageToVideo:
+    encode = execute  # TODO: remove
+
+
+class HunyuanImageToVideo(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "vae": ("VAE", ),
-                             "width": ("INT", {"default": 848, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                             "height": ("INT", {"default": 480, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                             "length": ("INT", {"default": 53, "min": 1, "max": nodes.MAX_RESOLUTION, "step": 4}),
-                             "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
-                             "guidance_type": (["v1 (concat)", "v2 (replace)", "custom"], )
-                },
-                "optional": {"start_image": ("IMAGE", ),
-                }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="HunyuanImageToVideo",
+            category="conditioning/video_models",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Vae.Input("vae"),
+                io.Int.Input("width", default=848, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("height", default=480, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("length", default=53, min=1, max=nodes.MAX_RESOLUTION, step=4),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+                io.Combo.Input("guidance_type", options=["v1 (concat)", "v2 (replace)", "custom"]),
+                io.Image.Input("start_image", optional=True),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Latent.Output(display_name="latent"),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "latent")
-    FUNCTION = "encode"
-
-    CATEGORY = "conditioning/video_models"
-
-    def encode(self, positive, vae, width, height, length, batch_size, guidance_type, start_image=None):
+    @classmethod
+    def execute(cls, positive, vae, width, height, length, batch_size, guidance_type, start_image=None) -> io.NodeOutput:
         latent = torch.zeros([batch_size, 16, ((length - 1) // 4) + 1, height // 8, width // 8], device=comfy.model_management.intermediate_device())
         out_latent = {}
 
@@ -111,51 +145,76 @@ class HunyuanImageToVideo:
             positive = node_helpers.conditioning_set_values(positive, cond)
 
         out_latent["samples"] = latent
-        return (positive, out_latent)
+        return io.NodeOutput(positive, out_latent)
 
-class EmptyHunyuanImageLatent:
+    encode = execute  # TODO: remove
+
+
+class EmptyHunyuanImageLatent(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "width": ("INT", {"default": 2048, "min": 64, "max": nodes.MAX_RESOLUTION, "step": 32}),
-                              "height": ("INT", {"default": 2048, "min": 64, "max": nodes.MAX_RESOLUTION, "step": 32}),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096})}}
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "generate"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="EmptyHunyuanImageLatent",
+            category="latent",
+            inputs=[
+                io.Int.Input("width", default=2048, min=64, max=nodes.MAX_RESOLUTION, step=32),
+                io.Int.Input("height", default=2048, min=64, max=nodes.MAX_RESOLUTION, step=32),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    CATEGORY = "latent"
-
-    def generate(self, width, height, batch_size=1):
+    @classmethod
+    def execute(cls, width, height, batch_size=1) -> io.NodeOutput:
         latent = torch.zeros([batch_size, 64, height // 32, width // 32], device=comfy.model_management.intermediate_device())
-        return ({"samples":latent}, )
+        return io.NodeOutput({"samples":latent})
 
-class HunyuanRefinerLatent:
+    generate = execute  # TODO: remove
+
+
+class HunyuanRefinerLatent(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "negative": ("CONDITIONING", ),
-                             "latent": ("LATENT", ),
-                             "noise_augmentation": ("FLOAT", {"default": 0.10, "min": 0.0, "max": 1.0, "step": 0.01}),
-                             }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="HunyuanRefinerLatent",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Latent.Input("latent"),
+                io.Float.Input("noise_augmentation", default=0.10, min=0.0, max=1.0, step=0.01),
 
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent"),
+            ],
+        )
 
-    FUNCTION = "execute"
-
-    def execute(self, positive, negative, latent, noise_augmentation):
+    @classmethod
+    def execute(cls, positive, negative, latent, noise_augmentation) -> io.NodeOutput:
         latent = latent["samples"]
         positive = node_helpers.conditioning_set_values(positive, {"concat_latent_image": latent, "noise_augmentation": noise_augmentation})
         negative = node_helpers.conditioning_set_values(negative, {"concat_latent_image": latent, "noise_augmentation": noise_augmentation})
         out_latent = {}
         out_latent["samples"] = torch.zeros([latent.shape[0], 32, latent.shape[-3], latent.shape[-2], latent.shape[-1]], device=comfy.model_management.intermediate_device())
-        return (positive, negative, out_latent)
+        return io.NodeOutput(positive, negative, out_latent)
 
 
-NODE_CLASS_MAPPINGS = {
-    "CLIPTextEncodeHunyuanDiT": CLIPTextEncodeHunyuanDiT,
-    "TextEncodeHunyuanVideo_ImageToVideo": TextEncodeHunyuanVideo_ImageToVideo,
-    "EmptyHunyuanLatentVideo": EmptyHunyuanLatentVideo,
-    "HunyuanImageToVideo": HunyuanImageToVideo,
-    "EmptyHunyuanImageLatent": EmptyHunyuanImageLatent,
-    "HunyuanRefinerLatent": HunyuanRefinerLatent,
-}
+class HunyuanExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            CLIPTextEncodeHunyuanDiT,
+            TextEncodeHunyuanVideo_ImageToVideo,
+            EmptyHunyuanLatentVideo,
+            HunyuanImageToVideo,
+            EmptyHunyuanImageLatent,
+            HunyuanRefinerLatent,
+        ]
+
+
+async def comfy_entrypoint() -> HunyuanExtension:
+    return HunyuanExtension()

comfy_extras/nodes_hypertile.py

@@ -1,9 +1,11 @@
 #Taken from: https://github.com/tfernd/HyperTile/
 
 import math
+from typing_extensions import override
 from einops import rearrange
 # Use torch rng for consistency across generations
 from torch import randint
+from comfy_api.latest import ComfyExtension, io
 
 def random_divisor(value: int, min_value: int, /, max_options: int = 1) -> int:
     min_value = min(min_value, value)
@@ -20,25 +22,31 @@ def random_divisor(value: int, min_value: int, /, max_options: int = 1) -> int:
 
     return ns[idx]
 
-class HyperTile:
+class HyperTile(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                             "tile_size": ("INT", {"default": 256, "min": 1, "max": 2048}),
-                             "swap_size": ("INT", {"default": 2, "min": 1, "max": 128}),
-                             "max_depth": ("INT", {"default": 0, "min": 0, "max": 10}),
-                             "scale_depth": ("BOOLEAN", {"default": False}),
-                              }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="HyperTile",
+            category="model_patches/unet",
+            inputs=[
+                io.Model.Input("model"),
+                io.Int.Input("tile_size", default=256, min=1, max=2048),
+                io.Int.Input("swap_size", default=2, min=1, max=128),
+                io.Int.Input("max_depth", default=0, min=0, max=10),
+                io.Boolean.Input("scale_depth", default=False),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    CATEGORY = "model_patches/unet"
-
-    def patch(self, model, tile_size, swap_size, max_depth, scale_depth):
+    @classmethod
+    def execute(cls, model, tile_size, swap_size, max_depth, scale_depth) -> io.NodeOutput:
         latent_tile_size = max(32, tile_size) // 8
-        self.temp = None
+        temp = None
 
         def hypertile_in(q, k, v, extra_options):
+            nonlocal temp
             model_chans = q.shape[-2]
             orig_shape = extra_options['original_shape']
             apply_to = []
@@ -58,14 +66,15 @@ class HyperTile:
 
                 if nh * nw > 1:
                     q = rearrange(q, "b (nh h nw w) c -> (b nh nw) (h w) c", h=h // nh, w=w // nw, nh=nh, nw=nw)
-                    self.temp = (nh, nw, h, w)
+                    temp = (nh, nw, h, w)
                 return q, k, v
 
             return q, k, v
         def hypertile_out(out, extra_options):
-            if self.temp is not None:
-                nh, nw, h, w = self.temp
-                self.temp = None
+            nonlocal temp
+            if temp is not None:
+                nh, nw, h, w = temp
+                temp = None
                 out = rearrange(out, "(b nh nw) hw c -> b nh nw hw c", nh=nh, nw=nw)
                 out = rearrange(out, "b nh nw (h w) c -> b (nh h nw w) c", h=h // nh, w=w // nw)
             return out
@@ -76,6 +85,14 @@ class HyperTile:
         m.set_model_attn1_output_patch(hypertile_out)
         return (m, )
 
-NODE_CLASS_MAPPINGS = {
-    "HyperTile": HyperTile,
-}
+
+class HyperTileExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            HyperTile,
+        ]
+
+
+async def comfy_entrypoint() -> HyperTileExtension:
+    return HyperTileExtension()

comfy_extras/nodes_ip2p.py

@@ -1,21 +1,30 @@
 import torch
 
-class InstructPixToPixConditioning:
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
+
+
+class InstructPixToPixConditioning(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "negative": ("CONDITIONING", ),
-                             "vae": ("VAE", ),
-                             "pixels": ("IMAGE", ),
-                             }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="InstructPixToPixConditioning",
+            category="conditioning/instructpix2pix",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Vae.Input("vae"),
+                io.Image.Input("pixels"),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent"),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING","CONDITIONING","LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
-    FUNCTION = "encode"
-
-    CATEGORY = "conditioning/instructpix2pix"
-
-    def encode(self, positive, negative, pixels, vae):
+    @classmethod
+    def execute(cls, positive, negative, pixels, vae) -> io.NodeOutput:
         x = (pixels.shape[1] // 8) * 8
         y = (pixels.shape[2] // 8) * 8
 
@@ -38,8 +47,17 @@ class InstructPixToPixConditioning:
                 n = [t[0], d]
                 c.append(n)
             out.append(c)
-        return (out[0], out[1], out_latent)
+        return io.NodeOutput(out[0], out[1], out_latent)
+
+
+class InstructPix2PixExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            InstructPixToPixConditioning,
+        ]
+
+
+async def comfy_entrypoint() -> InstructPix2PixExtension:
+    return InstructPix2PixExtension()
 
-NODE_CLASS_MAPPINGS = {
-    "InstructPixToPixConditioning": InstructPixToPixConditioning,
-}

comfy_extras/nodes_latent.py

@@ -2,6 +2,8 @@ import comfy.utils
 import comfy_extras.nodes_post_processing
 import torch
 import nodes
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 
 def reshape_latent_to(target_shape, latent, repeat_batch=True):
@@ -13,17 +15,23 @@ def reshape_latent_to(target_shape, latent, repeat_batch=True):
         return latent
 
 
-class LatentAdd:
+class LatentAdd(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples1": ("LATENT",), "samples2": ("LATENT",)}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentAdd",
+            category="latent/advanced",
+            inputs=[
+                io.Latent.Input("samples1"),
+                io.Latent.Input("samples2"),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced"
-
-    def op(self, samples1, samples2):
+    @classmethod
+    def execute(cls, samples1, samples2) -> io.NodeOutput:
         samples_out = samples1.copy()
 
         s1 = samples1["samples"]
@@ -31,19 +39,25 @@ class LatentAdd:
 
         s2 = reshape_latent_to(s1.shape, s2)
         samples_out["samples"] = s1 + s2
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentSubtract:
+class LatentSubtract(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples1": ("LATENT",), "samples2": ("LATENT",)}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentSubtract",
+            category="latent/advanced",
+            inputs=[
+                io.Latent.Input("samples1"),
+                io.Latent.Input("samples2"),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced"
-
-    def op(self, samples1, samples2):
+    @classmethod
+    def execute(cls, samples1, samples2) -> io.NodeOutput:
         samples_out = samples1.copy()
 
         s1 = samples1["samples"]
@@ -51,41 +65,49 @@ class LatentSubtract:
 
         s2 = reshape_latent_to(s1.shape, s2)
         samples_out["samples"] = s1 - s2
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentMultiply:
+class LatentMultiply(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples": ("LATENT",),
-                              "multiplier": ("FLOAT", {"default": 1.0, "min": -10.0, "max": 10.0, "step": 0.01}),
-                             }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentMultiply",
+            category="latent/advanced",
+            inputs=[
+                io.Latent.Input("samples"),
+                io.Float.Input("multiplier", default=1.0, min=-10.0, max=10.0, step=0.01),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced"
-
-    def op(self, samples, multiplier):
+    @classmethod
+    def execute(cls, samples, multiplier) -> io.NodeOutput:
         samples_out = samples.copy()
 
         s1 = samples["samples"]
         samples_out["samples"] = s1 * multiplier
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentInterpolate:
+class LatentInterpolate(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples1": ("LATENT",),
-                              "samples2": ("LATENT",),
-                              "ratio": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.01}),
-                              }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentInterpolate",
+            category="latent/advanced",
+            inputs=[
+                io.Latent.Input("samples1"),
+                io.Latent.Input("samples2"),
+                io.Float.Input("ratio", default=1.0, min=0.0, max=1.0, step=0.01),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced"
-
-    def op(self, samples1, samples2, ratio):
+    @classmethod
+    def execute(cls, samples1, samples2, ratio) -> io.NodeOutput:
         samples_out = samples1.copy()
 
         s1 = samples1["samples"]
@@ -104,19 +126,26 @@ class LatentInterpolate:
         st = torch.nan_to_num(t / mt)
 
         samples_out["samples"] = st * (m1 * ratio + m2 * (1.0 - ratio))
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentConcat:
+class LatentConcat(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples1": ("LATENT",), "samples2": ("LATENT",), "dim": (["x", "-x", "y", "-y", "t", "-t"], )}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentConcat",
+            category="latent/advanced",
+            inputs=[
+                io.Latent.Input("samples1"),
+                io.Latent.Input("samples2"),
+                io.Combo.Input("dim", options=["x", "-x", "y", "-y", "t", "-t"]),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced"
-
-    def op(self, samples1, samples2, dim):
+    @classmethod
+    def execute(cls, samples1, samples2, dim) -> io.NodeOutput:
         samples_out = samples1.copy()
 
         s1 = samples1["samples"]
@@ -136,22 +165,27 @@ class LatentConcat:
             dim = -3
 
         samples_out["samples"] = torch.cat(c, dim=dim)
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentCut:
+class LatentCut(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"samples": ("LATENT",),
-                             "dim": (["x", "y", "t"], ),
-                             "index": ("INT", {"default": 0, "min": -nodes.MAX_RESOLUTION, "max": nodes.MAX_RESOLUTION, "step": 1}),
-                             "amount": ("INT", {"default": 1, "min": 1, "max": nodes.MAX_RESOLUTION, "step": 1})}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentCut",
+            category="latent/advanced",
+            inputs=[
+                io.Latent.Input("samples"),
+                io.Combo.Input("dim", options=["x", "y", "t"]),
+                io.Int.Input("index", default=0, min=-nodes.MAX_RESOLUTION, max=nodes.MAX_RESOLUTION, step=1),
+                io.Int.Input("amount", default=1, min=1, max=nodes.MAX_RESOLUTION, step=1),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced"
-
-    def op(self, samples, dim, index, amount):
+    @classmethod
+    def execute(cls, samples, dim, index, amount) -> io.NodeOutput:
         samples_out = samples.copy()
 
         s1 = samples["samples"]
@@ -171,19 +205,25 @@ class LatentCut:
             amount = min(-index, amount)
 
         samples_out["samples"] = torch.narrow(s1, dim, index, amount)
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentBatch:
+class LatentBatch(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples1": ("LATENT",), "samples2": ("LATENT",)}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentBatch",
+            category="latent/batch",
+            inputs=[
+                io.Latent.Input("samples1"),
+                io.Latent.Input("samples2"),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "batch"
-
-    CATEGORY = "latent/batch"
-
-    def batch(self, samples1, samples2):
+    @classmethod
+    def execute(cls, samples1, samples2) -> io.NodeOutput:
         samples_out = samples1.copy()
         s1 = samples1["samples"]
         s2 = samples2["samples"]
@@ -192,20 +232,25 @@ class LatentBatch:
         s = torch.cat((s1, s2), dim=0)
         samples_out["samples"] = s
         samples_out["batch_index"] = samples1.get("batch_index", [x for x in range(0, s1.shape[0])]) + samples2.get("batch_index", [x for x in range(0, s2.shape[0])])
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentBatchSeedBehavior:
+class LatentBatchSeedBehavior(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples": ("LATENT",),
-                              "seed_behavior": (["random", "fixed"],{"default": "fixed"}),}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentBatchSeedBehavior",
+            category="latent/advanced",
+            inputs=[
+                io.Latent.Input("samples"),
+                io.Combo.Input("seed_behavior", options=["random", "fixed"], default="fixed"),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced"
-
-    def op(self, samples, seed_behavior):
+    @classmethod
+    def execute(cls, samples, seed_behavior) -> io.NodeOutput:
         samples_out = samples.copy()
         latent = samples["samples"]
         if seed_behavior == "random":
@@ -215,41 +260,50 @@ class LatentBatchSeedBehavior:
             batch_number = samples_out.get("batch_index", [0])[0]
             samples_out["batch_index"] = [batch_number] * latent.shape[0]
 
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentApplyOperation:
+class LatentApplyOperation(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "samples": ("LATENT",),
-                             "operation": ("LATENT_OPERATION",),
-                             }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentApplyOperation",
+            category="latent/advanced/operations",
+            is_experimental=True,
+            inputs=[
+                io.Latent.Input("samples"),
+                io.LatentOperation.Input("operation"),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced/operations"
-    EXPERIMENTAL = True
-
-    def op(self, samples, operation):
+    @classmethod
+    def execute(cls, samples, operation) -> io.NodeOutput:
         samples_out = samples.copy()
 
         s1 = samples["samples"]
         samples_out["samples"] = operation(latent=s1)
-        return (samples_out,)
+        return io.NodeOutput(samples_out)
 
