Build a recurring AI influencer identity first, then use that same character as…

Prompt metadata
Shared by	@Yashal Ai
Model	Seedance 2.5
Category	Build a recurring AI influencer identity first, then use that same character as the anchor for every Seedance 2.5 vlog to prevent face/outfit drift. Generate a clear opening selfie frame (face, hair, wardrobe, room layout, lighting, phone-camera perspective, mood) and use it as the visual reference input before running Seedance 2.5. Create a 16-beat storyboard from the opening frame and feed that storyboard plan into Seedance 2.5 so the 30-second video follows a continuous sequence instead of disconnected clips. Use Seedance 2.5 via Image to Video Ultra S to generate a single continuous video up to 30 seconds rather than stitching multiple short clips. When prompting Seedance 2.5, align the written prompt with the storyboard rhythm (camera behavior, timing, expressions, motion, tone) and avoid introducing new outfits/rooms/lighting that contradict the storyboard. Leverage Seedance 2.5’s multimodal references (up to 50 assets) to lock identity, styling, environment, motion, and camera direction more precisely. After generation, review for identity consistency, environment continuity, natural handheld camera feel, believable face/hands, and emotional pacing; fix issues by editing the storyboard/strengthening references/clarifying a specific timeline segment instead of restarting from scratch.
Dataset	renoise-ai · awesome-seedance-2-5-prompts

The prompt

Build a recurring AI influencer identity first, then use that same character as the anchor for every Seedance 2.5 vlog to prevent face/outfit drift.

Generate a clear opening selfie frame (face, hair, wardrobe, room layout, lighting, phone-camera perspective, mood) and use it as the visual reference input before running Seedance 2.5.

Create a 16-beat storyboard from the opening frame and feed that storyboard plan into Seedance 2.5 so the 30-second video follows a continuous sequence instead of disconnected clips.

Use Seedance 2.5 via Image to Video Ultra S to generate a single continuous video up to 30 seconds rather than stitching multiple short clips.

When prompting Seedance 2.5, align the written prompt with the storyboard rhythm (camera behavior, timing, expressions, motion, tone) and avoid introducing new outfits/rooms/lighting that contradict the storyboard.

Leverage Seedance 2.5’s multimodal references (up to 50 assets) to lock identity, styling, environment, motion, and camera direction more precisely.

After generation, review for identity consistency, environment continuity, natural handheld camera feel, believable face/hands, and emotional pacing; fix issues by editing the storyboard/strengthening references/clarifying a specific timeline segment instead of restarting from scratch.

Try in DraPub