【WAN2.1】IMG to VIDEO — 🟡 v2.1

Use this ComfyUI workflow with an input image, written prompt, and required model files to generate Wan Video clips; it also supports text-to-video generation, optional SeedVR2 tiled upscaling, and images saved with generation parameters.

Choose the auto version for automatic model downloads, or use the manual version with the listed model files.

Turn the built-in SeedVR2 tiled upscaler on or off.

Manual setup calls for an I2V Model or I2V Quant Model, plus CLIP, CLIP-VISION, VAE, Speed LoRA, and an upscale model.

The workflow outputs video, and saved images can include all generation parameters for online publishing and remixing.

In ComfyUI, enable Settings → “Use Nodes V2 (Vue),” refresh the page, and then load the workflow.

Install ComfyUI-UmeAiRT-Toolkit through ComfyUI Manager by searching “UmeAiRT,” or use the UmeAiRT Auto-Installer.

Listed model files and model sources are: clip_vision_h.safetensors, MiaoshouAI/Florence-2-base-PromptGen-v2.0, rife47.pth, umt5_xxl_fp8_e4m3fn_scaled.safetensors, wan_2.1_vae.safetensors, and wan2.1-i2v-14b-480p-Q5_K_M.gguf.

Listed ComfyUI node packs are: comfyui-custom-scripts, comfyui-easy-use, comfyui-florence2, comfyui-frame-interpolation, ComfyUI-GGUF, comfyui-kjnodes, comfyui-multigpu, comfyui-mxtoolkit, comfyui-videohelpersuite, rgthree-comfy, and was-node-suite-comfyui.

Use the three LoRA slots with individual on/off toggles and strength controls; additional LoRA modules can be connected.

For manual GGUF setup, choose Q8, Q5, or Q3 for the 480p or 720p I2V quantized model.

Enable Nodes 2.0 (Vue) before loading; otherwise, the workflow may have display problems.

Before running, verify that all six listed model entries and eleven listed node packs are available in ComfyUI.

Vorschlag · nicht geprüft