-class LatentApplyOperationCFG:
+class LatentApplyOperationCFG(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                             "operation": ("LATENT_OPERATION",),
-                              }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentApplyOperationCFG",
+            category="latent/advanced/operations",
+            is_experimental=True,
+            inputs=[
+                io.Model.Input("model"),
+                io.LatentOperation.Input("operation"),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    CATEGORY = "latent/advanced/operations"
-    EXPERIMENTAL = True
-
-    def patch(self, model, operation):
+    @classmethod
+    def execute(cls, model, operation) -> io.NodeOutput:
         m = model.clone()
 
         def pre_cfg_function(args):
@@ -261,21 +315,25 @@ class LatentApplyOperationCFG:
             return conds_out
 
         m.set_model_sampler_pre_cfg_function(pre_cfg_function)
-        return (m, )
+        return io.NodeOutput(m)
 
-class LatentOperationTonemapReinhard:
+class LatentOperationTonemapReinhard(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "multiplier": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 100.0, "step": 0.01}),
-                              }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentOperationTonemapReinhard",
+            category="latent/advanced/operations",
+            is_experimental=True,
+            inputs=[
+                io.Float.Input("multiplier", default=1.0, min=0.0, max=100.0, step=0.01),
+            ],
+            outputs=[
+                io.LatentOperation.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT_OPERATION",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced/operations"
-    EXPERIMENTAL = True
-
-    def op(self, multiplier):
+    @classmethod
+    def execute(cls, multiplier) -> io.NodeOutput:
         def tonemap_reinhard(latent, **kwargs):
             latent_vector_magnitude = (torch.linalg.vector_norm(latent, dim=(1)) + 0.0000000001)[:,None]
             normalized_latent = latent / latent_vector_magnitude
@@ -291,39 +349,27 @@ class LatentOperationTonemapReinhard:
             new_magnitude *= top
 
             return normalized_latent * new_magnitude
-        return (tonemap_reinhard,)
+        return io.NodeOutput(tonemap_reinhard)
 
-class LatentOperationSharpen:
+class LatentOperationSharpen(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-                "sharpen_radius": ("INT", {
-                    "default": 9,
-                    "min": 1,
-                    "max": 31,
-                    "step": 1
-                }),
-                "sigma": ("FLOAT", {
-                    "default": 1.0,
-                    "min": 0.1,
-                    "max": 10.0,
-                    "step": 0.1
-                }),
-                "alpha": ("FLOAT", {
-                    "default": 0.1,
-                    "min": 0.0,
-                    "max": 5.0,
-                    "step": 0.01
-                }),
-                              }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LatentOperationSharpen",
+            category="latent/advanced/operations",
+            is_experimental=True,
+            inputs=[
+                io.Int.Input("sharpen_radius", default=9, min=1, max=31, step=1),
+                io.Float.Input("sigma", default=1.0, min=0.1, max=10.0, step=0.1),
+                io.Float.Input("alpha", default=0.1, min=0.0, max=5.0, step=0.01),
+            ],
+            outputs=[
+                io.LatentOperation.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("LATENT_OPERATION",)
-    FUNCTION = "op"
-
-    CATEGORY = "latent/advanced/operations"
-    EXPERIMENTAL = True
-
-    def op(self, sharpen_radius, sigma, alpha):
+    @classmethod
+    def execute(cls, sharpen_radius, sigma, alpha) -> io.NodeOutput:
         def sharpen(latent, **kwargs):
             luminance = (torch.linalg.vector_norm(latent, dim=(1)) + 1e-6)[:,None]
             normalized_latent = latent / luminance
@@ -340,19 +386,27 @@ class LatentOperationSharpen:
             sharpened = torch.nn.functional.conv2d(padded_image, kernel.repeat(channels, 1, 1).unsqueeze(1), padding=kernel_size // 2, groups=channels)[:,:,sharpen_radius:-sharpen_radius, sharpen_radius:-sharpen_radius]
 
             return luminance * sharpened
-        return (sharpen,)
+        return io.NodeOutput(sharpen)
 
-NODE_CLASS_MAPPINGS = {
-    "LatentAdd": LatentAdd,
-    "LatentSubtract": LatentSubtract,
-    "LatentMultiply": LatentMultiply,
-    "LatentInterpolate": LatentInterpolate,
-    "LatentConcat": LatentConcat,
-    "LatentCut": LatentCut,
-    "LatentBatch": LatentBatch,
-    "LatentBatchSeedBehavior": LatentBatchSeedBehavior,
-    "LatentApplyOperation": LatentApplyOperation,
-    "LatentApplyOperationCFG": LatentApplyOperationCFG,
-    "LatentOperationTonemapReinhard": LatentOperationTonemapReinhard,
-    "LatentOperationSharpen": LatentOperationSharpen,
-}
+
+class LatentExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            LatentAdd,
+            LatentSubtract,
+            LatentMultiply,
+            LatentInterpolate,
+            LatentConcat,
+            LatentCut,
+            LatentBatch,
+            LatentBatchSeedBehavior,
+            LatentApplyOperation,
+            LatentApplyOperationCFG,
+            LatentOperationTonemapReinhard,
+            LatentOperationSharpen,
+        ]
+
+
+async def comfy_entrypoint() -> LatentExtension:
+    return LatentExtension()

comfy_extras/nodes_lora_extract.py

@@ -5,6 +5,8 @@ import folder_paths
 import os
 import logging
 from enum import Enum
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 CLAMP_QUANTILE = 0.99
 
@@ -71,32 +73,40 @@ def calc_lora_model(model_diff, rank, prefix_model, prefix_lora, output_sd, lora
             output_sd["{}{}.diff_b".format(prefix_lora, k[len(prefix_model):-5])] = sd[k].contiguous().half().cpu()
     return output_sd
 
-class LoraSave:
-    def __init__(self):
-        self.output_dir = folder_paths.get_output_directory()
+class LoraSave(io.ComfyNode):
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LoraSave",
+            display_name="Extract and Save Lora",
+            category="_for_testing",
+            inputs=[
+                io.String.Input("filename_prefix", default="loras/ComfyUI_extracted_lora"),
+                io.Int.Input("rank", default=8, min=1, max=4096, step=1),
+                io.Combo.Input("lora_type", options=tuple(LORA_TYPES.keys())),
+                io.Boolean.Input("bias_diff", default=True),
+                io.Model.Input(
+                    "model_diff",
+                    tooltip="The ModelSubtract output to be converted to a lora.",
+                    optional=True,
+                ),
+                io.Clip.Input(
+                  "text_encoder_diff",
+                    tooltip="The CLIPSubtract output to be converted to a lora.",
+                    optional=True,
+                ),
+            ],
+            is_experimental=True,
+            is_output_node=True,
+        )
 
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"filename_prefix": ("STRING", {"default": "loras/ComfyUI_extracted_lora"}),
-                              "rank": ("INT", {"default": 8, "min": 1, "max": 4096, "step": 1}),
-                              "lora_type": (tuple(LORA_TYPES.keys()),),
-                              "bias_diff": ("BOOLEAN", {"default": True}),
-                            },
-                "optional": {"model_diff": ("MODEL", {"tooltip": "The ModelSubtract output to be converted to a lora."}),
-                             "text_encoder_diff": ("CLIP", {"tooltip": "The CLIPSubtract output to be converted to a lora."})},
-    }
-    RETURN_TYPES = ()
-    FUNCTION = "save"
-    OUTPUT_NODE = True
-
-    CATEGORY = "_for_testing"
-
-    def save(self, filename_prefix, rank, lora_type, bias_diff, model_diff=None, text_encoder_diff=None):
+    def execute(cls, filename_prefix, rank, lora_type, bias_diff, model_diff=None, text_encoder_diff=None) -> io.NodeOutput:
         if model_diff is None and text_encoder_diff is None:
-            return {}
+            return io.NodeOutput()
 
         lora_type = LORA_TYPES.get(lora_type)
-        full_output_folder, filename, counter, subfolder, filename_prefix = folder_paths.get_save_image_path(filename_prefix, self.output_dir)
+        full_output_folder, filename, counter, subfolder, filename_prefix = folder_paths.get_save_image_path(filename_prefix, folder_paths.get_output_directory())
 
         output_sd = {}
         if model_diff is not None:
@@ -108,12 +118,16 @@ class LoraSave:
         output_checkpoint = os.path.join(full_output_folder, output_checkpoint)
 
         comfy.utils.save_torch_file(output_sd, output_checkpoint, metadata=None)
-        return {}
+        return io.NodeOutput()
 
-NODE_CLASS_MAPPINGS = {
-    "LoraSave": LoraSave
-}
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "LoraSave": "Extract and Save Lora"
-}
+class LoraSaveExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            LoraSave,
+        ]
+
+
+async def comfy_entrypoint() -> LoraSaveExtension:
+    return LoraSaveExtension()

comfy_extras/nodes_lotus.py

@@ -1,20 +1,22 @@
+from typing_extensions import override
+
 import torch
 import comfy.model_management as mm
+from comfy_api.latest import ComfyExtension, io
 
-class LotusConditioning:
+
+class LotusConditioning(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-            },
-        }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LotusConditioning",
+            category="conditioning/lotus",
+            inputs=[],
+            outputs=[io.Conditioning.Output(display_name="conditioning")],
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    RETURN_NAMES = ("conditioning",)
-    FUNCTION = "conditioning"
-    CATEGORY = "conditioning/lotus"
-
-    def conditioning(self):
+    @classmethod
+    def execute(cls) -> io.NodeOutput:
         device = mm.get_torch_device()
         #lotus uses a frozen encoder and null conditioning, i'm just inlining the results of that operation since it doesn't change
         #and getting parity with the reference implementation would otherwise require inference and 800mb of tensors
@@ -22,8 +24,16 @@ class LotusConditioning:
 
         cond = [[prompt_embeds, {}]]
 
-        return (cond,)
+        return io.NodeOutput(cond)
 
-NODE_CLASS_MAPPINGS = {
-    "LotusConditioning" : LotusConditioning,
-}
+
+class LotusExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            LotusConditioning,
+        ]
+
+
+async def comfy_entrypoint() -> LotusExtension:
+    return LotusExtension()

comfy_extras/nodes_lt.py

@@ -1,4 +1,3 @@
-import io
 import nodes
 import node_helpers
 import torch
@@ -8,46 +7,61 @@ import comfy.utils
 import math
 import numpy as np
 import av
+from io import BytesIO
+from typing_extensions import override
 from comfy.ldm.lightricks.symmetric_patchifier import SymmetricPatchifier, latent_to_pixel_coords
+from comfy_api.latest import ComfyExtension, io
 
-class EmptyLTXVLatentVideo:
+class EmptyLTXVLatentVideo(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "width": ("INT", {"default": 768, "min": 64, "max": nodes.MAX_RESOLUTION, "step": 32}),
-                              "height": ("INT", {"default": 512, "min": 64, "max": nodes.MAX_RESOLUTION, "step": 32}),
-                              "length": ("INT", {"default": 97, "min": 1, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096})}}
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "generate"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="EmptyLTXVLatentVideo",
+            category="latent/video/ltxv",
+            inputs=[
+                io.Int.Input("width", default=768, min=64, max=nodes.MAX_RESOLUTION, step=32),
+                io.Int.Input("height", default=512, min=64, max=nodes.MAX_RESOLUTION, step=32),
+                io.Int.Input("length", default=97, min=1, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    CATEGORY = "latent/video/ltxv"
-
-    def generate(self, width, height, length, batch_size=1):
+    @classmethod
+    def execute(cls, width, height, length, batch_size=1) -> io.NodeOutput:
         latent = torch.zeros([batch_size, 128, ((length - 1) // 8) + 1, height // 32, width // 32], device=comfy.model_management.intermediate_device())
-        return ({"samples": latent}, )
+        return io.NodeOutput({"samples": latent})
 
+    generate = execute  # TODO: remove
 
-class LTXVImgToVideo:
+class LTXVImgToVideo(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "negative": ("CONDITIONING", ),
-                             "vae": ("VAE",),
-                             "image": ("IMAGE",),
-                             "width": ("INT", {"default": 768, "min": 64, "max": nodes.MAX_RESOLUTION, "step": 32}),
-                             "height": ("INT", {"default": 512, "min": 64, "max": nodes.MAX_RESOLUTION, "step": 32}),
-                             "length": ("INT", {"default": 97, "min": 9, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                             "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
-                             "strength": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0}),
-                             }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LTXVImgToVideo",
+            category="conditioning/video_models",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Vae.Input("vae"),
+                io.Image.Input("image"),
+                io.Int.Input("width", default=768, min=64, max=nodes.MAX_RESOLUTION, step=32),
+                io.Int.Input("height", default=512, min=64, max=nodes.MAX_RESOLUTION, step=32),
+                io.Int.Input("length", default=97, min=9, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+                io.Float.Input("strength", default=1.0, min=0.0, max=1.0),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent"),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
-
-    CATEGORY = "conditioning/video_models"
-    FUNCTION = "generate"
-
-    def generate(self, positive, negative, image, vae, width, height, length, batch_size, strength):
+    @classmethod
+    def execute(cls, positive, negative, image, vae, width, height, length, batch_size, strength) -> io.NodeOutput:
         pixels = comfy.utils.common_upscale(image.movedim(-1, 1), width, height, "bilinear", "center").movedim(1, -1)
         encode_pixels = pixels[:, :, :, :3]
         t = vae.encode(encode_pixels)
@@ -62,7 +76,9 @@ class LTXVImgToVideo:
         )
         conditioning_latent_frames_mask[:, :, :t.shape[2]] = 1.0 - strength
 
-        return (positive, negative, {"samples": latent, "noise_mask": conditioning_latent_frames_mask}, )
+        return io.NodeOutput(positive, negative, {"samples": latent, "noise_mask": conditioning_latent_frames_mask})
+
+    generate = execute  # TODO: remove
 
 
 def conditioning_get_any_value(conditioning, key, default=None):
@@ -93,35 +109,46 @@ def get_keyframe_idxs(cond):
     num_keyframes = torch.unique(keyframe_idxs[:, 0]).shape[0]
     return keyframe_idxs, num_keyframes
 
-class LTXVAddGuide:
+class LTXVAddGuide(io.ComfyNode):
+    NUM_PREFIX_FRAMES = 2
+    PATCHIFIER = SymmetricPatchifier(1)
+
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "negative": ("CONDITIONING", ),
-                             "vae": ("VAE",),
-                             "latent": ("LATENT",),
-                             "image": ("IMAGE", {"tooltip": "Image or video to condition the latent video on. Must be 8*n + 1 frames."
-                                                 "If the video is not 8*n + 1 frames, it will be cropped to the nearest 8*n + 1 frames."}),
-                             "frame_idx": ("INT", {"default": 0, "min": -9999, "max": 9999,
-                                                   "tooltip": "Frame index to start the conditioning at. For single-frame images or "
-                                                   "videos with 1-8 frames, any frame_idx value is acceptable. For videos with 9+ "
-                                                   "frames, frame_idx must be divisible by 8, otherwise it will be rounded down to "
-                                                   "the nearest multiple of 8. Negative values are counted from the end of the video."}),
-                             "strength": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.01}),
-                             }
-            }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LTXVAddGuide",
+            category="conditioning/video_models",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Vae.Input("vae"),
+                io.Latent.Input("latent"),
+                io.Image.Input(
+                    "image",
+                    tooltip="Image or video to condition the latent video on. Must be 8*n + 1 frames. "
+                            "If the video is not 8*n + 1 frames, it will be cropped to the nearest 8*n + 1 frames.",
+                ),
+                io.Int.Input(
+                    "frame_idx",
+                    default=0,
+                    min=-9999,
+                    max=9999,
+                    tooltip="Frame index to start the conditioning at. "
+                            "For single-frame images or videos with 1-8 frames, any frame_idx value is acceptable. "
+                            "For videos with 9+ frames, frame_idx must be divisible by 8, otherwise it will be rounded "
+                            "down to the nearest multiple of 8. Negative values are counted from the end of the video.",
+                ),
+                io.Float.Input("strength", default=1.0, min=0.0, max=1.0, step=0.01),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent"),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
-
-    CATEGORY = "conditioning/video_models"
-    FUNCTION = "generate"
-
-    def __init__(self):
-        self._num_prefix_frames = 2
-        self._patchifier = SymmetricPatchifier(1)
-
-    def encode(self, vae, latent_width, latent_height, images, scale_factors):
+    @classmethod
+    def encode(cls, vae, latent_width, latent_height, images, scale_factors):
         time_scale_factor, width_scale_factor, height_scale_factor = scale_factors
         images = images[:(images.shape[0] - 1) // time_scale_factor * time_scale_factor + 1]
         pixels = comfy.utils.common_upscale(images.movedim(-1, 1), latent_width * width_scale_factor, latent_height * height_scale_factor, "bilinear", crop="disabled").movedim(1, -1)
@@ -129,7 +156,8 @@ class LTXVAddGuide:
         t = vae.encode(encode_pixels)
         return encode_pixels, t
 
