[Framepack] Intuitive left to right framepack workflow | 프레임팩 워크플로우 — v2.0

This Framepack workflow is for simplifying basic Framepack tasks in a left-to-right layout; the page does not specify its particular inputs or outputs.

The page names model files and ComfyUI folders for setup, but does not identify a generated output format.

Place FramePackI2V_HY_fp8_e4m3fn.safetensors in ComfyUI\models\diffusion_models, sigclip_vision_patch14_384.safetensors in ComfyUI\models\clip_vision, and hunyuan_video_vae_bf16.safetensors in ComfyUI\models\vae.

Vorschlag · nicht geprüft

Place llava_llama3_fp16.safetensors or the optional llava_llama3_fp8_scaled.safetensors, plus clip_l.safetensors, in ComfyUI\models\text_encoders.

Vorschlag · nicht geprüft

Model files named for this workflow include clip_l.safetensors, FramePackI2V_HY_fp8_e4m3fn.safetensors, hunyuan_video_vae_bf16.safetensors, llava_llama3_fp16.safetensors and sigclip_vision_patch14_384.safetensors; llava_llama3_fp8_scaled.safetensors is also listed as optional.

Listed ComfyUI node packs are comfyui_essentials, comfyui-kjnodes and comfyui-videohelpersuite.

Set Attention_mode to sdpa or sageattn; if sageattn fails, switch to sdpa.

Vorschlag · nicht geprüft

If TORCH_LOGS="+dynamo" causes an error, unset it or set it to an empty value.

Vorschlag · nicht geprüft

Before using the workflow, check that the named model files and node packs are available in your ComfyUI setup.

Vorschlag · nicht geprüft

The page does not list a license; review its terms before reusing the workflow.

Vorschlag