
Here is the workflow pictured:
https://civitai.com/models/1112672/fasthunyuanvideo6step
Posting two derivative HunyuanVideo models here for convenience for others. I originally found them here:
https://huggingface.co/Kijai/HunyuanVideo_comfy/tree/main
You can find the VAE there as well ^^^
Files Section has two files --->
-
Full model is the Regular HunyuanVideo FP8 model (CFG6, Flow 9).
-
Pruned model is the FastHunyuanVideo FP8 model (CFG 9, Flow 17).
N.B. I realize that they are the exact same size, but they are in fact different files for different purposes.
Instead of using the Fast checkpoint to lower steps, you may use this LORA.
LORA: https://huggingface.co/Kijai/HunyuanVideo_comfy/blob/main/hyvideo_FastVideo_LoRA-fp8.safetensors
描述:
FastHunyuanVideo Model
训练词语:
名称: fasthunyuanvideo720p_fp8.safetensors
大小 (KB): 12876011
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success