-    def get_latent_index(self, cond, latent_length, guide_length, frame_idx, scale_factors):
+    @classmethod
+    def get_latent_index(cls, cond, latent_length, guide_length, frame_idx, scale_factors):
         time_scale_factor, _, _ = scale_factors
         _, num_keyframes = get_keyframe_idxs(cond)
         latent_count = latent_length - num_keyframes
@@ -141,9 +169,10 @@ class LTXVAddGuide:
 
         return frame_idx, latent_idx
 
-    def add_keyframe_index(self, cond, frame_idx, guiding_latent, scale_factors):
+    @classmethod
+    def add_keyframe_index(cls, cond, frame_idx, guiding_latent, scale_factors):
         keyframe_idxs, _ = get_keyframe_idxs(cond)
-        _, latent_coords = self._patchifier.patchify(guiding_latent)
+        _, latent_coords = cls.PATCHIFIER.patchify(guiding_latent)
         pixel_coords = latent_to_pixel_coords(latent_coords, scale_factors, causal_fix=frame_idx == 0)  # we need the causal fix only if we're placing the new latents at index 0
         pixel_coords[:, 0] += frame_idx
         if keyframe_idxs is None:
@@ -152,8 +181,9 @@ class LTXVAddGuide:
             keyframe_idxs = torch.cat([keyframe_idxs, pixel_coords], dim=2)
         return node_helpers.conditioning_set_values(cond, {"keyframe_idxs": keyframe_idxs})
 
-    def append_keyframe(self, positive, negative, frame_idx, latent_image, noise_mask, guiding_latent, strength, scale_factors):
-        _, latent_idx = self.get_latent_index(
+    @classmethod
+    def append_keyframe(cls, positive, negative, frame_idx, latent_image, noise_mask, guiding_latent, strength, scale_factors):
+        _, latent_idx = cls.get_latent_index(
             cond=positive,
             latent_length=latent_image.shape[2],
             guide_length=guiding_latent.shape[2],
@@ -162,8 +192,8 @@ class LTXVAddGuide:
         )
         noise_mask[:, :, latent_idx:latent_idx + guiding_latent.shape[2]] = 1.0
 
-        positive = self.add_keyframe_index(positive, frame_idx, guiding_latent, scale_factors)
-        negative = self.add_keyframe_index(negative, frame_idx, guiding_latent, scale_factors)
+        positive = cls.add_keyframe_index(positive, frame_idx, guiding_latent, scale_factors)
+        negative = cls.add_keyframe_index(negative, frame_idx, guiding_latent, scale_factors)
 
         mask = torch.full(
             (noise_mask.shape[0], 1, guiding_latent.shape[2], noise_mask.shape[3], noise_mask.shape[4]),
@@ -176,7 +206,8 @@ class LTXVAddGuide:
         noise_mask = torch.cat([noise_mask, mask], dim=2)
         return positive, negative, latent_image, noise_mask
 
-    def replace_latent_frames(self, latent_image, noise_mask, guiding_latent, latent_idx, strength):
+    @classmethod
+    def replace_latent_frames(cls, latent_image, noise_mask, guiding_latent, latent_idx, strength):
         cond_length = guiding_latent.shape[2]
         assert latent_image.shape[2] >= latent_idx + cond_length, "Conditioning frames exceed the length of the latent sequence."
 
@@ -195,20 +226,21 @@ class LTXVAddGuide:
 
         return latent_image, noise_mask
 
-    def generate(self, positive, negative, vae, latent, image, frame_idx, strength):
+    @classmethod
+    def execute(cls, positive, negative, vae, latent, image, frame_idx, strength) -> io.NodeOutput:
         scale_factors = vae.downscale_index_formula
         latent_image = latent["samples"]
         noise_mask = get_noise_mask(latent)
 
         _, _, latent_length, latent_height, latent_width = latent_image.shape
-        image, t = self.encode(vae, latent_width, latent_height, image, scale_factors)
+        image, t = cls.encode(vae, latent_width, latent_height, image, scale_factors)
 
-        frame_idx, latent_idx = self.get_latent_index(positive, latent_length, len(image), frame_idx, scale_factors)
+        frame_idx, latent_idx = cls.get_latent_index(positive, latent_length, len(image), frame_idx, scale_factors)
         assert latent_idx + t.shape[2] <= latent_length, "Conditioning frames exceed the length of the latent sequence."
 
-        num_prefix_frames = min(self._num_prefix_frames, t.shape[2])
+        num_prefix_frames = min(cls.NUM_PREFIX_FRAMES, t.shape[2])
 
-        positive, negative, latent_image, noise_mask = self.append_keyframe(
+        positive, negative, latent_image, noise_mask = cls.append_keyframe(
             positive,
             negative,
             frame_idx,
@@ -223,9 +255,9 @@ class LTXVAddGuide:
 
         t = t[:, :, num_prefix_frames:]
         if t.shape[2] == 0:
-            return (positive, negative, {"samples": latent_image, "noise_mask": noise_mask},)
+            return io.NodeOutput(positive, negative, {"samples": latent_image, "noise_mask": noise_mask})
 
-        latent_image, noise_mask = self.replace_latent_frames(
+        latent_image, noise_mask = cls.replace_latent_frames(
             latent_image,
             noise_mask,
             t,
@@ -233,34 +265,37 @@ class LTXVAddGuide:
             strength,
         )
 
-        return (positive, negative, {"samples": latent_image, "noise_mask": noise_mask},)
+        return io.NodeOutput(positive, negative, {"samples": latent_image, "noise_mask": noise_mask})
+
+    generate = execute  # TODO: remove
 
 
-class LTXVCropGuides:
+class LTXVCropGuides(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "negative": ("CONDITIONING", ),
-                             "latent": ("LATENT",),
-                             }
-            }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LTXVCropGuides",
+            category="conditioning/video_models",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Latent.Input("latent"),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent"),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
-
-    CATEGORY = "conditioning/video_models"
-    FUNCTION = "crop"
-
-    def __init__(self):
-        self._patchifier = SymmetricPatchifier(1)
-
-    def crop(self, positive, negative, latent):
+    @classmethod
+    def execute(cls, positive, negative, latent) -> io.NodeOutput:
         latent_image = latent["samples"].clone()
         noise_mask = get_noise_mask(latent)
 
         _, num_keyframes = get_keyframe_idxs(positive)
         if num_keyframes == 0:
-            return (positive, negative, {"samples": latent_image, "noise_mask": noise_mask},)
+            return io.NodeOutput(positive, negative, {"samples": latent_image, "noise_mask": noise_mask},)
 
         latent_image = latent_image[:, :, :-num_keyframes]
         noise_mask = noise_mask[:, :, :-num_keyframes]
@@ -268,44 +303,54 @@ class LTXVCropGuides:
         positive = node_helpers.conditioning_set_values(positive, {"keyframe_idxs": None})
         negative = node_helpers.conditioning_set_values(negative, {"keyframe_idxs": None})
 
-        return (positive, negative, {"samples": latent_image, "noise_mask": noise_mask},)
+        return io.NodeOutput(positive, negative, {"samples": latent_image, "noise_mask": noise_mask})
+
+    crop = execute  # TODO: remove
 
 
-class LTXVConditioning:
+class LTXVConditioning(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "negative": ("CONDITIONING", ),
-                             "frame_rate": ("FLOAT", {"default": 25.0, "min": 0.0, "max": 1000.0, "step": 0.01}),
-                             }}
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING")
-    RETURN_NAMES = ("positive", "negative")
-    FUNCTION = "append"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LTXVConditioning",
+            category="conditioning/video_models",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Float.Input("frame_rate", default=25.0, min=0.0, max=1000.0, step=0.01),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+            ],
+        )
 
-    CATEGORY = "conditioning/video_models"
-
-    def append(self, positive, negative, frame_rate):
+    @classmethod
+    def execute(cls, positive, negative, frame_rate) -> io.NodeOutput:
         positive = node_helpers.conditioning_set_values(positive, {"frame_rate": frame_rate})
         negative = node_helpers.conditioning_set_values(negative, {"frame_rate": frame_rate})
-        return (positive, negative)
+        return io.NodeOutput(positive, negative)
 
 
-class ModelSamplingLTXV:
+class ModelSamplingLTXV(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                              "max_shift": ("FLOAT", {"default": 2.05, "min": 0.0, "max": 100.0, "step":0.01}),
-                              "base_shift": ("FLOAT", {"default": 0.95, "min": 0.0, "max": 100.0, "step":0.01}),
-                              },
-                "optional": {"latent": ("LATENT",), }
-                }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ModelSamplingLTXV",
+            category="advanced/model",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input("max_shift", default=2.05, min=0.0, max=100.0, step=0.01),
+                io.Float.Input("base_shift", default=0.95, min=0.0, max=100.0, step=0.01),
+                io.Latent.Input("latent", optional=True),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
-
-    CATEGORY = "advanced/model"
-
-    def patch(self, model, max_shift, base_shift, latent=None):
+    @classmethod
+    def execute(cls, model, max_shift, base_shift, latent=None) -> io.NodeOutput:
         m = model.clone()
 
         if latent is None:
@@ -329,37 +374,41 @@ class ModelSamplingLTXV:
         model_sampling.set_parameters(shift=shift)
         m.add_object_patch("model_sampling", model_sampling)
 
-        return (m, )
+        return io.NodeOutput(m)
 
 
-class LTXVScheduler:
+class LTXVScheduler(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required":
-                    {"steps": ("INT", {"default": 20, "min": 1, "max": 10000}),
-                     "max_shift": ("FLOAT", {"default": 2.05, "min": 0.0, "max": 100.0, "step":0.01}),
-                     "base_shift": ("FLOAT", {"default": 0.95, "min": 0.0, "max": 100.0, "step":0.01}),
-                     "stretch": ("BOOLEAN", {
-                        "default": True,
-                        "tooltip": "Stretch the sigmas to be in the range [terminal, 1]."
-                    }),
-                     "terminal": (
-                        "FLOAT",
-                        {
-                            "default": 0.1, "min": 0.0, "max": 0.99, "step": 0.01,
-                            "tooltip": "The terminal value of the sigmas after stretching."
-                        },
-                    ),
-                    },
-                "optional": {"latent": ("LATENT",), }
-               }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LTXVScheduler",
+            category="sampling/custom_sampling/schedulers",
+            inputs=[
+                io.Int.Input("steps", default=20, min=1, max=10000),
+                io.Float.Input("max_shift", default=2.05, min=0.0, max=100.0, step=0.01),
+                io.Float.Input("base_shift", default=0.95, min=0.0, max=100.0, step=0.01),
+                io.Boolean.Input(
+                    id="stretch",
+                    default=True,
+                    tooltip="Stretch the sigmas to be in the range [terminal, 1].",
+                ),
+                io.Float.Input(
+                    id="terminal",
+                    default=0.1,
+                    min=0.0,
+                    max=0.99,
+                    step=0.01,
+                    tooltip="The terminal value of the sigmas after stretching.",
+                ),
+                io.Latent.Input("latent", optional=True),
+            ],
+            outputs=[
+                io.Sigmas.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("SIGMAS",)
-    CATEGORY = "sampling/custom_sampling/schedulers"
-
-    FUNCTION = "get_sigmas"
-
-    def get_sigmas(self, steps, max_shift, base_shift, stretch, terminal, latent=None):
+    @classmethod
+    def execute(cls, steps, max_shift, base_shift, stretch, terminal, latent=None) -> io.NodeOutput:
         if latent is None:
             tokens = 4096
         else:
@@ -389,7 +438,7 @@ class LTXVScheduler:
             stretched = 1.0 - (one_minus_z / scale_factor)
             sigmas[non_zero_mask] = stretched
 
-        return (sigmas,)
+        return io.NodeOutput(sigmas)
 
 def encode_single_frame(output_file, image_array: np.ndarray, crf):
     container = av.open(output_file, "w", format="mp4")
@@ -423,52 +472,55 @@ def preprocess(image: torch.Tensor, crf=29):
         return image
 
     image_array = (image[:(image.shape[0] // 2) * 2, :(image.shape[1] // 2) * 2] * 255.0).byte().cpu().numpy()
-    with io.BytesIO() as output_file:
+    with BytesIO() as output_file:
         encode_single_frame(output_file, image_array, crf)
         video_bytes = output_file.getvalue()
-    with io.BytesIO(video_bytes) as video_file:
+    with BytesIO(video_bytes) as video_file:
         image_array = decode_single_frame(video_file)
     tensor = torch.tensor(image_array, dtype=image.dtype, device=image.device) / 255.0
     return tensor
 
 
-class LTXVPreprocess:
+class LTXVPreprocess(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "image": ("IMAGE",),
-                "img_compression": (
-                    "INT",
-                    {
-                        "default": 35,
-                        "min": 0,
-                        "max": 100,
-                        "tooltip": "Amount of compression to apply on image.",
-                    },
+    def define_schema(cls):
+        return io.Schema(
+            node_id="LTXVPreprocess",
+            category="image",
+            inputs=[
+                io.Image.Input("image"),
+                io.Int.Input(
+                    id="img_compression", default=35, min=0, max=100, tooltip="Amount of compression to apply on image."
                 ),
-            }
-        }
+            ],
+            outputs=[
+                io.Image.Output(display_name="output_image"),
+            ],
+        )
 
-    FUNCTION = "preprocess"
-    RETURN_TYPES = ("IMAGE",)
-    RETURN_NAMES = ("output_image",)
-    CATEGORY = "image"
-
-    def preprocess(self, image, img_compression):
+    @classmethod
+    def execute(cls, image, img_compression) -> io.NodeOutput:
         output_images = []
         for i in range(image.shape[0]):
             output_images.append(preprocess(image[i], img_compression))
-        return (torch.stack(output_images),)
+        return io.NodeOutput(torch.stack(output_images))
+
+    preprocess = execute  # TODO: remove
+
+class LtxvExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            EmptyLTXVLatentVideo,
+            LTXVImgToVideo,
+            ModelSamplingLTXV,
+            LTXVConditioning,
+            LTXVScheduler,
+            LTXVAddGuide,
+            LTXVPreprocess,
+            LTXVCropGuides,
+        ]
 
 
-NODE_CLASS_MAPPINGS = {
-    "EmptyLTXVLatentVideo": EmptyLTXVLatentVideo,
-    "LTXVImgToVideo": LTXVImgToVideo,
-    "ModelSamplingLTXV": ModelSamplingLTXV,
-    "LTXVConditioning": LTXVConditioning,
-    "LTXVScheduler": LTXVScheduler,
-    "LTXVAddGuide": LTXVAddGuide,
-    "LTXVPreprocess": LTXVPreprocess,
-    "LTXVCropGuides": LTXVCropGuides,
-}
+async def comfy_entrypoint() -> LtxvExtension:
+    return LtxvExtension()

comfy_extras/nodes_lumina2.py

@@ -1,20 +1,27 @@
-from comfy.comfy_types import IO, ComfyNodeABC, InputTypeDict
+from typing_extensions import override
 import torch
 
+from comfy_api.latest import ComfyExtension, io
 
-class RenormCFG:
+
+class RenormCFG(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                              "cfg_trunc": ("FLOAT", {"default": 100, "min": 0.0, "max": 100.0, "step": 0.01}),
-                              "renorm_cfg": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 100.0, "step": 0.01}),
-                              }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="RenormCFG",
+            category="advanced/model",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input("cfg_trunc", default=100, min=0.0, max=100.0, step=0.01),
+                io.Float.Input("renorm_cfg", default=1.0, min=0.0, max=100.0, step=0.01),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    CATEGORY = "advanced/model"
-
-    def patch(self, model, cfg_trunc, renorm_cfg):
+    @classmethod
+    def execute(cls, model, cfg_trunc, renorm_cfg) -> io.NodeOutput:
         def renorm_cfg_func(args):
             cond_denoised = args["cond_denoised"]
             uncond_denoised = args["uncond_denoised"]
@@ -53,10 +60,10 @@ class RenormCFG:
 
         m = model.clone()
         m.set_model_sampler_cfg_function(renorm_cfg_func)
-        return (m, )
+        return io.NodeOutput(m)
 
 
-class CLIPTextEncodeLumina2(ComfyNodeABC):
+class CLIPTextEncodeLumina2(io.ComfyNode):
     SYSTEM_PROMPT = {
         "superior": "You are an assistant designed to generate superior images with the superior "\
             "degree of image-text alignment based on textual prompts or user prompts.",
@@ -69,36 +76,52 @@ class CLIPTextEncodeLumina2(ComfyNodeABC):
         "Alignment: You are an assistant designed to generate high-quality images with the highest "\
         "degree of image-text alignment based on textual prompts."
     @classmethod
-    def INPUT_TYPES(s) -> InputTypeDict:
-        return {
-            "required": {
-                "system_prompt": (list(CLIPTextEncodeLumina2.SYSTEM_PROMPT.keys()), {"tooltip": CLIPTextEncodeLumina2.SYSTEM_PROMPT_TIP}),
-                "user_prompt": (IO.STRING, {"multiline": True, "dynamicPrompts": True, "tooltip": "The text to be encoded."}),
-                "clip": (IO.CLIP, {"tooltip": "The CLIP model used for encoding the text."})
-            }
-        }
-    RETURN_TYPES = (IO.CONDITIONING,)
-    OUTPUT_TOOLTIPS = ("A conditioning containing the embedded text used to guide the diffusion model.",)
-    FUNCTION = "encode"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodeLumina2",
+            display_name="CLIP Text Encode for Lumina2",
+            category="conditioning",
+            description="Encodes a system prompt and a user prompt using a CLIP model into an embedding "
+                        "that can be used to guide the diffusion model towards generating specific images.",
+            inputs=[
+                io.Combo.Input(
+                    "system_prompt",
+                    options=list(cls.SYSTEM_PROMPT.keys()),
+                    tooltip=cls.SYSTEM_PROMPT_TIP,
+                ),
+                io.String.Input(
+                    "user_prompt",
+                    multiline=True,
+                    dynamic_prompts=True,
+                    tooltip="The text to be encoded.",
+                ),
+                io.Clip.Input("clip", tooltip="The CLIP model used for encoding the text."),
+            ],
+            outputs=[
+                io.Conditioning.Output(
+                    tooltip="A conditioning containing the embedded text used to guide the diffusion model.",
+                ),
+            ],
+        )
 
