gpt-imageVIDEO
用 GPT-Image-2.5 生成分镜做动作迁移工作流
作者用 GPT-Image-2.5 加故事板生成帧，再配 h3 max 做动作迁移，成品未放大，附可试玩链接。
Azer Koçulu
@azerkoculu2026.09.23收录原推发布于 2026.09.22
Azer Koçulu @azerkoculu原推发布于 2026.09.22
h3 max is really good.made workflow for transferring motion control yesterday. this example is straight from h3 max, without upscaling. generated the frames with gpt-image-2.5 + story board. you can try it at https://mitte.ai/app/hotel-lobby-character-swap https://x.com/DesignArena/status/2102182411555790882
这条是引用着另一条写的：x.com/designarena/status/2102182411555790882
x.com/azerkoculu/status/2102358192063656240采集于 2026.09.23 10:15