AI Knowledge Central patreon

🚨 Emergency Wan 2.2 Workflow Fix! 🚨

🕑 Added 2025-10-29 08:34:17 +0000 UTC

Comments

Chris Wenzl

throwing out the clean VRAM node can help in that case

Iain Forbes

I am stil getting noise after the first generation. If I generate an image after loading comfyui it's fine. If I run it again I get noise unless I fully restart the program

awol

Also I'm starting to see frequent errors. it doesn't happen all the time but once in a while: KSamplerAdvanced CUDA error: unspecified launch failure

awol

Did something change with the random seed? I notice the same prompt seems to generate the same output every time and doesn't seem to be using a random seed.

Chris Wenzl

Hi, can you check this again please? I used the workflow today and it ran just fine.

Chris Wenzl

Hi, can you check your console when comfyUI is starting up maybe its some incompatibility. Without any error best i can do is guess work. You could go to the manager again and look for missing nodes. If any nodes cant be loaded use the "try-fix" button.

KillerteamX

Hello, i just try to make it work but i got some problems that some stuff dont get loaded. im not talented with this stuff but i got a screenshot of it would be Cool if you could help me with that and possible say what i did wrong. https://prnt.sc/fCRLVBsCqdNF . i just did it how you did in the Video.

Matt A

Not sure if it's affecting anyone else but latest comfyui broke this workflow again. They know about it and are fixing it even with the nightly update I can't get it to work. Default wan 2.2 template works fine still. Something in your workflow for hit. I think it's the ksampler now. Getting cuda errors

Matt A

There is a new ksampler node that apparently works well and gives more control over what parts of the generation high/low use a lightx2v lora and what doesn't. Would love to see this in a future workflow to further refine quality vs speed: https://github.com/VraethrDalkr/ComfyUI-TripleKSampler?tab=readme-ov-file