-    CATEGORY = "conditioning"
-    DESCRIPTION = "Encodes a system prompt and a user prompt using a CLIP model into an embedding that can be used to guide the diffusion model towards generating specific images."
-
-    def encode(self, clip, user_prompt, system_prompt):
+    @classmethod
+    def execute(cls, clip, user_prompt, system_prompt) -> io.NodeOutput:
         if clip is None:
             raise RuntimeError("ERROR: clip input is invalid: None\n\nIf the clip is from a checkpoint loader node your checkpoint does not contain a valid clip or text encoder model.")
-        system_prompt = CLIPTextEncodeLumina2.SYSTEM_PROMPT[system_prompt]
+        system_prompt = cls.SYSTEM_PROMPT[system_prompt]
         prompt = f'{system_prompt} <Prompt Start> {user_prompt}'
         tokens = clip.tokenize(prompt)
-        return (clip.encode_from_tokens_scheduled(tokens), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens))
 
 
-NODE_CLASS_MAPPINGS = {
-    "CLIPTextEncodeLumina2": CLIPTextEncodeLumina2,
-    "RenormCFG": RenormCFG
-}
+class Lumina2Extension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            CLIPTextEncodeLumina2,
+            RenormCFG,
+        ]
 
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "CLIPTextEncodeLumina2": "CLIP Text Encode for Lumina2",
-}
+async def comfy_entrypoint() -> Lumina2Extension:
+    return Lumina2Extension()

comfy_extras/nodes_mahiro.py

@@ -1,17 +1,29 @@
+from typing_extensions import override
 import torch
 import torch.nn.functional as F
 
-class Mahiro:
+from comfy_api.latest import ComfyExtension, io
+
+
+class Mahiro(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"model": ("MODEL",),
-                            }}
-    RETURN_TYPES = ("MODEL",)
-    RETURN_NAMES = ("patched_model",)
-    FUNCTION = "patch"
-    CATEGORY = "_for_testing"
-    DESCRIPTION = "Modify the guidance to scale more on the 'direction' of the positive prompt rather than the difference between the negative prompt."
-    def patch(self, model):
+    def define_schema(cls):
+        return io.Schema(
+            node_id="Mahiro",
+            display_name="Mahiro is so cute that she deserves a better guidance function!! (。・ω・。)",
+            category="_for_testing",
+            description="Modify the guidance to scale more on the 'direction' of the positive prompt rather than the difference between the negative prompt.",
+            inputs=[
+                io.Model.Input("model"),
+            ],
+            outputs=[
+                io.Model.Output(display_name="patched_model"),
+            ],
+            is_experimental=True,
+        )
+
+    @classmethod
+    def execute(cls, model) -> io.NodeOutput:
         m = model.clone()
         def mahiro_normd(args):
             scale: float = args['cond_scale']
@@ -30,12 +42,16 @@ class Mahiro:
             wm = (simsc*cfg + (4-simsc)*leap) / 4
             return wm
         m.set_model_sampler_post_cfg_function(mahiro_normd)
-        return (m, )
+        return io.NodeOutput(m)
 
-NODE_CLASS_MAPPINGS = {
-    "Mahiro": Mahiro
-}
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "Mahiro": "Mahiro is so cute that she deserves a better guidance function!! (。・ω・。)",
-}
+class MahiroExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            Mahiro,
+        ]
+
+
+async def comfy_entrypoint() -> MahiroExtension:
+    return MahiroExtension()

comfy_extras/nodes_mask.py

@@ -12,35 +12,38 @@ from nodes import MAX_RESOLUTION
 def composite(destination, source, x, y, mask = None, multiplier = 8, resize_source = False):
     source = source.to(destination.device)
     if resize_source:
-        source = torch.nn.functional.interpolate(source, size=(destination.shape[2], destination.shape[3]), mode="bilinear")
+        source = torch.nn.functional.interpolate(source, size=(destination.shape[-2], destination.shape[-1]), mode="bilinear")
 
     source = comfy.utils.repeat_to_batch_size(source, destination.shape[0])
 
-    x = max(-source.shape[3] * multiplier, min(x, destination.shape[3] * multiplier))
-    y = max(-source.shape[2] * multiplier, min(y, destination.shape[2] * multiplier))
+    x = max(-source.shape[-1] * multiplier, min(x, destination.shape[-1] * multiplier))
+    y = max(-source.shape[-2] * multiplier, min(y, destination.shape[-2] * multiplier))
 
     left, top = (x // multiplier, y // multiplier)
-    right, bottom = (left + source.shape[3], top + source.shape[2],)
+    right, bottom = (left + source.shape[-1], top + source.shape[-2],)
 
     if mask is None:
         mask = torch.ones_like(source)
     else:
         mask = mask.to(destination.device, copy=True)
-        mask = torch.nn.functional.interpolate(mask.reshape((-1, 1, mask.shape[-2], mask.shape[-1])), size=(source.shape[2], source.shape[3]), mode="bilinear")
+        mask = torch.nn.functional.interpolate(mask.reshape((-1, 1, mask.shape[-2], mask.shape[-1])), size=(source.shape[-2], source.shape[-1]), mode="bilinear")
         mask = comfy.utils.repeat_to_batch_size(mask, source.shape[0])
 
     # calculate the bounds of the source that will be overlapping the destination
     # this prevents the source trying to overwrite latent pixels that are out of bounds
     # of the destination
-    visible_width, visible_height = (destination.shape[3] - left + min(0, x), destination.shape[2] - top + min(0, y),)
+    visible_width, visible_height = (destination.shape[-1] - left + min(0, x), destination.shape[-2] - top + min(0, y),)
 
     mask = mask[:, :, :visible_height, :visible_width]
+    if mask.ndim < source.ndim:
+        mask = mask.unsqueeze(1)
+
     inverse_mask = torch.ones_like(mask) - mask
 
-    source_portion = mask * source[:, :, :visible_height, :visible_width]
-    destination_portion = inverse_mask  * destination[:, :, top:bottom, left:right]
+    source_portion = mask * source[..., :visible_height, :visible_width]
+    destination_portion = inverse_mask  * destination[..., top:bottom, left:right]
 
-    destination[:, :, top:bottom, left:right] = source_portion + destination_portion
+    destination[..., top:bottom, left:right] = source_portion + destination_portion
     return destination
 
 class LatentCompositeMasked:

comfy_extras/nodes_mochi.py

@@ -1,23 +1,40 @@
-import nodes
+from typing_extensions import override
 import torch
 import comfy.model_management
+import nodes
+from comfy_api.latest import ComfyExtension, io
 
-class EmptyMochiLatentVideo:
+
+class EmptyMochiLatentVideo(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "width": ("INT", {"default": 848, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                              "height": ("INT", {"default": 480, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                              "length": ("INT", {"default": 25, "min": 7, "max": nodes.MAX_RESOLUTION, "step": 6}),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096})}}
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "generate"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="EmptyMochiLatentVideo",
+            category="latent/video",
+            inputs=[
+                io.Int.Input("width", default=848, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("height", default=480, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("length", default=25, min=7, max=nodes.MAX_RESOLUTION, step=6),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
-    CATEGORY = "latent/video"
-
-    def generate(self, width, height, length, batch_size=1):
+    @classmethod
+    def execute(cls, width, height, length, batch_size=1) -> io.NodeOutput:
         latent = torch.zeros([batch_size, 12, ((length - 1) // 6) + 1, height // 8, width // 8], device=comfy.model_management.intermediate_device())
-        return ({"samples":latent}, )
+        return io.NodeOutput({"samples": latent})
 
-NODE_CLASS_MAPPINGS = {
-    "EmptyMochiLatentVideo": EmptyMochiLatentVideo,
-}
+
+class MochiExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            EmptyMochiLatentVideo,
+        ]
+
+
+async def comfy_entrypoint() -> MochiExtension:
+    return MochiExtension()

comfy_extras/nodes_model_downscale.py

@@ -1,24 +1,33 @@
+from typing_extensions import override
 import comfy.utils
+from comfy_api.latest import ComfyExtension, io
 
-class PatchModelAddDownscale:
-    upscale_methods = ["bicubic", "nearest-exact", "bilinear", "area", "bislerp"]
+
+class PatchModelAddDownscale(io.ComfyNode):
+    UPSCALE_METHODS = ["bicubic", "nearest-exact", "bilinear", "area", "bislerp"]
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                              "block_number": ("INT", {"default": 3, "min": 1, "max": 32, "step": 1}),
-                              "downscale_factor": ("FLOAT", {"default": 2.0, "min": 0.1, "max": 9.0, "step": 0.001}),
-                              "start_percent": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.001}),
-                              "end_percent": ("FLOAT", {"default": 0.35, "min": 0.0, "max": 1.0, "step": 0.001}),
-                              "downscale_after_skip": ("BOOLEAN", {"default": True}),
-                              "downscale_method": (s.upscale_methods,),
-                              "upscale_method": (s.upscale_methods,),
-                              }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="PatchModelAddDownscale",
+            display_name="PatchModelAddDownscale (Kohya Deep Shrink)",
+            category="model_patches/unet",
+            inputs=[
+                io.Model.Input("model"),
+                io.Int.Input("block_number", default=3, min=1, max=32, step=1),
+                io.Float.Input("downscale_factor", default=2.0, min=0.1, max=9.0, step=0.001),
+                io.Float.Input("start_percent", default=0.0, min=0.0, max=1.0, step=0.001),
+                io.Float.Input("end_percent", default=0.35, min=0.0, max=1.0, step=0.001),
+                io.Boolean.Input("downscale_after_skip", default=True),
+                io.Combo.Input("downscale_method", options=cls.UPSCALE_METHODS),
+                io.Combo.Input("upscale_method", options=cls.UPSCALE_METHODS),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    CATEGORY = "model_patches/unet"
-
-    def patch(self, model, block_number, downscale_factor, start_percent, end_percent, downscale_after_skip, downscale_method, upscale_method):
+    @classmethod
+    def execute(cls, model, block_number, downscale_factor, start_percent, end_percent, downscale_after_skip, downscale_method, upscale_method) -> io.NodeOutput:
         model_sampling = model.get_model_object("model_sampling")
         sigma_start = model_sampling.percent_to_sigma(start_percent)
         sigma_end = model_sampling.percent_to_sigma(end_percent)
@@ -41,13 +50,21 @@ class PatchModelAddDownscale:
         else:
             m.set_model_input_block_patch(input_block_patch)
         m.set_model_output_block_patch(output_block_patch)
-        return (m, )
+        return io.NodeOutput(m)
 
-NODE_CLASS_MAPPINGS = {
-    "PatchModelAddDownscale": PatchModelAddDownscale,
-}
 
 NODE_DISPLAY_NAME_MAPPINGS = {
     # Sampling
-    "PatchModelAddDownscale": "PatchModelAddDownscale (Kohya Deep Shrink)",
+    "PatchModelAddDownscale": "",
 }
+
+class ModelDownscaleExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            PatchModelAddDownscale,
+        ]
+
+
+async def comfy_entrypoint() -> ModelDownscaleExtension:
+    return ModelDownscaleExtension()

comfy_extras/nodes_morphology.py

@@ -1,24 +1,34 @@
 import torch
 import comfy.model_management
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 from kornia.morphology import dilation, erosion, opening, closing, gradient, top_hat, bottom_hat
 import kornia.color
 
 
-class Morphology:
+class Morphology(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"image": ("IMAGE",),
-                                "operation": (["erode",  "dilate", "open", "close", "gradient", "bottom_hat", "top_hat"],),
-                                "kernel_size": ("INT", {"default": 3, "min": 3, "max": 999, "step": 1}),
-                                }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="Morphology",
+            display_name="ImageMorphology",
+            category="image/postprocessing",
+            inputs=[
+                io.Image.Input("image"),
+                io.Combo.Input(
+                    "operation",
+                    options=["erode", "dilate", "open", "close", "gradient", "bottom_hat", "top_hat"],
+                ),
+                io.Int.Input("kernel_size", default=3, min=3, max=999, step=1),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "process"
-
-    CATEGORY = "image/postprocessing"
-
-    def process(self, image, operation, kernel_size):
+    @classmethod
+    def execute(cls, image, operation, kernel_size) -> io.NodeOutput:
         device = comfy.model_management.get_torch_device()
         kernel = torch.ones(kernel_size, kernel_size, device=device)
         image_k = image.to(device).movedim(-1, 1)
@@ -39,49 +49,63 @@ class Morphology:
         else:
             raise ValueError(f"Invalid operation {operation} for morphology. Must be one of 'erode', 'dilate', 'open', 'close', 'gradient', 'tophat', 'bottomhat'")
         img_out = output.to(comfy.model_management.intermediate_device()).movedim(1, -1)
-        return (img_out,)
+        return io.NodeOutput(img_out)
 
 
-class ImageRGBToYUV:
+class ImageRGBToYUV(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "image": ("IMAGE",),
-                              }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageRGBToYUV",
+            category="image/batch",
+            inputs=[
+                io.Image.Input("image"),
+            ],
+            outputs=[
+                io.Image.Output(display_name="Y"),
+                io.Image.Output(display_name="U"),
+                io.Image.Output(display_name="V"),
+            ],
+        )
 
-    RETURN_TYPES = ("IMAGE", "IMAGE", "IMAGE")
-    RETURN_NAMES = ("Y", "U", "V")
-    FUNCTION = "execute"
-
-    CATEGORY = "image/batch"
-
-    def execute(self, image):
+    @classmethod
+    def execute(cls, image) -> io.NodeOutput:
         out = kornia.color.rgb_to_ycbcr(image.movedim(-1, 1)).movedim(1, -1)
-        return (out[..., 0:1].expand_as(image), out[..., 1:2].expand_as(image), out[..., 2:3].expand_as(image))
+        return io.NodeOutput(out[..., 0:1].expand_as(image), out[..., 1:2].expand_as(image), out[..., 2:3].expand_as(image))
 
-class ImageYUVToRGB:
+class ImageYUVToRGB(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"Y": ("IMAGE",),
-                             "U": ("IMAGE",),
-                             "V": ("IMAGE",),
-                              }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageYUVToRGB",
+            category="image/batch",
+            inputs=[
+                io.Image.Input("Y"),
+                io.Image.Input("U"),
+                io.Image.Input("V"),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "execute"
-
-    CATEGORY = "image/batch"
-
-    def execute(self, Y, U, V):
+    @classmethod
+    def execute(cls, Y, U, V) -> io.NodeOutput:
         image = torch.cat([torch.mean(Y, dim=-1, keepdim=True), torch.mean(U, dim=-1, keepdim=True), torch.mean(V, dim=-1, keepdim=True)], dim=-1)
         out = kornia.color.ycbcr_to_rgb(image.movedim(-1, 1)).movedim(1, -1)
-        return (out,)
+        return io.NodeOutput(out)
 
-NODE_CLASS_MAPPINGS = {
-    "Morphology": Morphology,
-    "ImageRGBToYUV": ImageRGBToYUV,
-    "ImageYUVToRGB": ImageYUVToRGB,
-}
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "Morphology": "ImageMorphology",
-}
+class MorphologyExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            Morphology,
+            ImageRGBToYUV,
+            ImageYUVToRGB,
+        ]
+
+
+async def comfy_entrypoint() -> MorphologyExtension:
+    return MorphologyExtension()
+

comfy_extras/nodes_optimalsteps.py

@@ -1,9 +1,12 @@
 # from https://github.com/bebebe666/OptimalSteps
 
-
 import numpy as np
 import torch
 
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
+
+
 def loglinear_interp(t_steps, num_steps):
     """
     Performs log-linear interpolation of a given array of decreasing numbers.
@@ -23,25 +26,28 @@ NOISE_LEVELS = {"FLUX": [0.9968, 0.9886, 0.9819, 0.975, 0.966, 0.9471, 0.9158, 0
 "Chroma": [0.992, 0.99, 0.988, 0.985, 0.982, 0.978, 0.973, 0.968, 0.961, 0.953, 0.943, 0.931, 0.917, 0.9, 0.881, 0.858, 0.832, 0.802, 0.769, 0.731, 0.69, 0.646, 0.599, 0.55, 0.501, 0.451, 0.402, 0.355, 0.311, 0.27, 0.232, 0.199, 0.169, 0.143, 0.12, 0.101, 0.084, 0.07, 0.058, 0.048, 0.001],
 }
 
-class OptimalStepsScheduler:
+class OptimalStepsScheduler(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required":
-                    {"model_type": (["FLUX", "Wan", "Chroma"], ),
-                     "steps": ("INT", {"default": 20, "min": 3, "max": 1000}),
-                     "denoise": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.01}),
-                      }
-               }
-    RETURN_TYPES = ("SIGMAS",)
-    CATEGORY = "sampling/custom_sampling/schedulers"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="OptimalStepsScheduler",
+            category="sampling/custom_sampling/schedulers",
+            inputs=[
+                io.Combo.Input("model_type", options=["FLUX", "Wan", "Chroma"]),
+                io.Int.Input("steps", default=20, min=3, max=1000),
+                io.Float.Input("denoise", default=1.0, min=0.0, max=1.0, step=0.01),
+            ],
+            outputs=[
+                io.Sigmas.Output(),
+            ],
+        )
 
-    FUNCTION = "get_sigmas"
-
-    def get_sigmas(self, model_type, steps, denoise):
+    @classmethod
+    def execute(cls, model_type, steps, denoise) ->io.NodeOutput:
         total_steps = steps
         if denoise < 1.0:
             if denoise <= 0.0:
-                return (torch.FloatTensor([]),)
+                return io.NodeOutput(torch.FloatTensor([]))
             total_steps = round(steps * denoise)
 
         sigmas = NOISE_LEVELS[model_type][:]
@@ -50,8 +56,16 @@ class OptimalStepsScheduler:
 
         sigmas = sigmas[-(total_steps + 1):]
         sigmas[-1] = 0
-        return (torch.FloatTensor(sigmas), )
+        return io.NodeOutput(torch.FloatTensor(sigmas))
 
-NODE_CLASS_MAPPINGS = {
-    "OptimalStepsScheduler": OptimalStepsScheduler,
-}
+
+class OptimalStepsExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            OptimalStepsScheduler,
+        ]
+
+
+async def comfy_entrypoint() -> OptimalStepsExtension:
+    return OptimalStepsExtension()

comfy_extras/nodes_pag.py

@@ -3,25 +3,30 @@
 
 #My modified one here is more basic but has less chances of breaking with ComfyUI updates.
 
