WorkflowListedExternal

jhj Hunyuan Video Workflows — v1.0

Only txt2vid workflow for now.

TaskUpscalingFamilyHunyuanEstimated VRAMNot yet characterizedBgrade

Runtime profile

Source description

jhj Hunyuan Video workflows

Only txt2vid workflow for now.

It loads GGUF from:

It first generates video with smaller resolution -> then upscales for the limits of VRAM.

Probably you better find a better upscaling model for your own purpose.

The version is controlled in :

AI-generated commentary

AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.

This Hunyuan Video workflow creates video from a text prompt and upscales the result.

It is described as generating video at a smaller resolution first, then upscaling it for VRAM limits.

Named model files include hunyuan-video-t2v-720p-Q4_K_S.gguf, hunyuan_video_vae_bf16.safetensors, clip_l.safetensors, llava_llama3_fp8_scaled.safetensors, and 4x_foolhardy_Remacri.pth.
Named ComfyUI nodes include CLIPTextEncode, CR Apply LoRA Stack, DualCLIPLoader, easy loraStack, EmptyHunyuanLatentVideo, ImageScale, and ImageUpscaleWithModel.
Other named nodes are KSamplerAdvanced, UnetLoaderGGUF, UpscaleModelLoader, VAEDecodeTiled, VAELoader, and VHS_VideoCombine. The listed node packs are ComfyUI-Easy-Use and ComfyUI_Comfyroll_CustomNodes.

It loads GGUF from city96/HunyuanVideo-gguf on Hugging Face; its version is controlled in jhj0517/ComfyUI-workflows.
You may want to choose an upscaling model that suits your needs.
Suggestion · not verified

Does this need to be edited?

Sources

1 source

CivitaiPrimary

Verified Jul 20

Source excerpts

1 excerpt

Source context: 429 downloads · Type Workflows · Base model Hunyuan Video