Back to Prisms

Marketing·recipe
GPT Image 2 + Seedance 2.0
Text → multi-shot scene in one go
Describe a scene. We render a high-detail still with GPT Image 2 and animate it into a 5-second Seedance 2.0 clip (native audio). Pick 2-4 shots and Sonnet plans a continuous beat list — each shot gets its own still + 5s clip, so a 4-shot run gives you a ~20-second multi-shot sequence with character/setting continuity. Every still and clip lands in your library.
175 cr~2 minutes (per shot, runs in parallel)beginner· 16:9
workflowimagevideogpt-imageseedancei2vmultishot
Try these
Click an example to fill the prompt field.
e.g. "a vintage diner at dusk, neon sign flickering, rain on the window" — or write a multi-beat brief if you picked more than 1 shot
Camera + subject motion. Leave blank for a gentle cinematic push. With multiple shots, this is a fallback — Sonnet writes per-shot motion from the brief.
