Edit lab · 7.5 seconds
Hear the next shot first.
A vault cuts to a spinning wheel. Both passes use the same picture frames. In pass B, the next shot’s audio fades in 300 milliseconds before the cut.
Earlier Seedance 2.5 clips with staged action and fictional adults. Original generated audio, cut and level-matched locally. The source files are 720 × 1280 and the annotated export is 1080 × 1920.
Make the same J-cut
- Keep 0.250–1.750 s from Green Means Go and 1.000–2.500 s from Night Run. Use the source files linked on those pages.
- Keep the picture cut at 1.500 s. In pass B, extend the incoming audio edge to include its source from 0.700 s. Keep the original sound in sync once the second shot is on screen.
- Blend the two audio tracks over the 300 ms overlap. Compare at the same playback volume. Extend or shorten the overlap if it steps on dialogue.
Rebuild the two audio passes
Save the Green Means Go source, the Night Run source and the replay script in the same directory. With Python 3 and FFmpeg on your path, run the command below.
python replay.py green-means-go-original.mp4 night-run-original.mp4 j-cut-outputThe script writes two 3-second WAVs and checks that their samples match outside the 300 ms overlap. These are the audio passes before final mastering. Put either WAV under the same picture cut at 1.500 s.
Planning an action shot? Open the free prompt planner. Generation is a separate step.