Image-to-Video First and Last Frame v1 (Official 2 4)

1. Upload image

It is recommended to upload two images with the same resolution; otherwise, the last frame image will be cropped to match the size of the first frame image.

2. Select lora

Wan2.2 model is actually composed of two models, high noise and low noise. Therefore, when selecting lora, it is necessary to choose the corresponding high noise and low noise lora to achieve better results. If you are not familiar with this aspect, it is not recommended to use lora, and the weight can be kept at 0.

3. Prompt words

If it is a micro-animation video, there is no need to write storyboard prompt words. If you want the character or camera to have a larger range of motion, it is recommended to fill in the storyboard content. Below is the official guideline for using prompt words: https://alidocs.dingtalk.com/i/nodes/jb9Y4gmKWrx9eo4dCql9LlbYJGXn6lpz

3. Video duration
For videos longer than 5 seconds, it is recommended to fill in complete storyboard prompt words to make the video’s camera and character movements more detailed. If the effective content of the prompt words is too little, after the model generates the corresponding content, the remaining duration will be generated again based on the prompt words, resulting in the final effect showing repetitive character movements.