Replies: 2 comments
-
I did the update last night but it was still the same, when I test other samplers I get some kind of output blurry jumpy colors, but the one from default WF tiny file saved black frames. So far none work well. I have 12gb VRAM and I can use full FLUX dev no issue, LTX works so the basic T2V for this should be easy enough? I will try a gguf see if that makes any odds of course seems nobody did a gguf of the smallest as I guess there was not much reason to. |
Beta Was this translation helpful? Give feedback.
0 replies
-
tied gguf same issue |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
I updated Pytorch 2.5.1+cu121 as it said 2.4+ and tried again but same issue. Tried setting weight type to all the FP8 types (using model wan2.1_t2v_1.3B_fp16.safetensors and have the right vae) comfyui at v0.3.24-13-g29832b3b about to update again see if the FP8 thing helps.
Any tips, tricks, dependencies to check, issues with specific cards (on Nvidia)?
Also I tried with option --force-upcast-attention made no change.
Beta Was this translation helpful? Give feedback.
All reactions