Files
ultimate-upscale-for-automa…/scripts/ultimate-upscale.py
2023-01-04 14:14:44 +03:00

294 lines
12 KiB
Python

import math
import modules.scripts as scripts
import gradio as gr
from PIL import Image, ImageDraw
from modules import processing, shared, sd_samplers, images, devices
from modules.processing import Processed
from modules.shared import opts, cmd_opts, state
def getFactor(num):
if num == 1:
return 2
if num % 4 == 0:
return 4
if num % 3 == 0:
return 3
if num % 2 == 0:
return 2
return 0
def upscale(p, init_img, upscaler_index, tileSize, padding):
scale_factor = max(p.width, p.height) // max(init_img.width, init_img.height)
print(f"Canva size: {p.width}x{p.height}")
print(f"Image size: {init_img.width}x{init_img.height}")
print(f"Scale factor: {scale_factor}")
upscaler = shared.sd_upscalers[upscaler_index]
p.extra_generation_params["SD upscale overlap"] = padding
p.extra_generation_params["SD upscale upscaler"] = upscaler.name
initial_info = None
seed = p.seed
upscaled_img = init_img
if upscaler.name == "None":
return upscaled_img.resize((p.width, p.height), resample=Image.LANCZOS)
current_scale = 1
iteration = 0
current_scale_factor = getFactor(scale_factor)
while current_scale_factor == 0:
scale_factor += 1
current_scale_factor = getFactor(scale_factor)
while current_scale < scale_factor:
iteration += 1
current_scale_factor = getFactor(scale_factor // current_scale)
current_scale = current_scale * current_scale_factor
if current_scale_factor == 0:
break
print(f"Upscaling iteration {iteration} with scale factor {current_scale_factor}")
upscaled_img = upscaler.scaler.upscale(upscaled_img, current_scale_factor, upscaler.data_path)
return upscaled_img.resize((p.width, p.height), resample=Image.LANCZOS)
def redraw_image(p, upscaled_img, rows, cols, tileSize, padding):
initial_info = None
for yi in range(rows):
for xi in range(cols):
p.width = tileSize
p.height = tileSize
p.inpaint_full_res = True
p.inpaint_full_res_padding = padding
mask = Image.new("L", (upscaled_img.width, upscaled_img.height), "black")
draw = ImageDraw.Draw(mask)
draw.rectangle((
xi * tileSize,
yi * tileSize,
xi * tileSize + tileSize,
yi * tileSize + tileSize
), fill="white")
p.init_images = [upscaled_img]
p.image_mask = mask
processed = processing.process_images(p)
initial_info = processed.info
if (len(processed.images) > 0):
upscaled_img = processed.images[0]
return upscaled_img, initial_info
def redraw_middle_offset_image(p, upscaled_img, rows, cols, tileSize, padding, seams_fix_denoise, seams_fix_mask_blur):
initial_info = None
gradient = Image.linear_gradient("L")
row_gradient = Image.new("L", (tileSize, tileSize), "black")
row_gradient.paste(gradient.resize((tileSize, tileSize//2), resample=Image.BICUBIC), (0, 0))
row_gradient.paste(gradient.rotate(180).resize((tileSize, tileSize//2), resample=Image.BICUBIC), (0, tileSize//2))
col_gradient = Image.new("L", (tileSize, tileSize), "black")
col_gradient.paste(gradient.rotate(90).resize((tileSize//2, tileSize), resample=Image.BICUBIC), (0, 0))
col_gradient.paste(gradient.rotate(270).resize((tileSize//2, tileSize), resample=Image.BICUBIC), (tileSize//2, 0))
p.denoising_strength = seams_fix_denoise
p.mask_blur = seams_fix_mask_blur
for yi in range(rows-1):
for xi in range(cols):
p.width = tileSize
p.height = tileSize
p.inpaint_full_res = True
p.inpaint_full_res_padding = padding
mask = Image.new("L", (upscaled_img.width, upscaled_img.height), "black")
mask.paste(row_gradient, (xi*tileSize, yi*tileSize + tileSize//2))
