SakeTami
Innovate Futures @ Benji
Innovate Futures @ Benji

patreon


(Workflow) AnimateDiff_IPA_Mask_DifferentialDiffusionLatentNoise Version 20240419

Related Tutorial : https://www.patreon.com/posts/102610521

Related topics of this workflow:

a-person-mask-generator: https://github.com/djbielejeski/a-person-mask-generator

Differential Diffusion: https://differential-diffusion.github.io/

Please remember to check out the previous post for Essential Models and Custom Nodes You Need : 

https://www.patreon.com/posts/rave-animatediff-98757420

https://www.patreon.com/posts/rave-animatediff-101438763

Comments

For that error, suspect your test_image has an additional alpha channel per pixel, thus it has 4 channels instead of only three. Have you try new tourch version? IPAdapter clip vision use the latest? But looks like IPAdapter node got some update after this April 2024 release workflow. Maybe try to reconnect IPA nodes if you use updated IPAdapter custom node most recent version.

Benjamin Law

I built it by following YouTube but got an error. I tried the workflow and got the same error: IPAdapterTiled The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 1" What should I do? Thanks!

Luc Job

If you can't install that, you can use other integer node for setting number value there. Comfy is flexible not a textbook answer.

Benjamin Law

error: Derfuu ComfyUI ModdedNodes error Nodes that have failed to load will show as red on the graph

Creed

derfuu node not working! got prompt Error handling request Traceback (most recent call last): File "E:\IA TOTAL\ComfyUI_windows_portable\python_embeded\Lib\site-packages\aiohttp\web_protocol.py", line 452, in _handle_request resp = await request_handler(request)

Creed

Have you enable Save video in the last Video Combine node? And you can right click and Save as well. By the way, that workflow is a good old one in Dec, 2023 :). Its still work with IPA for character and Bg, same theory but different setting.

Benjamin Law

Thanks for the quick reply. I took another shot with your workflow on Open Art: https://openart.ai/workflows/futurebenji/animatediff-flicker-free-animation-video-workflow/A9ZE35kkDazgWGXhnyXh After i hit queue prompt, everything works fine, but i'm stuck with the last node "Video Save - After Faceswap". There is no video nor anything to download. Am i missing something?

Schlecky Silberstein

If both not work , I think your system are missing something. Check out their Github and see the installation or the requirements.txt

Benjamin Law

The JWinteger and Derfuu Packs do not work. Are there any other nodes i can try to get theses values?

Schlecky Silberstein

If you have the Derfuu Nodes install, then it should be in there. Have you successfully loaded when Comfyui boot up? If not , then just replace any Integer set node for that 3 values.

Benjamin Law

Hey Benji, i ran in the same integer issue. I git cloned Derfuu Nodes in the custom node folder but i still get the integer not found error. On reddit someone advised me to replace it with any primitive but i don't want to mess up the workflow. It's about number of frames, height and width. I searched the internet twice for a solution, please help.

Schlecky Silberstein

cool thanks. I ended up using JWInteger package. but then I rebuilt the whole thing from scratch and the Load Video has a "frame load cap" input field and image resize includes fields for height and width so those nodes arent even necessary anymore.

Tom Sepe

hi , the Integer node are coming from a custom node package called " Derfuu_ComfyUI_ModdedNodes" https://github.com/Derfuu/Derfuu_ComfyUI_ModdedNodes

Benjamin Law

I can't seem to find the custom node "Integer" When loading the graph, the following node types were not found: Integer

Tom Sepe


More Creators