This is the Wan 2.2 Image to Video workflow. You can generate videos based on uploaded images and prompts.

Uploadable images are the Start Image (the image that will be the first frame of the video) and the End Image (the image that will be the final frame of the video).

If you upload two images, Wan 2.2 will adjust the animation between them so that the final frame is the uploaded image.

If you don't use an End Image, you can generate a video using only the Start Image by right-clicking and selecting "Bypass."

In addition, three Wan Video Select Multi nodes, each with five LoRA slots, are used, for a maximum of 15 slots (only 10 can actually be used).

Block Swap is also implemented to reduce VRAM consumption to a certain extent.

TIME COST

For Free Generation users, here are the estimated time costs.

When generated with the default settings (without specifying the final frame), generation took approximately 2 minutes 30 seconds to 3 minutes.

If you specify an final frame, the generation time will be extended, and may take 3 to 7 minutes.

Generation times may vary depending on the settings. Additionally, generation times may change during peak service periods. Please consider this time as an estimate only.

HOW TO USE

1. Upload an image.

If you do not need an End Image, right-click the node and select "Bypass."

2. Enter a prompt.

What kind of video do you want? Negative prompts are optional.

3. Video Length

Specify the number of frames. This is the number of images per FPS (frames per second). The default FPS is 16 and the length is 81 frames. This creates a video of approximately 5 seconds.

4. Resolution Settings

Set the long side of the resolution. Increasing the resolution beyond the default may result in generation failure or significantly longer generation times, so only tweak this setting if you have the knowledge to change other settings.

5. Loop Count

This function combines multiple loops of a video. In other words, setting this to 1 will create a single video by repeating the exact same 5-second video twice. This setting is rarely necessary.

6. Press the Generate button.

LoRA

There are a total of 15 LoRA slots (three loaders, each with five LoRA slots each).

However, only a maximum of 10 LoRA slots can be used at one time.

Wan2.2 divides sampling processing into two systems: high noise and low noise. LoRA for Wan2.2 is usually available in HIGH/LOW versions, and two must be loaded simultaneously to achieve 100% performance.

Red nodes are HIGH nodes, blue are LOW nodes, and gray nodes are used to load LoRA not separated into HIGH/LOW or LoRA for Wan2.1.

You can load any LoRA and adjust its strength.

Known issues.

1. Pipe Bug

Occasionally, a workflow may fail to load, causing pipes (wires connecting nodes that run freely across the screen) to become disconnected.

Reloading the workflow will fix this.

2. VRAM Bug

Occasionally, an error due to insufficient VRAM may occur.

This is an error on the SeaArt site's side, and it often occurs during busy times. Please try again later.

3. Model and LoRA Loading Bug

This occurs randomly. Regenerating the model may resolve the issue.

UPDATE-------------------------------------------------------------

2025/11/16

The layout has been changed to make pipes invisible.

WAN2.2 Image to Video 10 LoRA ワークフロー

評価とレビュー