continue-revolution / sd-webui-animatediff

AnimateDiff for AUTOMATIC1111 Stable Diffusion WebUI
Other
3.11k stars 258 forks source link

[Bug]: #361

Closed HellKaiser45 closed 11 months ago

HellKaiser45 commented 11 months ago

Is there an existing issue for this?

Have you read FAQ on README?

What happened?

Hello,

I'm having trouble getting a simple animation right. The results are not satisfactory. My basic workflow involves vid2vid with controlnet openpose on comfyui. First, I extract the poses from my video. Then, I use the Ksampler (I've tried several combinations of samplers and schedulers).

myworkflow

I've also tried various checkpoints (even the recommended ones), but the results are quite odd. It's as if the image is overly "baked". For example:

image

The quality of this image is far inferior to what the model is capable of at this resolution, as you can see here:

image

I suspect the issue might be due to an incompatible Python library, possibly PyTorch. I can't pinpoint the exact problem.

Can anyone help me resolve this issue?

Steps to reproduce the problem

  1. Go to ....
  2. Press ....
  3. ...

What should have happened?

a good quality animation

Commit where the problem happens

webui: comfyui extension: impactpack

What browsers do you use to access the UI ?

No response

Command Line Arguments

default

Console logs

nothing interesting

Additional information

No response

continue-revolution commented 11 months ago

This repo is the extension for A1111 users, not for Comfy. Please post your issue to the Comfy node you use. That said, this extension can achieve most features you want, especially CN V2V. If you have any problem using my extension in A1111, you are welcome to post another issue to ask.

HellKaiser45 commented 11 months ago

sorry for that I'll definitely try it with A1111.