+from typing_extensions import override
+
 import comfy.model_patcher
 import comfy.samplers
+from comfy_api.latest import ComfyExtension, io
 
-class PerturbedAttentionGuidance:
+
+class PerturbedAttentionGuidance(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "model": ("MODEL",),
-                "scale": ("FLOAT", {"default": 3.0, "min": 0.0, "max": 100.0, "step": 0.01, "round": 0.01}),
-            }
-        }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="PerturbedAttentionGuidance",
+            category="model_patches/unet",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input("scale", default=3.0, min=0.0, max=100.0, step=0.01, round=0.01),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
-
-    CATEGORY = "model_patches/unet"
-
-    def patch(self, model, scale):
+    @classmethod
+    def execute(cls, model, scale) -> io.NodeOutput:
         unet_block = "middle"
         unet_block_id = 0
         m = model.clone()
@@ -49,8 +54,16 @@ class PerturbedAttentionGuidance:
 
         m.set_model_sampler_post_cfg_function(post_cfg_function)
 
-        return (m,)
+        return io.NodeOutput(m)
 
-NODE_CLASS_MAPPINGS = {
-    "PerturbedAttentionGuidance": PerturbedAttentionGuidance,
-}
+
+class PAGExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            PerturbedAttentionGuidance,
+        ]
+
+
+async def comfy_entrypoint() -> PAGExtension:
+    return PAGExtension()

comfy_extras/nodes_perpneg.py

@@ -5,6 +5,9 @@ import comfy.samplers
 import comfy.utils
 import node_helpers
 import math
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
+
 
 def perp_neg(x, noise_pred_pos, noise_pred_neg, noise_pred_nocond, neg_scale, cond_scale):
     pos = noise_pred_pos - noise_pred_nocond
@@ -16,20 +19,27 @@ def perp_neg(x, noise_pred_pos, noise_pred_neg, noise_pred_nocond, neg_scale, co
     return cfg_result
 
 #TODO: This node should be removed, it has been replaced with PerpNegGuider
-class PerpNeg:
+class PerpNeg(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"model": ("MODEL", ),
-                             "empty_conditioning": ("CONDITIONING", ),
-                             "neg_scale": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 100.0, "step": 0.01}),
-                            }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="PerpNeg",
+            display_name="Perp-Neg (DEPRECATED by PerpNegGuider)",
+            category="_for_testing",
+            inputs=[
+                io.Model.Input("model"),
+                io.Conditioning.Input("empty_conditioning"),
+                io.Float.Input("neg_scale", default=1.0, min=0.0, max=100.0, step=0.01),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+            is_experimental=True,
+            is_deprecated=True,
+        )
 
-    CATEGORY = "_for_testing"
-    DEPRECATED = True
-
-    def patch(self, model, empty_conditioning, neg_scale):
+    @classmethod
+    def execute(cls, model, empty_conditioning, neg_scale) -> io.NodeOutput:
         m = model.clone()
         nocond = comfy.sampler_helpers.convert_cond(empty_conditioning)
 
@@ -50,7 +60,7 @@ class PerpNeg:
 
         m.set_model_sampler_cfg_function(cfg_function)
 
-        return (m, )
+        return io.NodeOutput(m)
 
 
 class Guider_PerpNeg(comfy.samplers.CFGGuider):
@@ -112,35 +122,42 @@ class Guider_PerpNeg(comfy.samplers.CFGGuider):
 
         return cfg_result
 
-class PerpNegGuider:
+class PerpNegGuider(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required":
-                    {"model": ("MODEL",),
-                    "positive": ("CONDITIONING", ),
-                    "negative": ("CONDITIONING", ),
-                    "empty_conditioning": ("CONDITIONING", ),
-                    "cfg": ("FLOAT", {"default": 8.0, "min": 0.0, "max": 100.0, "step":0.1, "round": 0.01}),
-                    "neg_scale": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 100.0, "step": 0.01}),
-                     }
-                }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="PerpNegGuider",
+            category="_for_testing",
+            inputs=[
+                io.Model.Input("model"),
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Conditioning.Input("empty_conditioning"),
+                io.Float.Input("cfg", default=8.0, min=0.0, max=100.0, step=0.1, round=0.01),
+                io.Float.Input("neg_scale", default=1.0, min=0.0, max=100.0, step=0.01),
+            ],
+            outputs=[
+                io.Guider.Output(),
+            ],
+            is_experimental=True,
+        )
 
-    RETURN_TYPES = ("GUIDER",)
-
-    FUNCTION = "get_guider"
-    CATEGORY = "_for_testing"
-
-    def get_guider(self, model, positive, negative, empty_conditioning, cfg, neg_scale):
+    @classmethod
+    def execute(cls, model, positive, negative, empty_conditioning, cfg, neg_scale) -> io.NodeOutput:
         guider = Guider_PerpNeg(model)
         guider.set_conds(positive, negative, empty_conditioning)
         guider.set_cfg(cfg, neg_scale)
-        return (guider,)
+        return io.NodeOutput(guider)
 
-NODE_CLASS_MAPPINGS = {
-    "PerpNeg": PerpNeg,
-    "PerpNegGuider": PerpNegGuider,
-}
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "PerpNeg": "Perp-Neg (DEPRECATED by PerpNegGuider)",
-}
+class PerpNegExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            PerpNeg,
+            PerpNegGuider,
+        ]
+
+
+async def comfy_entrypoint() -> PerpNegExtension:
+    return PerpNegExtension()

comfy_extras/nodes_photomaker.py

@@ -4,6 +4,8 @@ import folder_paths
 import comfy.clip_model
 import comfy.clip_vision
 import comfy.ops
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 # code for model from: https://github.com/TencentARC/PhotoMaker/blob/main/photomaker/model.py under Apache License Version 2.0
 VISION_CONFIG_DICT = {
@@ -116,41 +118,52 @@ class PhotoMakerIDEncoder(comfy.clip_model.CLIPVisionModelProjection):
         return updated_prompt_embeds
 
 
-class PhotoMakerLoader:
+class PhotoMakerLoader(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "photomaker_model_name": (folder_paths.get_filename_list("photomaker"), )}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="PhotoMakerLoader",
+            category="_for_testing/photomaker",
+            inputs=[
+                io.Combo.Input("photomaker_model_name", options=folder_paths.get_filename_list("photomaker")),
+            ],
+            outputs=[
+                io.Photomaker.Output(),
+            ],
+            is_experimental=True,
+        )
 
-    RETURN_TYPES = ("PHOTOMAKER",)
-    FUNCTION = "load_photomaker_model"
-
-    CATEGORY = "_for_testing/photomaker"
-
-    def load_photomaker_model(self, photomaker_model_name):
+    @classmethod
+    def execute(cls, photomaker_model_name):
         photomaker_model_path = folder_paths.get_full_path_or_raise("photomaker", photomaker_model_name)
         photomaker_model = PhotoMakerIDEncoder()
         data = comfy.utils.load_torch_file(photomaker_model_path, safe_load=True)
         if "id_encoder" in data:
             data = data["id_encoder"]
         photomaker_model.load_state_dict(data)
-        return (photomaker_model,)
+        return io.NodeOutput(photomaker_model)
 
 
-class PhotoMakerEncode:
+class PhotoMakerEncode(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "photomaker": ("PHOTOMAKER",),
-                              "image": ("IMAGE",),
-                              "clip": ("CLIP", ),
-                              "text": ("STRING", {"multiline": True, "dynamicPrompts": True, "default": "photograph of photomaker"}),
-                             }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="PhotoMakerEncode",
+            category="_for_testing/photomaker",
+            inputs=[
+                io.Photomaker.Input("photomaker"),
+                io.Image.Input("image"),
+                io.Clip.Input("clip"),
+                io.String.Input("text", multiline=True, dynamic_prompts=True, default="photograph of photomaker"),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+            is_experimental=True,
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "apply_photomaker"
-
-    CATEGORY = "_for_testing/photomaker"
-
-    def apply_photomaker(self, photomaker, image, clip, text):
+    @classmethod
+    def execute(cls, photomaker, image, clip, text):
         special_token = "photomaker"
         pixel_values = comfy.clip_vision.clip_preprocess(image.to(photomaker.load_device)).float()
         try:
@@ -178,11 +191,16 @@ class PhotoMakerEncode:
         else:
             out = cond
 
-        return ([[out, {"pooled_output": pooled}]], )
+        return io.NodeOutput([[out, {"pooled_output": pooled}]])
 
 
-NODE_CLASS_MAPPINGS = {
-    "PhotoMakerLoader": PhotoMakerLoader,
-    "PhotoMakerEncode": PhotoMakerEncode,
-}
+class PhotomakerExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            PhotoMakerLoader,
+            PhotoMakerEncode,
+        ]
 
+async def comfy_entrypoint() -> PhotomakerExtension:
+    return PhotomakerExtension()

comfy_extras/nodes_pixart.py

@@ -1,24 +1,38 @@
-from nodes import MAX_RESOLUTION
+from typing_extensions import override
+import nodes
+from comfy_api.latest import ComfyExtension, io
 
-class CLIPTextEncodePixArtAlpha:
+class CLIPTextEncodePixArtAlpha(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "width": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            "height": ("INT", {"default": 1024.0, "min": 0, "max": MAX_RESOLUTION}),
-            # "aspect_ratio": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 10.0, "step": 0.01}),
-            "text": ("STRING", {"multiline": True, "dynamicPrompts": True}), "clip": ("CLIP", ),
-            }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodePixArtAlpha",
+            category="advanced/conditioning",
+            description="Encodes text and sets the resolution conditioning for PixArt Alpha. Does not apply to PixArt Sigma.",
+            inputs=[
+                io.Int.Input("width", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                io.Int.Input("height", default=1024, min=0, max=nodes.MAX_RESOLUTION),
+                # "aspect_ratio": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 10.0, "step": 0.01}),
+                io.String.Input("text", multiline=True, dynamic_prompts=True),
+                io.Clip.Input("clip"),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
-    CATEGORY = "advanced/conditioning"
-    DESCRIPTION = "Encodes text and sets the resolution conditioning for PixArt Alpha. Does not apply to PixArt Sigma."
-
-    def encode(self, clip, width, height, text):
+    @classmethod
+    def execute(cls, clip, width, height, text):
         tokens = clip.tokenize(text)
-        return (clip.encode_from_tokens_scheduled(tokens, add_dict={"width": width, "height": height}),)
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens, add_dict={"width": width, "height": height}))
 
-NODE_CLASS_MAPPINGS = {
-    "CLIPTextEncodePixArtAlpha": CLIPTextEncodePixArtAlpha,
-}
+
+class PixArtExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            CLIPTextEncodePixArtAlpha,
+        ]
+
+async def comfy_entrypoint() -> PixArtExtension:
+    return PixArtExtension()

comfy_extras/nodes_post_processing.py

@@ -1,3 +1,4 @@
+from typing_extensions import override
 import numpy as np
 import torch
 import torch.nn.functional as F
@@ -7,33 +8,27 @@ import math
 import comfy.utils
 import comfy.model_management
 import node_helpers
+from comfy_api.latest import ComfyExtension, io
 
-class Blend:
-    def __init__(self):
-        pass
+class Blend(io.ComfyNode):
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageBlend",
+            category="image/postprocessing",
+            inputs=[
+                io.Image.Input("image1"),
+                io.Image.Input("image2"),
+                io.Float.Input("blend_factor", default=0.5, min=0.0, max=1.0, step=0.01),
+                io.Combo.Input("blend_mode", options=["normal", "multiply", "screen", "overlay", "soft_light", "difference"]),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "image1": ("IMAGE",),
-                "image2": ("IMAGE",),
-                "blend_factor": ("FLOAT", {
-                    "default": 0.5,
-                    "min": 0.0,
-                    "max": 1.0,
-                    "step": 0.01
-                }),
-                "blend_mode": (["normal", "multiply", "screen", "overlay", "soft_light", "difference"],),
-            },
-        }
-
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "blend_images"
-
-    CATEGORY = "image/postprocessing"
-
-    def blend_images(self, image1: torch.Tensor, image2: torch.Tensor, blend_factor: float, blend_mode: str):
+    def execute(cls, image1: torch.Tensor, image2: torch.Tensor, blend_factor: float, blend_mode: str) -> io.NodeOutput:
         image1, image2 = node_helpers.image_alpha_fix(image1, image2)
         image2 = image2.to(image1.device)
         if image1.shape != image2.shape:
@@ -41,12 +36,13 @@ class Blend:
             image2 = comfy.utils.common_upscale(image2, image1.shape[2], image1.shape[1], upscale_method='bicubic', crop='center')
             image2 = image2.permute(0, 2, 3, 1)
 
-        blended_image = self.blend_mode(image1, image2, blend_mode)
+        blended_image = cls.blend_mode(image1, image2, blend_mode)
         blended_image = image1 * (1 - blend_factor) + blended_image * blend_factor
         blended_image = torch.clamp(blended_image, 0, 1)
-        return (blended_image,)
+        return io.NodeOutput(blended_image)
 
-    def blend_mode(self, img1, img2, mode):
+    @classmethod
+    def blend_mode(cls, img1, img2, mode):
         if mode == "normal":
             return img2
         elif mode == "multiply":
@@ -56,13 +52,13 @@ class Blend:
         elif mode == "overlay":
             return torch.where(img1 <= 0.5, 2 * img1 * img2, 1 - 2 * (1 - img1) * (1 - img2))
         elif mode == "soft_light":
-            return torch.where(img2 <= 0.5, img1 - (1 - 2 * img2) * img1 * (1 - img1), img1 + (2 * img2 - 1) * (self.g(img1) - img1))
+            return torch.where(img2 <= 0.5, img1 - (1 - 2 * img2) * img1 * (1 - img1), img1 + (2 * img2 - 1) * (cls.g(img1) - img1))
         elif mode == "difference":
             return img1 - img2
-        else:
-            raise ValueError(f"Unsupported blend mode: {mode}")
+        raise ValueError(f"Unsupported blend mode: {mode}")
 
-    def g(self, x):
+    @classmethod
+    def g(cls, x):
         return torch.where(x <= 0.25, ((16 * x - 12) * x + 4) * x, torch.sqrt(x))
 
 def gaussian_kernel(kernel_size: int, sigma: float, device=None):
@@ -71,38 +67,26 @@ def gaussian_kernel(kernel_size: int, sigma: float, device=None):
     g = torch.exp(-(d * d) / (2.0 * sigma * sigma))
     return g / g.sum()
 
-class Blur:
-    def __init__(self):
-        pass
+class Blur(io.ComfyNode):
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageBlur",
+            category="image/postprocessing",
+            inputs=[
+                io.Image.Input("image"),
+                io.Int.Input("blur_radius", default=1, min=1, max=31, step=1),
+                io.Float.Input("sigma", default=1.0, min=0.1, max=10.0, step=0.1),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "image": ("IMAGE",),
-                "blur_radius": ("INT", {
-                    "default": 1,
-                    "min": 1,
-                    "max": 31,
-                    "step": 1
-                }),
-                "sigma": ("FLOAT", {
-                    "default": 1.0,
-                    "min": 0.1,
-                    "max": 10.0,
-                    "step": 0.1
-                }),
-            },
-        }
-
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "blur"
-
-    CATEGORY = "image/postprocessing"
-
-    def blur(self, image: torch.Tensor, blur_radius: int, sigma: float):
+    def execute(cls, image: torch.Tensor, blur_radius: int, sigma: float) -> io.NodeOutput:
         if blur_radius == 0:
-            return (image,)
+            return io.NodeOutput(image)
 
         image = image.to(comfy.model_management.get_torch_device())
         batch_size, height, width, channels = image.shape
@@ -115,31 +99,24 @@ class Blur:
         blurred = F.conv2d(padded_image, kernel, padding=kernel_size // 2, groups=channels)[:,:,blur_radius:-blur_radius, blur_radius:-blur_radius]
         blurred = blurred.permute(0, 2, 3, 1)
 
-        return (blurred.to(comfy.model_management.intermediate_device()),)
+        return io.NodeOutput(blurred.to(comfy.model_management.intermediate_device()))
 
