Serelius Flow Video
Turn text scripts into complete videos with audio and subtitles inside Google Flow.
Serelius Flow Video is a specialised Flow Tool designed to run directly inside Google Flow. It guides creators, agencies, and businesses from a written script to production-ready videos across a structured 5-phase pipeline: Script → Style & Identity → Storyboard → Videos → Final Assembly.
1. What is Serelius Flow Video
The core philosophy of Serelius Flow Video is plan before generating. By validating character character sheets, visual style rules, and storyboard frames beforehand, users prevent wasted AI generation credits and ensure visual continuity.
2. Key Facts
| Fact | Detail |
|---|---|
| Execution Environment | Runs inside Google Flow as a tool built in the Tool Creator. Operates entirely in the browser without external servers or databases. |
| AI Generation Models | Uses Google Flow native AI models (Nano Banana for images, Omni and Veo 3.1 for video, Gemini 2.0 Flash for text). The tool orchestrates and edits. |
| Credit Consumption | Image and video generations consume Google Flow workspace credits. The tool itself charges no recurring generation fee. |
| Zero-Cost Operations | Structural script breakdown (scenes, speakers, dialogue) is deterministic code requiring zero AI credits. Optical transitions, captions, and export run locally without AI. |
| Interface Languages | English, Brazilian Portuguese, and Spanish, switchable via header toggle. |
| Spoken Video Languages | Brazilian Portuguese, US English, and neutral Spanish, configured per project. |
| Aspect Ratios | 16:9 widescreen (horizontal) and 9:16 vertical. |
| Project Persistence | Saved locally in browser IndexedDB, with full JSON project export and import. |
3. How It Works
The tool organizes the entire production pipeline into 5 distinct phases:
ROTEIRO ──► ESTILO & IDENTIDADE ──► STORYBOARD ──► VÍDEOS ──► MONTAGEM FINAL
(texto) (aparência + (um quadro (um clipe (legenda, transições,
fichas dos por cena, por cena, download dos clipes,
personagens) aprovado) com som) EDL, legendas, JSON)
| Phase | Input | Processing | Output |
|---|---|---|---|
| 1. Script | Structured script or raw story/article/lesson | Divides into 8s scenes, detects characters, lines, voice-over, locations, and SFX without AI. Optional AI script generation from raw text. | Analyzed scenes list + character profiles |
| 2. Style & Identity | Analyzed scenes | Defines FACT and NARRATIVE style rules; generates 1:1 character reference sheets; registers logos and props; tests visual style. | Saved style rules + visual character references |
| 3. Storyboard | Visual style + character sheets + script | Generates one framing-accurate frame per scene in chosen aspect ratio before video generation. | Approved storyboard frames per scene |
| 4. Videos | Storyboards + script | Generates scene video clips with voice, SFX, speech pacing, and continuity anchors. | Rendered MP4 video clips per scene |
| 5. Final Assembly | Rendered video clips | Applies zero-credit optical transitions and burned-in captions, generates AI Match-Cut transitions (optional), and exports. | MP4 clips, SRT/VTT subtitles, transcript, EDL, JSON |
4. Operation Modes
The tool provides two operating modes suited for different stages of production:
- Review Mode (Default): The creator triggers each phase manually, inspects results, and approves visual assets before spending credits on videos.
- Automatic Mode: Clicking START automatically sequences character sheets, storyboards, videos, and final assembly according to active settings.
Ready to build and launch your SaaS?
Get 100% full source code ownership, no proprietary wrappers, with architecture ready for millions of requests on Cloudflare.