JonXL's Multi Wan2.2 T2I / T2V / I2V + Clip Extender Workflow (+Upscaling & Interpolation) — v2a

A multipurpose easy to use workflow for generating text to video and (infinite) image to video, using 2-3 simple toggles.

Configure all models once and then easily switch between generation modes.

(See example video of a 25 second clip generated through frame extension)

Stay tuned for future updates.

Pro tip:

Use my Quick Video Stitcher workflow in conjunction to quickly stitch two (generated) video clips together (rinse and repeat):

https://civitai.com/models/1864873

Versions:

V1 - Has examples for all modes

V2 - Now only uses 1 Image loader that can be refreshed + Video Frame Dump (for manual frame picking) .. recommended.

Features:

Allows for T2V and I2V generation (using a single toggle)

Allows for single text to image generation with resolution up to 1920x1080

Allows for infinite video generation, by using the last frame as the input frame (for V1 turn on the toggle and generate to your heart's content :) )

Includes upscaling and frame interpolation

Supports (per toggle) simple resize and rescaling of input images instead of cropping

Multi Lora loading

Clip Vision for input image (though disabled by default)

Sage Attention for performance

Very useful for infinite video generation, just change the prompt and click on the "Generate button" once a clip has rendered, to start using the last frame for a new clip.

If needed, you can manually select the last frame too, if you didn't like the last clip generated. Use 'reset' if you want the latest generated frame again.

Toggles can be found in "Step 3 - Choose parameters".

Optionally you can quickly enable or disable frame interpolation and upscaling in the control panel above step 5 "Prompt".

Make sure to load a random image (perhaps the one that came with this workflow zip archive) to avoid the initial error for the "Load image" node. (while the image is not needed for text 2 image or video, ComfyUI likes to complains about it)

The example Lora used in this model can be found here: https://civitai.com/models/1862320/wan22-t2v-lora-cartoon-style )

This workflow does not stitch videos together, but you can use my Quick Video Stitcher Workflow to quickly stitch videos together : https://civitai.com/models/1864873

The zip file (V1) contains one workflow and three example images.

The workflow is preconfigured for Image to Video with Frame Extending on, 960x720x81 frames.

You can also drop in the example images to get a pre-configuration for T2I, T2V, I2V, I2V with frame extending.

For T2I it is recommend to install the following custom_node to use "res_2s" + "bong tangent" sampler: https://github.com/ClownsharkBatwing/RES4LYF

The 4 steps lightning Loras can be found here:

https://huggingface.co/Kijai/WanVideo_comfy/tree/main/Wan22-Lightning

Text encoder · Unknown