All work fal.ai · Spec Work · 2026 06 / 31

fal — Seedance 2.5 Text to Video

A 26-second concept explainer for Seedance 2.5 on fal: one prompt goes into the playground, a whole scene comes out.

Unofficial concept. Not affiliated with or endorsed by fal.ai.

Concept explainer for the Seedance 2.5 text-to-video endpoint on fal. The film follows a single prompt — Ember vs Gale, a live-action anime duel — from the model picker, through the prompt and its settings, to the generated clip, and lands on the line: a few words in, a whole scene out. The clip in the result panel is a real Seedance 2.5 generation on fal from that exact prompt, not a mock-up.

Built in After Effects

The whole fal playground is rebuilt as one flat UI in world space and filmed with a single 3D camera, so the moves between prompt, settings and result are camera moves rather than cuts. The prompt types itself with text animators, the status readout and waveform run on expressions, and Geist and Geist Mono keep the type on-product. Dark theme, ember glow to match the prompt.

Storyboard

Storyboard frames from the Seedance 2.5 explainer

More work

All work (31) ↗