https://huggingface.co/datasets/StefanFalkok/ComfyUI_portable_torch_2.11.0_cu130_cp313_sageattention_triton - my new ComfyUI build with torch 2.11.0cu130 + sageattention + my workflows
Hi! I intrduce my WAN 2.2 Models 8 steps
NEW! Wan_2.2_I2V_SVI_2_PRO_8steps models
merged with SVI 2 PRO loras - you don't need choose SVI lora to make long video
https://huggingface.co/StefanFalkok/Wan_2.2_I2V_SVI_2_PRO_8steps - merged models + special workflow for those models + hunyuanvideofoley to add sound towards videos
Download
FP8/FP16 + Workflows - https://huggingface.co/StefanFalkok/Wan_2.2_10steps/tree/main
GGUF + Workflows - https://huggingface.co/StefanFalkok/Wan_2.2_10steps_GGUF/tree/main (Q8_0 and Q4_K_M)
Everything u need - just load model without light lora, set 8 steps (4/4), set 2cfg on High Noise and cfg1 on low noise, set 81 frame (16 framerate), set resolution from 576p to 720p
I merged original WAN 2.2 Models from ComfyUI repository with LightX T2V Rank 256 bf 16 (https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Lightx2v/lightx2v_T2V_14B_cfg_step_distill_v2_lora_rank256_bf16.safetensors)
fp8 results on rtx 5080
1024x576 - around 3 minutes
720p - around 5 minutes
Q8_0 and FP16 take a 20% more time to generate video, but you get more quality and stable result
My TG Channel - https://t.me/StefanFalkokAI
My TG Chat - https://t.me/+y4R5JybDZcFjMjFi
Description
Wan 2.2 I2V HighNoise 10 Steps Q4_K_M
FAQ
Comments (7)
Just so you know, your T2V fp8 examples are great, and your I2V, while they look nice, are all in slow motion, which is something people are avoiding with this lightning stuff. Is that because you only got slow motion videos out of the I2V model, or you happened to only choose slow motion videos?
Isn't the slow motion fixed by now? I don't know, haven't kept up much lately, but I know that at least for the i2v models, there are more recent versions of the light loras. It's a bit chaotic but on kijai's huggingface in the loras folder there are newer ones, like v1030. - no clue about updated ones for t2v.
Hi, Jellai. I've just demonstrated examples for i2v model. To avoid slowmotion, write prompts better and more detail, and specify that you don't want slowmo video. Also adjust resolution from 480p to 1024x576 . Also you can set 720p, but it's longer generation for i2v and u may get unstable result with 720p
@Amatiramisu I have tested many light loras, older and newer. In my opinion, lightx t2v distill rank 256 bf 16 is the best light loras I have used
@Stefan_Falkok That's fine. Just letting you know how the model is being presented, in case you wanted people to see the capability, and not make them think it has a common problem that people experienced with lightning.
Hello, could you please help and describe in more details how to get sage attention work? I get some errors when running the workflow
U can turn off sageattention - the speed of generation will be within the margin of error