-class Quantize:
-    def __init__(self):
-        pass
 
+class Quantize(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "image": ("IMAGE",),
-                "colors": ("INT", {
-                    "default": 256,
-                    "min": 1,
-                    "max": 256,
-                    "step": 1
-                }),
-                "dither": (["none", "floyd-steinberg", "bayer-2", "bayer-4", "bayer-8", "bayer-16"],),
-            },
-        }
-
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "quantize"
-
-    CATEGORY = "image/postprocessing"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageQuantize",
+            category="image/postprocessing",
+            inputs=[
+                io.Image.Input("image"),
+                io.Int.Input("colors", default=256, min=1, max=256, step=1),
+                io.Combo.Input("dither", options=["none", "floyd-steinberg", "bayer-2", "bayer-4", "bayer-8", "bayer-16"]),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
     @staticmethod
     def bayer(im, pal_im, order):
@@ -167,7 +144,8 @@ class Quantize:
         im = im.quantize(palette=pal_im, dither=Image.Dither.NONE)
         return im
 
-    def quantize(self, image: torch.Tensor, colors: int, dither: str):
+    @classmethod
+    def execute(cls, image: torch.Tensor, colors: int, dither: str) -> io.NodeOutput:
         batch_size, height, width, _ = image.shape
         result = torch.zeros_like(image)
 
@@ -187,46 +165,29 @@ class Quantize:
             quantized_array = torch.tensor(np.array(quantized_image.convert("RGB"))).float() / 255
             result[b] = quantized_array
 
-        return (result,)
+        return io.NodeOutput(result)
 
-class Sharpen:
-    def __init__(self):
-        pass
+class Sharpen(io.ComfyNode):
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageSharpen",
+            category="image/postprocessing",
+            inputs=[
+                io.Image.Input("image"),
+                io.Int.Input("sharpen_radius", default=1, min=1, max=31, step=1),
+                io.Float.Input("sigma", default=1.0, min=0.1, max=10.0, step=0.01),
+                io.Float.Input("alpha", default=1.0, min=0.0, max=5.0, step=0.01),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
     @classmethod
-    def INPUT_TYPES(s):
-        return {
-            "required": {
-                "image": ("IMAGE",),
-                "sharpen_radius": ("INT", {
-                    "default": 1,
-                    "min": 1,
-                    "max": 31,
-                    "step": 1
-                }),
-                "sigma": ("FLOAT", {
-                    "default": 1.0,
-                    "min": 0.1,
-                    "max": 10.0,
-                    "step": 0.01
-                }),
-                "alpha": ("FLOAT", {
-                    "default": 1.0,
-                    "min": 0.0,
-                    "max": 5.0,
-                    "step": 0.01
-                }),
-            },
-        }
-
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "sharpen"
-
-    CATEGORY = "image/postprocessing"
-
-    def sharpen(self, image: torch.Tensor, sharpen_radius: int, sigma:float, alpha: float):
+    def execute(cls, image: torch.Tensor, sharpen_radius: int, sigma:float, alpha: float) -> io.NodeOutput:
         if sharpen_radius == 0:
-            return (image,)
+            return io.NodeOutput(image)
 
         batch_size, height, width, channels = image.shape
         image = image.to(comfy.model_management.get_torch_device())
@@ -245,23 +206,29 @@ class Sharpen:
 
         result = torch.clamp(sharpened, 0, 1)
 
-        return (result.to(comfy.model_management.intermediate_device()),)
+        return io.NodeOutput(result.to(comfy.model_management.intermediate_device()))
 
-class ImageScaleToTotalPixels:
+class ImageScaleToTotalPixels(io.ComfyNode):
     upscale_methods = ["nearest-exact", "bilinear", "area", "bicubic", "lanczos"]
     crop_methods = ["disabled", "center"]
 
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "image": ("IMAGE",), "upscale_method": (s.upscale_methods,),
-                              "megapixels": ("FLOAT", {"default": 1.0, "min": 0.01, "max": 16.0, "step": 0.01}),
-                            }}
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "upscale"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageScaleToTotalPixels",
+            category="image/upscaling",
+            inputs=[
+                io.Image.Input("image"),
+                io.Combo.Input("upscale_method", options=cls.upscale_methods),
+                io.Float.Input("megapixels", default=1.0, min=0.01, max=16.0, step=0.01),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
-    CATEGORY = "image/upscaling"
-
-    def upscale(self, image, upscale_method, megapixels):
+    @classmethod
+    def execute(cls, image, upscale_method, megapixels) -> io.NodeOutput:
         samples = image.movedim(-1,1)
         total = int(megapixels * 1024 * 1024)
 
@@ -271,12 +238,18 @@ class ImageScaleToTotalPixels:
 
         s = comfy.utils.common_upscale(samples, width, height, upscale_method, "disabled")
         s = s.movedim(1,-1)
-        return (s,)
+        return io.NodeOutput(s)
 
-NODE_CLASS_MAPPINGS = {
-    "ImageBlend": Blend,
-    "ImageBlur": Blur,
-    "ImageQuantize": Quantize,
-    "ImageSharpen": Sharpen,
-    "ImageScaleToTotalPixels": ImageScaleToTotalPixels,
-}
+class PostProcessingExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            Blend,
+            Blur,
+            Quantize,
+            Sharpen,
+            ImageScaleToTotalPixels,
+        ]
+
+async def comfy_entrypoint() -> PostProcessingExtension:
+    return PostProcessingExtension()

comfy_extras/nodes_preview_any.py

@@ -25,7 +25,7 @@ class PreviewAny():
             value = str(source)
         elif source is not None:
             try:
-                value = json.dumps(source)
+                value = json.dumps(source, indent=4)
             except Exception:
                 try:
                     value = str(source)

comfy_extras/nodes_qwen.py

@@ -1,24 +1,29 @@
 import node_helpers
 import comfy.utils
 import math
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 
-class TextEncodeQwenImageEdit:
+class TextEncodeQwenImageEdit(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "prompt": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            },
-            "optional": {"vae": ("VAE", ),
-                         "image": ("IMAGE", ),}}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="TextEncodeQwenImageEdit",
+            category="advanced/conditioning",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.String.Input("prompt", multiline=True, dynamic_prompts=True),
+                io.Vae.Input("vae", optional=True),
+                io.Image.Input("image", optional=True),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
-
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, prompt, vae=None, image=None):
+    @classmethod
+    def execute(cls, clip, prompt, vae=None, image=None) -> io.NodeOutput:
         ref_latent = None
         if image is None:
             images = []
@@ -40,28 +45,30 @@ class TextEncodeQwenImageEdit:
         conditioning = clip.encode_from_tokens_scheduled(tokens)
         if ref_latent is not None:
             conditioning = node_helpers.conditioning_set_values(conditioning, {"reference_latents": [ref_latent]}, append=True)
-        return (conditioning, )
+        return io.NodeOutput(conditioning)
 
 
-class TextEncodeQwenImageEditPlus:
+class TextEncodeQwenImageEditPlus(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "prompt": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            },
-            "optional": {"vae": ("VAE", ),
-                         "image1": ("IMAGE", ),
-                         "image2": ("IMAGE", ),
-                         "image3": ("IMAGE", ),
-                         }}
+    def define_schema(cls):
+        return io.Schema(
+            node_id="TextEncodeQwenImageEditPlus",
+            category="advanced/conditioning",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.String.Input("prompt", multiline=True, dynamic_prompts=True),
+                io.Vae.Input("vae", optional=True),
+                io.Image.Input("image1", optional=True),
+                io.Image.Input("image2", optional=True),
+                io.Image.Input("image3", optional=True),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
-
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, prompt, vae=None, image1=None, image2=None, image3=None):
+    @classmethod
+    def execute(cls, clip, prompt, vae=None, image1=None, image2=None, image3=None) -> io.NodeOutput:
         ref_latents = []
         images = [image1, image2, image3]
         images_vl = []
@@ -94,10 +101,17 @@ class TextEncodeQwenImageEditPlus:
         conditioning = clip.encode_from_tokens_scheduled(tokens)
         if len(ref_latents) > 0:
             conditioning = node_helpers.conditioning_set_values(conditioning, {"reference_latents": ref_latents}, append=True)
-        return (conditioning, )
+        return io.NodeOutput(conditioning)
 
 
-NODE_CLASS_MAPPINGS = {
-    "TextEncodeQwenImageEdit": TextEncodeQwenImageEdit,
-    "TextEncodeQwenImageEditPlus": TextEncodeQwenImageEditPlus,
-}
+class QwenExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            TextEncodeQwenImageEdit,
+            TextEncodeQwenImageEditPlus,
+        ]
+
+
+async def comfy_entrypoint() -> QwenExtension:
+    return QwenExtension()

comfy_extras/nodes_rebatch.py

@@ -1,18 +1,25 @@
+from typing_extensions import override
 import torch
 
-class LatentRebatch:
+from comfy_api.latest import ComfyExtension, io
+
+
+class LatentRebatch(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "latents": ("LATENT",),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
-                              }}
-    RETURN_TYPES = ("LATENT",)
-    INPUT_IS_LIST = True
-    OUTPUT_IS_LIST = (True, )
-
-    FUNCTION = "rebatch"
-
-    CATEGORY = "latent/batch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="RebatchLatents",
+            display_name="Rebatch Latents",
+            category="latent/batch",
+            is_input_list=True,
+            inputs=[
+                io.Latent.Input("latents"),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+            ],
+            outputs=[
+                io.Latent.Output(is_output_list=True),
+            ],
+        )
 
     @staticmethod
     def get_batch(latents, list_ind, offset):
@@ -53,7 +60,8 @@ class LatentRebatch:
         result = [torch.cat((b1, b2)) if torch.is_tensor(b1) else b1 + b2 for b1, b2 in zip(batch1, batch2)]
         return result
 
-    def rebatch(self, latents, batch_size):
+    @classmethod
+    def execute(cls, latents, batch_size):
         batch_size = batch_size[0]
 
         output_list = []
@@ -63,24 +71,24 @@ class LatentRebatch:
         for i in range(len(latents)):
             # fetch new entry of list
             #samples, masks, indices = self.get_batch(latents, i)
-            next_batch = self.get_batch(latents, i, processed)
+            next_batch = cls.get_batch(latents, i, processed)
             processed += len(next_batch[2])
             # set to current if current is None
             if current_batch[0] is None:
                 current_batch = next_batch
             # add previous to list if dimensions do not match
             elif next_batch[0].shape[-1] != current_batch[0].shape[-1] or next_batch[0].shape[-2] != current_batch[0].shape[-2]:
-                sliced, _ = self.slice_batch(current_batch, 1, batch_size)
+                sliced, _ = cls.slice_batch(current_batch, 1, batch_size)
                 output_list.append({'samples': sliced[0][0], 'noise_mask': sliced[1][0], 'batch_index': sliced[2][0]})
                 current_batch = next_batch
             # cat if everything checks out
             else:
-                current_batch = self.cat_batch(current_batch, next_batch)
+                current_batch = cls.cat_batch(current_batch, next_batch)
 
             # add to list if dimensions gone above target batch size
             if current_batch[0].shape[0] > batch_size:
                 num = current_batch[0].shape[0] // batch_size
-                sliced, remainder = self.slice_batch(current_batch, num, batch_size)
+                sliced, remainder = cls.slice_batch(current_batch, num, batch_size)
 
                 for i in range(num):
                     output_list.append({'samples': sliced[0][i], 'noise_mask': sliced[1][i], 'batch_index': sliced[2][i]})
@@ -89,7 +97,7 @@ class LatentRebatch:
 
         #add remainder
         if current_batch[0] is not None:
-            sliced, _ = self.slice_batch(current_batch, 1, batch_size)
+            sliced, _ = cls.slice_batch(current_batch, 1, batch_size)
             output_list.append({'samples': sliced[0][0], 'noise_mask': sliced[1][0], 'batch_index': sliced[2][0]})
 
         #get rid of empty masks
@@ -97,23 +105,27 @@ class LatentRebatch:
             if s['noise_mask'].mean() == 1.0:
                 del s['noise_mask']
 
-        return (output_list,)
+        return io.NodeOutput(output_list)
 
-class ImageRebatch:
+class ImageRebatch(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "images": ("IMAGE",),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
-                              }}
-    RETURN_TYPES = ("IMAGE",)
-    INPUT_IS_LIST = True
-    OUTPUT_IS_LIST = (True, )
+    def define_schema(cls):
+        return io.Schema(
+            node_id="RebatchImages",
+            display_name="Rebatch Images",
+            category="image/batch",
+            is_input_list=True,
+            inputs=[
+                io.Image.Input("images"),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+            ],
+            outputs=[
+                io.Image.Output(is_output_list=True),
+            ],
+        )
 
-    FUNCTION = "rebatch"
-
-    CATEGORY = "image/batch"
-
-    def rebatch(self, images, batch_size):
+    @classmethod
+    def execute(cls, images, batch_size):
         batch_size = batch_size[0]
 
         output_list = []
@@ -125,14 +137,17 @@ class ImageRebatch:
         for i in range(0, len(all_images), batch_size):
             output_list.append(torch.cat(all_images[i:i+batch_size], dim=0))
 
-        return (output_list,)
+        return io.NodeOutput(output_list)
 
-NODE_CLASS_MAPPINGS = {
-    "RebatchLatents": LatentRebatch,
-    "RebatchImages": ImageRebatch,
-}
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "RebatchLatents": "Rebatch Latents",
-    "RebatchImages": "Rebatch Images",
-}
+class RebatchExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            LatentRebatch,
+            ImageRebatch,
+        ]
+
+
+async def comfy_entrypoint() -> RebatchExtension:
+    return RebatchExtension()

comfy_extras/nodes_sag.py

@@ -2,10 +2,13 @@ import torch
 from torch import einsum
 import torch.nn.functional as F
 import math
+from typing_extensions import override
 
 from einops import rearrange, repeat
 from comfy.ldm.modules.attention import optimized_attention
 import comfy.samplers
+from comfy_api.latest import ComfyExtension, io
+
 
 # from comfy/ldm/modules/attention.py
 # but modified to return attention scores as well as output
@@ -104,19 +107,26 @@ def gaussian_blur_2d(img, kernel_size, sigma):
     img = F.conv2d(img, kernel2d, groups=img.shape[-3])
     return img
 
-class SelfAttentionGuidance:
+class SelfAttentionGuidance(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                             "scale": ("FLOAT", {"default": 0.5, "min": -2.0, "max": 5.0, "step": 0.01}),
-                             "blur_sigma": ("FLOAT", {"default": 2.0, "min": 0.0, "max": 10.0, "step": 0.1}),
-                              }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="SelfAttentionGuidance",
+            display_name="Self-Attention Guidance",
+            category="_for_testing",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input("scale", default=0.5, min=-2.0, max=5.0, step=0.01),
+                io.Float.Input("blur_sigma", default=2.0, min=0.0, max=10.0, step=0.1),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+            is_experimental=True,
+        )
 
-    CATEGORY = "_for_testing"
-
-    def patch(self, model, scale, blur_sigma):
+    @classmethod
+    def execute(cls, model, scale, blur_sigma):
         m = model.clone()
 
         attn_scores = None
@@ -170,12 +180,16 @@ class SelfAttentionGuidance:
         # unet.mid_block.attentions[0].transformer_blocks[0].attn1.patch
         m.set_model_attn1_replace(attn_and_record, "middle", 0, 0)
 
-        return (m, )
+        return io.NodeOutput(m)
 
-NODE_CLASS_MAPPINGS = {
-    "SelfAttentionGuidance": SelfAttentionGuidance,
-}
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "SelfAttentionGuidance": "Self-Attention Guidance",
-}
+class SagExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            SelfAttentionGuidance,
+        ]
+
+
+async def comfy_entrypoint() -> SagExtension:
+    return SagExtension()

comfy_extras/nodes_sd3.py

@@ -3,64 +3,83 @@ import comfy.sd
 import comfy.model_management
 import nodes
 import torch
-import comfy_extras.nodes_slg
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
+from comfy_extras.nodes_slg import SkipLayerGuidanceDiT
 
 
-class TripleCLIPLoader:
+class TripleCLIPLoader(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "clip_name1": (folder_paths.get_filename_list("text_encoders"), ), "clip_name2": (folder_paths.get_filename_list("text_encoders"), ), "clip_name3": (folder_paths.get_filename_list("text_encoders"), )
-                             }}
-    RETURN_TYPES = ("CLIP",)
-    FUNCTION = "load_clip"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="TripleCLIPLoader",
+            category="advanced/loaders",
+            description="[Recipes]\n\nsd3: clip-l, clip-g, t5",
+            inputs=[
+                io.Combo.Input("clip_name1", options=folder_paths.get_filename_list("text_encoders")),
+                io.Combo.Input("clip_name2", options=folder_paths.get_filename_list("text_encoders")),
+                io.Combo.Input("clip_name3", options=folder_paths.get_filename_list("text_encoders")),
+            ],
+            outputs=[
+                io.Clip.Output(),
+            ],
+        )
 
-    CATEGORY = "advanced/loaders"
-
-    DESCRIPTION = "[Recipes]\n\nsd3: clip-l, clip-g, t5"
-
-    def load_clip(self, clip_name1, clip_name2, clip_name3):
+    @classmethod
+    def execute(cls, clip_name1, clip_name2, clip_name3) -> io.NodeOutput:
         clip_path1 = folder_paths.get_full_path_or_raise("text_encoders", clip_name1)
         clip_path2 = folder_paths.get_full_path_or_raise("text_encoders", clip_name2)
         clip_path3 = folder_paths.get_full_path_or_raise("text_encoders", clip_name3)
         clip = comfy.sd.load_clip(ckpt_paths=[clip_path1, clip_path2, clip_path3], embedding_directory=folder_paths.get_folder_paths("embeddings"))
-        return (clip,)
+        return io.NodeOutput(clip)
+
+    load_clip = execute  # TODO: remove
 
