tomclive/seedance2stitcher

SeeDance 2 Stitcher is a small local tool for joining together two AI generated video clips from a system called SeeDance 2, specifically for the common case where the end of one clip and the start of the next were generated separately and do not line up perfectly in brightness or contrast. Rather than manually trimming and color matching clips by hand, this tool automates the process. The workflow starts by dragging two video files into a browser based interface. The tool then looks at the frames near where the two clips would meet and detects where they overlap, suggesting which frames to trim so the join feels continuous rather than repeated or jumpy. It can smooth out any mismatch in brightness or contrast right at the seam, or apply that same color correction across the entire second clip if the mismatch runs throughout. Audio from both clips is preserved through the export. Finished files are given descriptive names that record which trimming and correction settings were used, and the tool can also export labeled side by side comparison images showing the original unedited join next to the corrected result. Running it requires Node.js version 18 or newer along with FFmpeg and FFprobe installed on the machine, either available on the system path or pointed to directly through environment variables. Starting the tool launches a small local web server, and the interface is reached by opening a browser to that local address. Sample videos can optionally be placed in a folder for testing before using your own footage. The readme notes that results work best when the original clips only contain natural sound effects and dialogue, since adding a music soundtrack afterward avoids the problem of two separately generated music beds rarely lining up cleanly once stitched. The project is written in JavaScript and released under the MIT license.