A brief.
A first draft you can edit.
Describe the product, audience and pictures. A local model suggests short spoken lines. Review the wording, then prepare complete voices and editable typography in the same Film project.
A useful starting point.
This brief asked for warm, simple coffee copy. The picture descriptions were checked against actual decoded footage.
Your judgment stays in the cut.
The model proposal left existing lines untouched. Two suggestions were edited before narration; this is reviewed assistance, not automatic factual verification.
Keep the whole line.
Three complete local voices, eleven editable word timings and a 5.33-second portrait export. The final sequence is grounds, water over grounds, then coffee into the cup.
The suggestion, and the authored cut.
| Original local model proposal | Reviewed spoken line |
|---|---|
| Pouring coffee beans into a filter. | Start with the grounds. |
| Water flows over the grounds. | Water meets the grounds. |
| Make mornings yours. | Make mornings yours. |
A supplied closing line is retained exactly. Suggestions use written descriptions, not video vision, and can add unsupported details. The native UI requires an explicit action to copy them into your written script; it does not automatically voice or export them.
The finished cut stays editable.
Actual native UI export · 720 × 1280 · 24 fps · 128 frames. The original model proposal, authored edits, complete voices, source hashes and measured evidence are included. Local Qwen3-4B copy took 5.92 seconds of CPU generation on the recorded machine; no general speed claim is made. New local suggestions require the optional model setup. The public page plays recorded output and accepts no inference requests. Footage: Michael Burrows / Pexels, under the Pexels License. Fictional coffee concept, generic local Kokoro voices, no music or identity cloning. Model licenses apply separately; app code is MIT. Final listening, creative approval and submission remain with Jay.