xixihhhh/clipforge ? reverse-engineered prompt

Build me an open source app that turns a single product photo into a short selling video.

I want to upload an image, have AI pull out the main selling points, write a catchy script, then automatically create a finished video with matching visuals, voiceover, subtitles, and background music. The product image should stay clear and not get warped, and the final result should be ready for places like Douyin, Kuaishou, Xiaohongshu, TikTok Shop, and similar short video platforms.

Please make it so I can also generate videos from just one theme or idea, not only product images. It should support self hosting, work locally, and let me make lots of videos quickly without watermarks. If possible, make the flow simple enough that a beginner can use it, but still useful for power users who want to tweak things. Use current docs online if you need to.