p.init_images = [upscaled_img]
p.image_mask = mask
processed = processing.process_images(p)
initial_info = processed.info
if (len(processed.images) > 0):
upscaled_img = processed.images[0]
for yi in range(rows):
for xi in range(cols-1):
p.width = tileSize
p.height = tileSize
p.inpaint_full_res = True
p.inpaint_full_res_padding = padding
mask = Image.new("L", (upscaled_img.width, upscaled_img.height), "black")
mask.paste(col_gradient, (xi*tileSize+tileSize//2, yi*tileSize))
p.init_images = [upscaled_img]
p.image_mask = mask
processed = processing.process_images(p)
initial_info = processed.info
if (len(processed.images) > 0):
upscaled_img = processed.images[0]
return upscaled_img, initial_info
def seam_draw(p, upscaled_img, seams_fix_width, seams_fix_padding, seams_fix_denoise, padding, tileSize, cols, rows, mask_blur):
p.denoising_strength = seams_fix_denoise
p.mask_blur = 0
gradient = Image.linear_gradient("L")
mirror_gradient = Image.new("L", (256, 256), "black")
mirror_gradient.paste(gradient.resize((256, 128), resample=Image.BICUBIC), (0, 0))
mirror_gradient.paste(gradient.rotate(180).resize((256, 128), resample=Image.BICUBIC), (0, 128))
row_gradient = mirror_gradient.resize((upscaled_img.width, seams_fix_width), resample=Image.BICUBIC)
col_gradient = mirror_gradient.rotate(90).resize((seams_fix_width, upscaled_img.height), resample=Image.BICUBIC)
for xi in range(1, cols):
p.width = seams_fix_width + seams_fix_padding * 2
p.height = upscaled_img.height
p.inpaint_full_res = True
p.inpaint_full_res_padding = seams_fix_padding
mask = Image.new("L", (upscaled_img.width, upscaled_img.height), "black")
mask.paste(col_gradient, (xi * tileSize - seams_fix_width // 2, 0))
p.init_images = [upscaled_img]
p.image_mask = mask
processed = processing.process_images(p)
if (len(processed.images) > 0):
upscaled_img = processed.images[0]
for yi in range(1, rows):
p.width = upscaled_img.width
p.height = seams_fix_width + seams_fix_padding * 2
p.inpaint_full_res = True
p.inpaint_full_res_padding = seams_fix_padding
mask = Image.new("L", (upscaled_img.width, upscaled_img.height), "black")
mask.paste(row_gradient, (0, yi * tileSize - seams_fix_width // 2))
p.init_images = [upscaled_img]
p.image_mask = mask
processed = processing.process_images(p)
if (len(processed.images) > 0):
upscaled_img = processed.images[0]
return upscaled_img
class Script(scripts.Script):
def title(self):
return "Ultimate SD upscale"
def show(self, is_img2img):
return is_img2img
def ui(self, is_img2img):
seams_fix_types = [
"None",
"Band pass",
"Half tile offset pass"
]
info = gr.HTML(
"<p style=\"margin-bottom:0.75em\">Will upscale the image to selected with and height</p>")
gr.HTML("<p style=\"margin-bottom:0.75em\">Redraw options:</p>")
with gr.Row():
redraw_enabled = gr.Checkbox(label="Enabled", value=True)
tileSize = gr.Slider(minimum=256, maximum=2048, step=64, label='Tile size', value=512)
mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=8)
padding = gr.Slider(label='Padding', minimum=0, maximum=128, step=1, value=32)
upscaler_index = gr.Radio(label='Upscaler', choices=[x.name for x in shared.sd_upscalers],
value=shared.sd_upscalers[0].name, type="index")
gr.HTML("<p style=\"margin-bottom:0.75em\">Seams fix:</p>")
with gr.Row():
seams_fix_type = gr.Dropdown(label="Type", choices=[k for k in seams_fix_types], type="index", value=next(iter(seams_fix_types)))
seams_fix_denoise = gr.Slider(label='Denoise', minimum=0, maximum=1, step=0.01, value=0.35, visible=False, interactive=True)
