Before you begin
You’ll have a repeatable reference-to-reference workflow, a fallback path for source clips that a platform will not accept, and copyable prompts for locking two replacement characters to the original performers without identity or wardrobe mixing.
Tools used
- Seedance 2.5
- Flova AI or another platform with Seedance 2.5
- TECH4SSD Video → Depth tool
- CapCut
Requirements
- A source/reference video you are permitted to use
- Two replacement character images or character sheets
- Access to an AI platform offering Seedance 2.5
- CapCut or another editor for final audio/editing if needed
On this page · 10 steps
Your progress
0 of 10 steps
The walkthrough
Step 01 / 10
Prepare the exact reference clip
Start with the exact section of the video whose movement you want to recreate. Trim it to the duration you plan to generate. The reference clip should be treated as choreography and camera guidance: movement, timing, body positions, interaction, framing and camera trajectory. Use footage you own, licensed footage, public-domain footage, or material you are otherwise permitted to use.
Step 02 / 10
Prepare Character 1 and Character 2
Create one clear reference for each replacement character. A character sheet is preferable when the performer turns around or changes angle. Include front/full-body, back/full-body, face close-up, outfit and important accessories. Character 1 should map to Source Performer A; Character 2 should map to Source Performer B. Keep those mappings fixed for the entire video.
Check before continuing
Continue from here