 
-class EmptySD3LatentImage:
-    def __init__(self):
-        self.device = comfy.model_management.intermediate_device()
+class EmptySD3LatentImage(io.ComfyNode):
+    @classmethod
+    def define_schema(cls):
+        return io.Schema(
+            node_id="EmptySD3LatentImage",
+            category="latent/sd3",
+            inputs=[
+                io.Int.Input("width", default=1024, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("height", default=1024, min=16, max=nodes.MAX_RESOLUTION, step=16),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+            ],
+            outputs=[
+                io.Latent.Output(),
+            ],
+        )
 
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "width": ("INT", {"default": 1024, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                              "height": ("INT", {"default": 1024, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 16}),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096})}}
-    RETURN_TYPES = ("LATENT",)
-    FUNCTION = "generate"
+    def execute(cls, width, height, batch_size=1) -> io.NodeOutput:
+        latent = torch.zeros([batch_size, 16, height // 8, width // 8], device=comfy.model_management.intermediate_device())
+        return io.NodeOutput({"samples":latent})
 
-    CATEGORY = "latent/sd3"
-
-    def generate(self, width, height, batch_size=1):
-        latent = torch.zeros([batch_size, 16, height // 8, width // 8], device=self.device)
-        return ({"samples":latent}, )
+    generate = execute  # TODO: remove
 
 
-class CLIPTextEncodeSD3:
+class CLIPTextEncodeSD3(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {
-            "clip": ("CLIP", ),
-            "clip_l": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "clip_g": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "t5xxl": ("STRING", {"multiline": True, "dynamicPrompts": True}),
-            "empty_padding": (["none", "empty_prompt"], )
-            }}
-    RETURN_TYPES = ("CONDITIONING",)
-    FUNCTION = "encode"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="CLIPTextEncodeSD3",
+            category="advanced/conditioning",
+            inputs=[
+                io.Clip.Input("clip"),
+                io.String.Input("clip_l", multiline=True, dynamic_prompts=True),
+                io.String.Input("clip_g", multiline=True, dynamic_prompts=True),
+                io.String.Input("t5xxl", multiline=True, dynamic_prompts=True),
+                io.Combo.Input("empty_padding", options=["none", "empty_prompt"]),
+            ],
+            outputs=[
+                io.Conditioning.Output(),
+            ],
+        )
 
-    CATEGORY = "advanced/conditioning"
-
-    def encode(self, clip, clip_l, clip_g, t5xxl, empty_padding):
+    @classmethod
+    def execute(cls, clip, clip_l, clip_g, t5xxl, empty_padding) -> io.NodeOutput:
         no_padding = empty_padding == "none"
 
         tokens = clip.tokenize(clip_g)
@@ -82,57 +101,112 @@ class CLIPTextEncodeSD3:
                 tokens["l"] += empty["l"]
             while len(tokens["l"]) > len(tokens["g"]):
                 tokens["g"] += empty["g"]
-        return (clip.encode_from_tokens_scheduled(tokens), )
+        return io.NodeOutput(clip.encode_from_tokens_scheduled(tokens))
+
+    encode = execute  # TODO: remove
 
 
-class ControlNetApplySD3(nodes.ControlNetApplyAdvanced):
+class ControlNetApplySD3(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"positive": ("CONDITIONING", ),
-                             "negative": ("CONDITIONING", ),
-                             "control_net": ("CONTROL_NET", ),
-                             "vae": ("VAE", ),
-                             "image": ("IMAGE", ),
-                             "strength": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 10.0, "step": 0.01}),
-                             "start_percent": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.001}),
-                             "end_percent": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.001})
-                             }}
-    CATEGORY = "conditioning/controlnet"
-    DEPRECATED = True
+    def define_schema(cls) -> io.Schema:
+        return io.Schema(
+            node_id="ControlNetApplySD3",
+            display_name="Apply Controlnet with VAE",
+            category="conditioning/controlnet",
+            inputs=[
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.ControlNet.Input("control_net"),
+                io.Vae.Input("vae"),
+                io.Image.Input("image"),
+                io.Float.Input("strength", default=1.0, min=0.0, max=10.0, step=0.01),
+                io.Float.Input("start_percent", default=0.0, min=0.0, max=1.0, step=0.001),
+                io.Float.Input("end_percent", default=1.0, min=0.0, max=1.0, step=0.001),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+            ],
+            is_deprecated=True,
+        )
+
+    @classmethod
+    def execute(cls, positive, negative, control_net, image, strength, start_percent, end_percent, vae=None) -> io.NodeOutput:
+        if strength == 0:
+            return io.NodeOutput(positive, negative)
+
+        control_hint = image.movedim(-1, 1)
+        cnets = {}
+
+        out = []
+        for conditioning in [positive, negative]:
+            c = []
+            for t in conditioning:
+                d = t[1].copy()
+
+                prev_cnet = d.get('control', None)
+                if prev_cnet in cnets:
+                    c_net = cnets[prev_cnet]
+                else:
+                    c_net = control_net.copy().set_cond_hint(control_hint, strength, (start_percent, end_percent),
+                                                             vae=vae, extra_concat=[])
+                    c_net.set_previous_controlnet(prev_cnet)
+                    cnets[prev_cnet] = c_net
+
+                d['control'] = c_net
+                d['control_apply_to_uncond'] = False
+                n = [t[0], d]
+                c.append(n)
+            out.append(c)
+        return io.NodeOutput(out[0], out[1])
+
+    apply_controlnet = execute  # TODO: remove
 
 
-class SkipLayerGuidanceSD3(comfy_extras.nodes_slg.SkipLayerGuidanceDiT):
+class SkipLayerGuidanceSD3(io.ComfyNode):
     '''
     Enhance guidance towards detailed dtructure by having another set of CFG negative with skipped layers.
     Inspired by Perturbed Attention Guidance (https://arxiv.org/abs/2403.17377)
     Experimental implementation by Dango233@StabilityAI.
     '''
+
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"model": ("MODEL", ),
-                             "layers": ("STRING", {"default": "7, 8, 9", "multiline": False}),
-                             "scale": ("FLOAT", {"default": 3.0, "min": 0.0, "max": 10.0, "step": 0.1}),
-                             "start_percent": ("FLOAT", {"default": 0.01, "min": 0.0, "max": 1.0, "step": 0.001}),
-                             "end_percent": ("FLOAT", {"default": 0.15, "min": 0.0, "max": 1.0, "step": 0.001})
-                                }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "skip_guidance_sd3"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="SkipLayerGuidanceSD3",
+            category="advanced/guidance",
+            description="Generic version of SkipLayerGuidance node that can be used on every DiT model.",
+            inputs=[
+                io.Model.Input("model"),
+                io.String.Input("layers", default="7, 8, 9", multiline=False),
+                io.Float.Input("scale", default=3.0, min=0.0, max=10.0, step=0.1),
+                io.Float.Input("start_percent", default=0.01, min=0.0, max=1.0, step=0.001),
+                io.Float.Input("end_percent", default=0.15, min=0.0, max=1.0, step=0.001),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+            is_experimental=True,
+        )
 
-    CATEGORY = "advanced/guidance"
+    @classmethod
+    def execute(cls, model, layers, scale, start_percent, end_percent) -> io.NodeOutput:
+        return SkipLayerGuidanceDiT().execute(model=model, scale=scale, start_percent=start_percent, end_percent=end_percent, double_layers=layers)
 
-    def skip_guidance_sd3(self, model, layers, scale, start_percent, end_percent):
-        return self.skip_guidance(model=model, scale=scale, start_percent=start_percent, end_percent=end_percent, double_layers=layers)
+    skip_guidance_sd3 = execute  # TODO: remove
 
 
-NODE_CLASS_MAPPINGS = {
-    "TripleCLIPLoader": TripleCLIPLoader,
-    "EmptySD3LatentImage": EmptySD3LatentImage,
-    "CLIPTextEncodeSD3": CLIPTextEncodeSD3,
-    "ControlNetApplySD3": ControlNetApplySD3,
-    "SkipLayerGuidanceSD3": SkipLayerGuidanceSD3,
-}
+class SD3Extension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            TripleCLIPLoader,
+            EmptySD3LatentImage,
+            CLIPTextEncodeSD3,
+            ControlNetApplySD3,
+            SkipLayerGuidanceSD3,
+        ]
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    # Sampling
-    "ControlNetApplySD3": "Apply Controlnet with VAE",
-}
+
+async def comfy_entrypoint() -> SD3Extension:
+    return SD3Extension()

comfy_extras/nodes_sdupscale.py

@@ -1,23 +1,31 @@
+from typing_extensions import override
+
 import torch
 import comfy.utils
+from comfy_api.latest import ComfyExtension, io
 
-class SD_4XUpscale_Conditioning:
+class SD_4XUpscale_Conditioning(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "images": ("IMAGE",),
-                              "positive": ("CONDITIONING",),
-                              "negative": ("CONDITIONING",),
-                              "scale_ratio": ("FLOAT", {"default": 4.0, "min": 0.0, "max": 10.0, "step": 0.01}),
-                              "noise_augmentation": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.001}),
-                             }}
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
+    def define_schema(cls):
+        return io.Schema(
+            node_id="SD_4XUpscale_Conditioning",
+            category="conditioning/upscale_diffusion",
+            inputs=[
+                io.Image.Input("images"),
+                io.Conditioning.Input("positive"),
+                io.Conditioning.Input("negative"),
+                io.Float.Input("scale_ratio", default=4.0, min=0.0, max=10.0, step=0.01),
+                io.Float.Input("noise_augmentation", default=0.0, min=0.0, max=1.0, step=0.001),
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent"),
+            ],
+        )
 
-    FUNCTION = "encode"
-
-    CATEGORY = "conditioning/upscale_diffusion"
-
-    def encode(self, images, positive, negative, scale_ratio, noise_augmentation):
+    @classmethod
+    def execute(cls, images, positive, negative, scale_ratio, noise_augmentation):
         width = max(1, round(images.shape[-2] * scale_ratio))
         height = max(1, round(images.shape[-3] * scale_ratio))
 
@@ -39,8 +47,16 @@ class SD_4XUpscale_Conditioning:
             out_cn.append(n)
 
         latent = torch.zeros([images.shape[0], 4, height // 4, width // 4])
-        return (out_cp, out_cn, {"samples":latent})
+        return io.NodeOutput(out_cp, out_cn, {"samples":latent})
 
-NODE_CLASS_MAPPINGS = {
-    "SD_4XUpscale_Conditioning": SD_4XUpscale_Conditioning,
-}
+
+class SdUpscaleExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            SD_4XUpscale_Conditioning,
+        ]
+
+
+async def comfy_entrypoint() -> SdUpscaleExtension:
+    return SdUpscaleExtension()

comfy_extras/nodes_slg.py

@@ -1,33 +1,40 @@
 import comfy.model_patcher
 import comfy.samplers
 import re
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 
-class SkipLayerGuidanceDiT:
+class SkipLayerGuidanceDiT(io.ComfyNode):
     '''
     Enhance guidance towards detailed dtructure by having another set of CFG negative with skipped layers.
     Inspired by Perturbed Attention Guidance (https://arxiv.org/abs/2403.17377)
     Original experimental implementation for SD3 by Dango233@StabilityAI.
     '''
+
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"model": ("MODEL", ),
-                             "double_layers": ("STRING", {"default": "7, 8, 9", "multiline": False}),
-                             "single_layers": ("STRING", {"default": "7, 8, 9", "multiline": False}),
-                             "scale": ("FLOAT", {"default": 3.0, "min": 0.0, "max": 10.0, "step": 0.1}),
-                             "start_percent": ("FLOAT", {"default": 0.01, "min": 0.0, "max": 1.0, "step": 0.001}),
-                             "end_percent": ("FLOAT", {"default": 0.15, "min": 0.0, "max": 1.0, "step": 0.001}),
-                             "rescaling_scale": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 10.0, "step": 0.01}),
-                                }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "skip_guidance"
-    EXPERIMENTAL = True
+    def define_schema(cls):
+        return io.Schema(
+            node_id="SkipLayerGuidanceDiT",
+            category="advanced/guidance",
+            description="Generic version of SkipLayerGuidance node that can be used on every DiT model.",
+            is_experimental=True,
+            inputs=[
+                io.Model.Input("model"),
+                io.String.Input("double_layers", default="7, 8, 9"),
+                io.String.Input("single_layers", default="7, 8, 9"),
+                io.Float.Input("scale", default=3.0, min=0.0, max=10.0, step=0.1),
+                io.Float.Input("start_percent", default=0.01, min=0.0, max=1.0, step=0.001),
+                io.Float.Input("end_percent", default=0.15, min=0.0, max=1.0, step=0.001),
+                io.Float.Input("rescaling_scale", default=0.0, min=0.0, max=10.0, step=0.01),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    DESCRIPTION = "Generic version of SkipLayerGuidance node that can be used on every DiT model."
-
-    CATEGORY = "advanced/guidance"
-
-    def skip_guidance(self, model, scale, start_percent, end_percent, double_layers="", single_layers="", rescaling_scale=0):
+    @classmethod
+    def execute(cls, model, scale, start_percent, end_percent, double_layers="", single_layers="", rescaling_scale=0) -> io.NodeOutput:
         # check if layer is comma separated integers
         def skip(args, extra_args):
             return args
@@ -43,7 +50,7 @@ class SkipLayerGuidanceDiT:
         single_layers = [int(i) for i in single_layers]
 
         if len(double_layers) == 0 and len(single_layers) == 0:
-            return (model, )
+            return io.NodeOutput(model)
 
         def post_cfg_function(args):
             model = args["model"]
@@ -76,29 +83,36 @@ class SkipLayerGuidanceDiT:
         m = model.clone()
         m.set_model_sampler_post_cfg_function(post_cfg_function)
 
-        return (m, )
+        return io.NodeOutput(m)
 
-class SkipLayerGuidanceDiTSimple:
+    skip_guidance = execute  # TODO: remove
+
+
+class SkipLayerGuidanceDiTSimple(io.ComfyNode):
     '''
     Simple version of the SkipLayerGuidanceDiT node that only modifies the uncond pass.
     '''
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": {"model": ("MODEL", ),
-                             "double_layers": ("STRING", {"default": "7, 8, 9", "multiline": False}),
-                             "single_layers": ("STRING", {"default": "7, 8, 9", "multiline": False}),
-                             "start_percent": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1.0, "step": 0.001}),
-                             "end_percent": ("FLOAT", {"default": 1.0, "min": 0.0, "max": 1.0, "step": 0.001}),
-                                }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "skip_guidance"
-    EXPERIMENTAL = True
+    def define_schema(cls):
+        return io.Schema(
+            node_id="SkipLayerGuidanceDiTSimple",
+            category="advanced/guidance",
+            description="Simple version of the SkipLayerGuidanceDiT node that only modifies the uncond pass.",
+            is_experimental=True,
+            inputs=[
+                io.Model.Input("model"),
+                io.String.Input("double_layers", default="7, 8, 9"),
+                io.String.Input("single_layers", default="7, 8, 9"),
+                io.Float.Input("start_percent", default=0.0, min=0.0, max=1.0, step=0.001),
+                io.Float.Input("end_percent", default=1.0, min=0.0, max=1.0, step=0.001),
+            ],
+            outputs=[
+                io.Model.Output(),
+            ],
+        )
 
-    DESCRIPTION = "Simple version of the SkipLayerGuidanceDiT node that only modifies the uncond pass."
-
-    CATEGORY = "advanced/guidance"
-
-    def skip_guidance(self, model, start_percent, end_percent, double_layers="", single_layers=""):
+    @classmethod
+    def execute(cls, model, start_percent, end_percent, double_layers="", single_layers="") -> io.NodeOutput:
         def skip(args, extra_args):
             return args
 
@@ -113,7 +127,7 @@ class SkipLayerGuidanceDiTSimple:
         single_layers = [int(i) for i in single_layers]
 
         if len(double_layers) == 0 and len(single_layers) == 0:
-            return (model, )
+            return io.NodeOutput(model)
 
         def calc_cond_batch_function(args):
             x = args["input"]
@@ -144,9 +158,19 @@ class SkipLayerGuidanceDiTSimple:
         m = model.clone()
         m.set_model_sampler_calc_cond_batch_function(calc_cond_batch_function)
 
-        return (m, )
+        return io.NodeOutput(m)
 
-NODE_CLASS_MAPPINGS = {
-    "SkipLayerGuidanceDiT": SkipLayerGuidanceDiT,
-    "SkipLayerGuidanceDiTSimple": SkipLayerGuidanceDiTSimple,
-}
+    skip_guidance = execute  # TODO: remove
+
+
+class SkipLayerGuidanceExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            SkipLayerGuidanceDiT,
+            SkipLayerGuidanceDiTSimple,
+        ]
+
+
+async def comfy_entrypoint() -> SkipLayerGuidanceExtension:
+    return SkipLayerGuidanceExtension()

comfy_extras/nodes_stable3d.py

@@ -1,6 +1,8 @@
 import torch
 import nodes
 import comfy.utils
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 def camera_embeddings(elevation, azimuth):
     elevation = torch.as_tensor([elevation])
@@ -20,26 +22,31 @@ def camera_embeddings(elevation, azimuth):
     return embeddings
 
