I2V Long Video Extension Master — I2V Infinite Extension

This image-to-video workflow uses an input image and text to generate video, offers extending options, and provides preview images, saved images, and combined video output.

The workflow loads an image, resizes, scales, and batches it, encodes visual and text inputs, runs WanImageToVideo, decodes frames, and combines them into video.

Image inputs can be loaded, resized, scaled, batched, and vision-encoded before video generation.

Text input uses CLIPTextEncode and ttN text; outputs include PreviewImage, SaveImage, and VHS_VideoCombine.

The listed base model is Wan Video 2.2 I2V-A14B.

The listed version is I2V Infinite Extension, published by Sinusplusminus.

The workflow requires these ComfyUI node packs: comfy-mtb, comfyui_essentials, comfyui_llm_party, comfyui_tinyterranodes, and ComfyUI-Crystools.

It also requires comfyui-easy-use, comfyui-kjnodes, ComfyUI-QualityOfLifeSuit_Omar92, comfyui-workflow-encrypt, and mikey_nodes.

The workflow lists these model files: clip_vision_h.safetensors, high_noise_model.safetensors, low_noise_model.safetensors, umt5_xxl_fp8_e4m3fn_scaled.safetensors, wan_2.1_vae.safetensors, wan2.2_i2v_high_noise_14B_fp8_scaled.safetensors, and wan2.2_i2v_low_noise_14B_fp8_scaled.s

Controls include batch count, integer and float values, loops, comparisons, and strings.

Other controls cover image resizing and scaling, LoRA loading, model settings, sampling, previews, and video combining.

Before use, verify that every listed model file is installed under its exact filename.

Suggestion · non vérifié

No license terms are provided, so check Sinusplusminus’s terms before reusing the workflow.

Suggestion · non vérifié