Simple Lora Loader for Hunyuan Video

Simple Lora Loader for Hunyuan Video is a basic workflow for loading one or more LoRAs for use with Hunyuan Video; it uses Hunyuan, LoRA, VAE, and text-encoder model files and includes SaveAnimatedWEBP for animated WEBP output.

The graph includes CLIPTextEncode, EmptyHunyuanLatentVideo, LoraLoaderModelOnly, KSamplerSelect, SamplerCustomAdvanced, VAEDecodeTiled, and SaveAnimatedWEBP nodes.

The workflow names CLIPTextEncode and EmptyHunyuanLatentVideo, and includes SaveAnimatedWEBP for animated WEBP output.

The workflow is published by Bellydreams and is described as modified from a ComfyUI example.

The workflow names these model files: clip_l.safetensors; hathaway style_epoch22.safetensors; hunyuan_video_vae_bf16.safetensors; Hunyuan/hunyuan-video-t2v-720p-Q3_K_S.gguf; llava-llama-3-8b-v1_1-int4.gguf.

Named nodes include BasicGuider, BasicScheduler, CLIPTextEncode, DualCLIPLoaderGGUF, EmptyHunyuanLatentVideo, FluxGuidance, KSamplerSelect, and LoraLoaderModelOnly.

Additional named nodes are ModelSamplingSD3, RandomNoise, SamplerCustomAdvanced, SaveAnimatedWEBP, UnetLoaderGGUF, VAEDecodeTiled, and VAELoader.

Stack LoRAs by adding more LoraLoaderModelOnly nodes as needed.

A version for GGUF-quantized Hunyuan Video models is included.

Before use, verify that your setup includes the listed model files.

Vorschlag · nicht geprüft

Before use, verify that your ComfyUI setup includes the listed node types.

Vorschlag · nicht geprüft