mirror of
https://github.com/muerrilla/sd-webui-detail-daemon.git
synced 2026-11-05 01:37:56 +00:00
[GH-ISSUE #12] index 49 is out of bounds for the schedule array, #8
Labels
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
muerrilla/sd-webui-detail-daemon#8
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @Krek12 on GitHub (Sep 27, 2024).
Original GitHub issue: https://github.com/muerrilla/sd-webui-detail-daemon/issues/12
*** Error executing callback cfg_denoiser_callback for /home/studio-lab-user/asd/extensions/sd-webui-detail-daemon/scripts/detail_daemon.py
Traceback (most recent call last):
File "/home/studio-lab-user/asd/modules/script_callbacks.py", line 333, in cfg_denoiser_callback
c.callback(params)
File "/home/studio-lab-user/asd/extensions/sd-webui-detail-daemon/scripts/detail_daemon.py", line 157, in denoiser_callback
multiplier = self.schedule[idx] * .1
IndexError: index 49 is out of bounds for axis 0 with size 25
@muerrilla commented on GitHub (Oct 10, 2024):
Need more context.
@Krek12 commented on GitHub (Oct 10, 2024):
I basically don't understand it myself, the extension seems to work but every step shows this is what you saw above.
@blepping commented on GitHub (Oct 24, 2024):
the logic for detecting 2 step samplers isn't working in this case. the extension makes a schedule with
steps * 2 - 1for those, not so coincidentally25 * 2 - 1is 49. you're probably using a sampler with a name it doesn't recognize as requiring that logic.@muerrilla commented on GitHub (Oct 24, 2024):
Half correct: Thesteps * 2 - 1thing is done for a few samplers specified by name. So if the sampler name is not recognized it would not be triggered. So this can't be a case of sampler name not being recognized; hence the need for more context.Oops, my bad. You're absolutely correct, I wasn't paying enough attention to which line is throwing the error. So, this might be caused by the 1.10.0 update. Haven't had time to update yet.
@Krek12 what sampler and webui version are you using?
@blepping commented on GitHub (Oct 26, 2024):
@muerrilla not sure if it would be useful to you, but this is how i tried to deal with the problem:
github.com/Jonseed/ComfyUI-Detail-Daemon@6d89e33265/detail_daemon_node.py (L199)in the case where we want to call the model with a sigma in between two of the schedule sigmas, we interpolate between the two DD adjustments proportionately. for example, if the sigma is 10% of the way from the low sigma to the high sigma then you'd get a mix of 90% of the DD schedule at the low sigma and 10% from the high sigma. that's what it's supposed to do anyway, not sure it actually works correctly.
with that approach, you don't have to care about how many steps a sampler uses.
@muerrilla commented on GitHub (Oct 26, 2024):
Interesting approach, but this is both impossible to pull off feasibly in a1111 (as far as I understand) and also way more complex than my one-line workaround - which only requires regular manual maintenance 😅 to keep the list of 2-step samplers up-to-date.
Cheers.
@Krek12 commented on GitHub (Oct 27, 2024):
AUTOMATIC, sampler DYN
https://github.com/Koishi-Star/Euler-Smea-Dyn-Sampler
@Krek12 commented on GitHub (Mar 15, 2025):
@muerrilla interesing https://github.com/Jonseed/ComfyUI-Detail-Daemon.git
@muerrilla commented on GitHub (Mar 15, 2025):
@Krek12 Hey! Yup, but what exactly about it (besides what was discussed above)?
@muerrilla commented on GitHub (Mar 15, 2025):
@Krek12 Sorry for the super late response, I had missed this reply. I don't think I can add support for these custom samplers without rewriting the whole thing. So probably won't happen.
@Krek12 commented on GitHub (Mar 15, 2025):
Is it possible to learn a little instruction? How would it work? Through the code.
Well, more precisely, what is the reason for the error? maybe I'll try to fix it myself later.
@Krek12 commented on GitHub (Mar 15, 2025):
An idea, by the way. Why don't you add a button to the extension that enables the operation of the detail daemon in hires-fix or turns it off.
Example to this:
https://github.com/Bing-su/adetailer/compare/main...MisterChief95:adetailer:main
@muerrilla commented on GitHub (Mar 15, 2025):
DD creates a schedule with a value for each step. The number of steps you set in the generation parameters are not necessarily the actual number of times the sampling function is run during generation. This depends on the sampler. For example second order samplers (e.g. Heun) do 2x the number of steps you set. DD handles this by checking the sampler name and when it's a second order sampler, it makes the schedule twice as long. Smea Dyn samplers do some extra steps, so they are neither x steps nor 2x steps but x + c steps or something. DD assumes they are normal samplers (since the sampler name is not one of the second order samplers) so it creates a schedule with the wrong number of steps, hence the error. Hope this helps.
@muerrilla commented on GitHub (Mar 15, 2025):
That one I've already done (plus some other features) but I need to do some code cleanup before I can release it. So maybe in like a month or so, when I have the free time to do that.