seams_fix_width = gr.Slider(label='Width', minimum=0, maximum=128, step=1, value=64, visible=False, interactive=True)
seams_fix_mask_blur = gr.Slider(label='Mask blur', minimum=0, maximum=64, step=1, value=4, visible=False, interactive=True)
seams_fix_padding = gr.Slider(label='Padding', minimum=0, maximum=128, step=1, value=16, visible=False, interactive=True)
gr.HTML("<p style=\"margin-bottom:0.75em\">Save options:</p>")
with gr.Row():
save_upscaled_image = gr.Checkbox(label="Upscaled", value=True)
save_seams_fix_image = gr.Checkbox(label="Seams fix", value=False)
def select_fix_type(fix_index):
all_visible = fix_index != 0
mask_blur_visible = fix_index == 2
width_visible = fix_index == 1
return [gr.update(visible=all_visible),
gr.update(visible=width_visible),
gr.update(visible=mask_blur_visible),
gr.update(visible=all_visible)]
seams_fix_type.change(
fn=select_fix_type,
inputs=seams_fix_type,
outputs=[seams_fix_denoise, seams_fix_width, seams_fix_mask_blur, seams_fix_padding]
)
return [info, tileSize, mask_blur, padding, seams_fix_width, seams_fix_denoise, seams_fix_padding,
upscaler_index, save_upscaled_image, redraw_enabled, save_seams_fix_image, seams_fix_mask_blur,
seams_fix_type]
def run(self, p, _, tileSize, mask_blur, padding, seams_fix_width, seams_fix_denoise, seams_fix_padding,
upscaler_index, save_upscaled_image, redraw_enabled, save_seams_fix_image, seams_fix_mask_blur,
seams_fix_type):
processing.fix_seed(p)
p.extra_generation_params["SD upscale tileSize"] = tileSize
p.mask_blur = mask_blur
seed = p.seed
initial_info = None
init_img = p.init_images[0]
init_img = images.flatten(init_img, opts.img2img_background_color)
# Upscaling
upscaled_img = upscale(p, init_img, upscaler_index, tileSize, padding)
# Drawing
devices.torch_gc()
p.do_not_save_grid = True
p.do_not_save_samples = True
p.inpaint_full_res = False
rows = math.ceil(p.height / tileSize)
cols = math.ceil(p.width / tileSize)
print(f"Tiles amount: {rows * cols}")
print(f"Grid: {rows}x{cols}")
print(f"Seam path: {seams_fix_type}")
seams = 0
if seams_fix_type > 0:
seams = rows - 1 + cols - 1
state.job_count = ((rows * cols) if redraw_enabled else 0) + (seams if seams_fix_type == 1 else 0) + (
(rows * (cols - 1) + (rows - 1) * cols) if seams_fix_type == 2 else 0)
result_images = []
result_image = upscaled_img
if redraw_enabled:
result_image, initial_info = redraw_image(p, upscaled_img, rows, cols, tileSize, padding)
result_images.append(result_image)
if save_upscaled_image:
images.save_image(result_image, p.outpath_samples, "", seed, p.prompt, opts.grid_format, info=initial_info, p=p)
if seams_fix_type == 2:
print(f"Starting offset pass drawing")
result_image, initial_info = redraw_middle_offset_image(p, result_image, rows, cols, tileSize, seams_fix_padding, seams_fix_denoise, seams_fix_mask_blur)
result_images.append(result_image)
if save_seams_fix_image:
images.save_image(result_image, p.outpath_samples, "", seed, p.prompt, opts.grid_format, info=initial_info, p=p)
if seams_fix_type == 1:
print(f"Starting bands pass drawing")
result_image = seam_draw(p, result_image, seams_fix_width, seams_fix_padding, seams_fix_denoise, padding, tileSize, cols, rows, 0)
result_images.append(result_image)
if save_seams_fix_image:
images.save_image(result_image, p.outpath_samples, "", seed, p.prompt, opts.grid_format, info=initial_info, p=p)
processed = Processed(p, result_images, seed, initial_info if initial_info is not None else "")
return processed