Wan 2.2 Fast Img2Video + Auto-Captioning (Low VRAM Friendly) — v1.0

This ComfyUI workflow turns a starting image and your motion prompt into a 2.5-second video, using automatically generated image tags to guide image-to-video generation.

WD14 Tagger analyzes your image and generates tags, then the workflow combines those tags with your motion prompt before generating the video.

Upload your starting image in the “Load Image” node.

Enter the movement you want in the “ttN text” node; examples include smiling, wind blowing, or blinking.

The video is saved in WebP and WebM formats.

Use ComfyUI Manager’s “Install Missing Custom Nodes” option if custom nodes are missing.

Sugerencia · sin verificar

Required node packs: comfyui-custom-scripts, ComfyUI_JPS-Nodes, comfyui_tinyterranodes, and comfyui-wd14-tagger.

Required model files: UNET `wan2.2_ti2v_5B_fp16.safetensors`, VAE `wan2.2_vae.safetensors`, and CLIP `umt5_xxl_fp16.safetensors`.

Generation is described as using the Wan Video 2.2 TI2V-5B base model and 10 steps.

The listing does not specify a license for this workflow.