model_download:

https://huggingface.co/ali-vilab/VACE-Wan2.1-1.3B-Preview/tree/main

wan2.1模型对于尺寸的输入总是苛刻的，所以这又是一个自动计算尺寸的分组。不论你输入的尺寸是什么比例，结果总是正方形的，因为他有着严格的尺寸需求，并且正方形可以更好的展示效果。除非你的原始图像也是正方形，那么你可以关闭这个分组并使用自定义外补尺寸分组。

The wan2.1 model is always very strict about the input size, so this is another group that automatically calculates the size. No matter what ratio you input, the result is always square, because it has strict size requirements and the square can show the effect better. Unless your original image is also square, you can turn off this group and use a custom external size group.

如果你真的决定使用自定义尺寸，尽量通过计算保证你外补以后的图像尺寸为480*832/832*480。这对于结果有着强烈的影响。

If you do decide to use a custom size, try to calculate and ensure that the image size after your padding is 480*832/832*480. This has a strong impact on the result.

注意最终输出和输入帧数也有很大关系，最好控制在20帧。这个节点仍然属于实验性，很多方面仍然在测试。

Note that the final output and input frame number are also closely related, and it is best to control it to 20 frames.I think this node is still experimental and many aspects are still being tested.