 
-class StableZero123_Conditioning:
+class StableZero123_Conditioning(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "clip_vision": ("CLIP_VISION",),
-                              "init_image": ("IMAGE",),
-                              "vae": ("VAE",),
-                              "width": ("INT", {"default": 256, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                              "height": ("INT", {"default": 256, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
-                              "elevation": ("FLOAT", {"default": 0.0, "min": -180.0, "max": 180.0, "step": 0.1, "round": False}),
-                              "azimuth": ("FLOAT", {"default": 0.0, "min": -180.0, "max": 180.0, "step": 0.1, "round": False}),
-                             }}
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
+    def define_schema(cls):
+        return io.Schema(
+            node_id="StableZero123_Conditioning",
+            category="conditioning/3d_models",
+            inputs=[
+                io.ClipVision.Input("clip_vision"),
+                io.Image.Input("init_image"),
+                io.Vae.Input("vae"),
+                io.Int.Input("width", default=256, min=16, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("height", default=256, min=16, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+                io.Float.Input("elevation", default=0.0, min=-180.0, max=180.0, step=0.1, round=False),
+                io.Float.Input("azimuth", default=0.0, min=-180.0, max=180.0, step=0.1, round=False)
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent")
+            ]
+        )
 
-    FUNCTION = "encode"
-
-    CATEGORY = "conditioning/3d_models"
-
-    def encode(self, clip_vision, init_image, vae, width, height, batch_size, elevation, azimuth):
+    @classmethod
+    def execute(cls, clip_vision, init_image, vae, width, height, batch_size, elevation, azimuth) -> io.NodeOutput:
         output = clip_vision.encode_image(init_image)
         pooled = output.image_embeds.unsqueeze(0)
         pixels = comfy.utils.common_upscale(init_image.movedim(-1,1), width, height, "bilinear", "center").movedim(1,-1)
@@ -51,30 +58,35 @@ class StableZero123_Conditioning:
         positive = [[cond, {"concat_latent_image": t}]]
         negative = [[torch.zeros_like(pooled), {"concat_latent_image": torch.zeros_like(t)}]]
         latent = torch.zeros([batch_size, 4, height // 8, width // 8])
-        return (positive, negative, {"samples":latent})
+        return io.NodeOutput(positive, negative, {"samples":latent})
 
-class StableZero123_Conditioning_Batched:
+class StableZero123_Conditioning_Batched(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "clip_vision": ("CLIP_VISION",),
-                              "init_image": ("IMAGE",),
-                              "vae": ("VAE",),
-                              "width": ("INT", {"default": 256, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                              "height": ("INT", {"default": 256, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                              "batch_size": ("INT", {"default": 1, "min": 1, "max": 4096}),
-                              "elevation": ("FLOAT", {"default": 0.0, "min": -180.0, "max": 180.0, "step": 0.1, "round": False}),
-                              "azimuth": ("FLOAT", {"default": 0.0, "min": -180.0, "max": 180.0, "step": 0.1, "round": False}),
-                              "elevation_batch_increment": ("FLOAT", {"default": 0.0, "min": -180.0, "max": 180.0, "step": 0.1, "round": False}),
-                              "azimuth_batch_increment": ("FLOAT", {"default": 0.0, "min": -180.0, "max": 180.0, "step": 0.1, "round": False}),
-                             }}
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
+    def define_schema(cls):
+        return io.Schema(
+            node_id="StableZero123_Conditioning_Batched",
+            category="conditioning/3d_models",
+            inputs=[
+                io.ClipVision.Input("clip_vision"),
+                io.Image.Input("init_image"),
+                io.Vae.Input("vae"),
+                io.Int.Input("width", default=256, min=16, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("height", default=256, min=16, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("batch_size", default=1, min=1, max=4096),
+                io.Float.Input("elevation", default=0.0, min=-180.0, max=180.0, step=0.1, round=False),
+                io.Float.Input("azimuth", default=0.0, min=-180.0, max=180.0, step=0.1, round=False),
+                io.Float.Input("elevation_batch_increment", default=0.0, min=-180.0, max=180.0, step=0.1, round=False),
+                io.Float.Input("azimuth_batch_increment", default=0.0, min=-180.0, max=180.0, step=0.1, round=False)
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent")
+            ]
+        )
 
-    FUNCTION = "encode"
-
-    CATEGORY = "conditioning/3d_models"
-
-    def encode(self, clip_vision, init_image, vae, width, height, batch_size, elevation, azimuth, elevation_batch_increment, azimuth_batch_increment):
+    @classmethod
+    def execute(cls, clip_vision, init_image, vae, width, height, batch_size, elevation, azimuth, elevation_batch_increment, azimuth_batch_increment) -> io.NodeOutput:
         output = clip_vision.encode_image(init_image)
         pooled = output.image_embeds.unsqueeze(0)
         pixels = comfy.utils.common_upscale(init_image.movedim(-1,1), width, height, "bilinear", "center").movedim(1,-1)
@@ -93,27 +105,32 @@ class StableZero123_Conditioning_Batched:
         positive = [[cond, {"concat_latent_image": t}]]
         negative = [[torch.zeros_like(pooled), {"concat_latent_image": torch.zeros_like(t)}]]
         latent = torch.zeros([batch_size, 4, height // 8, width // 8])
-        return (positive, negative, {"samples":latent, "batch_index": [0] * batch_size})
+        return io.NodeOutput(positive, negative, {"samples":latent, "batch_index": [0] * batch_size})
 
-class SV3D_Conditioning:
+class SV3D_Conditioning(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "clip_vision": ("CLIP_VISION",),
-                              "init_image": ("IMAGE",),
-                              "vae": ("VAE",),
-                              "width": ("INT", {"default": 576, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                              "height": ("INT", {"default": 576, "min": 16, "max": nodes.MAX_RESOLUTION, "step": 8}),
-                              "video_frames": ("INT", {"default": 21, "min": 1, "max": 4096}),
-                              "elevation": ("FLOAT", {"default": 0.0, "min": -90.0, "max": 90.0, "step": 0.1, "round": False}),
-                             }}
-    RETURN_TYPES = ("CONDITIONING", "CONDITIONING", "LATENT")
-    RETURN_NAMES = ("positive", "negative", "latent")
+    def define_schema(cls):
+        return io.Schema(
+            node_id="SV3D_Conditioning",
+            category="conditioning/3d_models",
+            inputs=[
+                io.ClipVision.Input("clip_vision"),
+                io.Image.Input("init_image"),
+                io.Vae.Input("vae"),
+                io.Int.Input("width", default=576, min=16, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("height", default=576, min=16, max=nodes.MAX_RESOLUTION, step=8),
+                io.Int.Input("video_frames", default=21, min=1, max=4096),
+                io.Float.Input("elevation", default=0.0, min=-90.0, max=90.0, step=0.1, round=False)
+            ],
+            outputs=[
+                io.Conditioning.Output(display_name="positive"),
+                io.Conditioning.Output(display_name="negative"),
+                io.Latent.Output(display_name="latent")
+            ]
+        )
 
-    FUNCTION = "encode"
-
-    CATEGORY = "conditioning/3d_models"
-
-    def encode(self, clip_vision, init_image, vae, width, height, video_frames, elevation):
+    @classmethod
+    def execute(cls, clip_vision, init_image, vae, width, height, video_frames, elevation) -> io.NodeOutput:
         output = clip_vision.encode_image(init_image)
         pooled = output.image_embeds.unsqueeze(0)
         pixels = comfy.utils.common_upscale(init_image.movedim(-1,1), width, height, "bilinear", "center").movedim(1,-1)
@@ -133,11 +150,17 @@ class SV3D_Conditioning:
         positive = [[pooled, {"concat_latent_image": t, "elevation": elevations, "azimuth": azimuths}]]
         negative = [[torch.zeros_like(pooled), {"concat_latent_image": torch.zeros_like(t), "elevation": elevations, "azimuth": azimuths}]]
         latent = torch.zeros([video_frames, 4, height // 8, width // 8])
-        return (positive, negative, {"samples":latent})
+        return io.NodeOutput(positive, negative, {"samples":latent})
 
 
-NODE_CLASS_MAPPINGS = {
-    "StableZero123_Conditioning": StableZero123_Conditioning,
-    "StableZero123_Conditioning_Batched": StableZero123_Conditioning_Batched,
-    "SV3D_Conditioning": SV3D_Conditioning,
-}
+class Stable3DExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            StableZero123_Conditioning,
+            StableZero123_Conditioning_Batched,
+            SV3D_Conditioning,
+        ]
+
+async def comfy_entrypoint() -> Stable3DExtension:
+    return Stable3DExtension()

comfy_extras/nodes_tcfg.py

@@ -1,8 +1,9 @@
 # TCFG: Tangential Damping Classifier-free Guidance - (arXiv: https://arxiv.org/abs/2503.18137)
 
+from typing_extensions import override
 import torch
 
-from comfy.comfy_types import IO, ComfyNodeABC, InputTypeDict
+from comfy_api.latest import ComfyExtension, io
 
 
 def score_tangential_damping(cond_score: torch.Tensor, uncond_score: torch.Tensor) -> torch.Tensor:
@@ -26,23 +27,24 @@ def score_tangential_damping(cond_score: torch.Tensor, uncond_score: torch.Tenso
     return uncond_score_td.reshape_as(uncond_score).to(uncond_score.dtype)
 
 
-class TCFG(ComfyNodeABC):
+class TCFG(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(cls) -> InputTypeDict:
-        return {
-            "required": {
-                "model": (IO.MODEL, {}),
-            }
-        }
+    def define_schema(cls):
+        return io.Schema(
+            node_id="TCFG",
+            display_name="Tangential Damping CFG",
+            category="advanced/guidance",
+            description="TCFG – Tangential Damping CFG (2503.18137)\n\nRefine the uncond (negative) to align with the cond (positive) for improving quality.",
+            inputs=[
+                io.Model.Input("model"),
+            ],
+            outputs=[
+                io.Model.Output(display_name="patched_model"),
+            ],
+        )
 
-    RETURN_TYPES = (IO.MODEL,)
-    RETURN_NAMES = ("patched_model",)
-    FUNCTION = "patch"
-
-    CATEGORY = "advanced/guidance"
-    DESCRIPTION = "TCFG – Tangential Damping CFG (2503.18137)\n\nRefine the uncond (negative) to align with the cond (positive) for improving quality."
-
-    def patch(self, model):
+    @classmethod
+    def execute(cls, model):
         m = model.clone()
 
         def tangential_damping_cfg(args):
@@ -59,13 +61,16 @@ class TCFG(ComfyNodeABC):
             return [cond_pred, uncond_pred_td] + conds_out[2:]
 
         m.set_model_sampler_pre_cfg_function(tangential_damping_cfg)
-        return (m,)
+        return io.NodeOutput(m)
 
 
-NODE_CLASS_MAPPINGS = {
-    "TCFG": TCFG,
-}
+class TcfgExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            TCFG,
+        ]
 
-NODE_DISPLAY_NAME_MAPPINGS = {
-    "TCFG": "Tangential Damping CFG",
-}
+
+async def comfy_entrypoint() -> TcfgExtension:
+    return TcfgExtension()

comfy_extras/nodes_tomesd.py

@@ -1,7 +1,9 @@
 #Taken from: https://github.com/dbolya/tomesd
 
 import torch
-from typing import Tuple, Callable
+from typing import Tuple, Callable, Optional
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 import math
 
 def do_nothing(x: torch.Tensor, mode:str=None):
@@ -144,33 +146,45 @@ def get_functions(x, ratio, original_shape):
 
 
 
-class TomePatchModel:
+class TomePatchModel(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                              "ratio": ("FLOAT", {"default": 0.3, "min": 0.0, "max": 1.0, "step": 0.01}),
-                              }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="TomePatchModel",
+            category="model_patches/unet",
+            inputs=[
+                io.Model.Input("model"),
+                io.Float.Input("ratio", default=0.3, min=0.0, max=1.0, step=0.01),
+            ],
+            outputs=[io.Model.Output()],
+        )
 
-    CATEGORY = "model_patches/unet"
-
-    def patch(self, model, ratio):
-        self.u = None
+    @classmethod
+    def execute(cls, model, ratio) -> io.NodeOutput:
+        u: Optional[Callable] = None
         def tomesd_m(q, k, v, extra_options):
+            nonlocal u
             #NOTE: In the reference code get_functions takes x (input of the transformer block) as the argument instead of q
             #however from my basic testing it seems that using q instead gives better results
-            m, self.u = get_functions(q, ratio, extra_options["original_shape"])
+            m, u = get_functions(q, ratio, extra_options["original_shape"])
             return m(q), k, v
         def tomesd_u(n, extra_options):
-            return self.u(n)
+            nonlocal u
+            return u(n)
 
         m = model.clone()
         m.set_model_attn1_patch(tomesd_m)
         m.set_model_attn1_output_patch(tomesd_u)
-        return (m, )
+        return io.NodeOutput(m)
 
 
-NODE_CLASS_MAPPINGS = {
-    "TomePatchModel": TomePatchModel,
-}
+class TomePatchModelExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            TomePatchModel,
+        ]
+
+
+async def comfy_entrypoint() -> TomePatchModelExtension:
+    return TomePatchModelExtension()

comfy_extras/nodes_torch_compile.py

@@ -1,23 +1,39 @@
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 from comfy_api.torch_helpers import set_torch_compile_wrapper
 
 
-class TorchCompileModel:
+class TorchCompileModel(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model": ("MODEL",),
-                             "backend": (["inductor", "cudagraphs"],),
-                              }}
-    RETURN_TYPES = ("MODEL",)
-    FUNCTION = "patch"
+    def define_schema(cls) -> io.Schema:
+        return io.Schema(
+            node_id="TorchCompileModel",
+            category="_for_testing",
+            inputs=[
+                io.Model.Input("model"),
+                io.Combo.Input(
+                    "backend",
+                    options=["inductor", "cudagraphs"],
+                ),
+            ],
+            outputs=[io.Model.Output()],
+            is_experimental=True,
+        )
 
-    CATEGORY = "_for_testing"
-    EXPERIMENTAL = True
-
-    def patch(self, model, backend):
+    @classmethod
+    def execute(cls, model, backend) -> io.NodeOutput:
         m = model.clone()
         set_torch_compile_wrapper(model=m, backend=backend)
-        return (m, )
+        return io.NodeOutput(m)
 
-NODE_CLASS_MAPPINGS = {
-    "TorchCompileModel": TorchCompileModel,
-}
+
+class TorchCompileExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            TorchCompileModel,
+        ]
+
+
+async def comfy_entrypoint() -> TorchCompileExtension:
+    return TorchCompileExtension()

comfy_extras/nodes_upscale_model.py

@@ -4,6 +4,8 @@ from comfy import model_management
 import torch
 import comfy.utils
 import folder_paths
+from typing_extensions import override
+from comfy_api.latest import ComfyExtension, io
 
 try:
     from spandrel_extra_arches import EXTRA_REGISTRY
@@ -13,17 +15,23 @@ try:
 except:
     pass
 
-class UpscaleModelLoader:
+class UpscaleModelLoader(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "model_name": (folder_paths.get_filename_list("upscale_models"), ),
-                             }}
-    RETURN_TYPES = ("UPSCALE_MODEL",)
-    FUNCTION = "load_model"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="UpscaleModelLoader",
+            display_name="Load Upscale Model",
+            category="loaders",
+            inputs=[
+                io.Combo.Input("model_name", options=folder_paths.get_filename_list("upscale_models")),
+            ],
+            outputs=[
+                io.UpscaleModel.Output(),
+            ],
+        )
 
-    CATEGORY = "loaders"
-
-    def load_model(self, model_name):
+    @classmethod
+    def execute(cls, model_name) -> io.NodeOutput:
         model_path = folder_paths.get_full_path_or_raise("upscale_models", model_name)
         sd = comfy.utils.load_torch_file(model_path, safe_load=True)
         if "module.layers.0.residual_group.blocks.0.norm1.weight" in sd:
@@ -33,21 +41,29 @@ class UpscaleModelLoader:
         if not isinstance(out, ImageModelDescriptor):
             raise Exception("Upscale model must be a single-image model.")
 
-        return (out, )
+        return io.NodeOutput(out)
+
+    load_model = execute  # TODO: remove
 
 
-class ImageUpscaleWithModel:
+class ImageUpscaleWithModel(io.ComfyNode):
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "upscale_model": ("UPSCALE_MODEL",),
-                              "image": ("IMAGE",),
-                              }}
-    RETURN_TYPES = ("IMAGE",)
-    FUNCTION = "upscale"
+    def define_schema(cls):
+        return io.Schema(
+            node_id="ImageUpscaleWithModel",
+            display_name="Upscale Image (using Model)",
+            category="image/upscaling",
+            inputs=[
+                io.UpscaleModel.Input("upscale_model"),
+                io.Image.Input("image"),
+            ],
+            outputs=[
+                io.Image.Output(),
+            ],
+        )
 
-    CATEGORY = "image/upscaling"
-
-    def upscale(self, upscale_model, image):
+    @classmethod
+    def execute(cls, upscale_model, image) -> io.NodeOutput:
         device = model_management.get_torch_device()
 
         memory_required = model_management.module_size(upscale_model.model)
@@ -75,9 +91,19 @@ class ImageUpscaleWithModel:
 
         upscale_model.to("cpu")
         s = torch.clamp(s.movedim(-3,-1), min=0, max=1.0)
-        return (s,)
+        return io.NodeOutput(s)
 
-NODE_CLASS_MAPPINGS = {
-    "UpscaleModelLoader": UpscaleModelLoader,
-    "ImageUpscaleWithModel": ImageUpscaleWithModel
-}
+    upscale = execute  # TODO: remove
+
+
+class UpscaleModelExtension(ComfyExtension):
+    @override
+    async def get_node_list(self) -> list[type[io.ComfyNode]]:
+        return [
+            UpscaleModelLoader,
+            ImageUpscaleWithModel,
+        ]
+
+
+async def comfy_entrypoint() -> UpscaleModelExtension:
+    return UpscaleModelExtension()