Dear author, thank you very much for your hard work. Your project is excellent. However, I have encountered an issue. The generation effect is great when the structures of images and videos are consistent, but it becomes much worse when they are inconsistent. For example, if the video shot pans upward from the lower body to the upper body, the upper body turns out poorly when the reference image only shows the lower body; the lower body looks bad when the reference image only shows the upper body; and the overall effect is poor if a full‑body image is provided. How can this problem be solved?
Dear author, thank you very much for your hard work. Your project is excellent. However, I have encountered an issue. The generation effect is great when the structures of images and videos are consistent, but it becomes much worse when they are inconsistent. For example, if the video shot pans upward from the lower body to the upper body, the upper body turns out poorly when the reference image only shows the lower body; the lower body looks bad when the reference image only shows the upper body; and the overall effect is poor if a full‑body image is provided. How can this problem be solved?