Technical deep-dive
How VeoStudio works.
The models, the running order and the reasons behind it. Six stages take a paragraph of description to a finished 1080p film, and each one hands a real artefact to the next.
7-day trial · 50 tokens · no credit card
The pipeline
Six stages, one film.
Every stage produces something you can look at and change before the next one runs. Nothing is a black box, and the expensive step is always the last one.
End to end
- Stages
- 6
- Scenes
- Up to 99
- Shot
- 8s, audio included
- Scene
- Up to 148s
- Export
- 1080p MP4
- 01
AI script writer
The AI Director
Structured, multi-scene scripts with precise control over every beat.
- Up to 99 scenes per project
- Each scene carries prompt, camera work, lighting and timeline
- Automatic scene continuity tracking
- Character placement and action
- Dialogue and narration
- Cost
- Included with the Pro plan ($25/month)
- Output
- Structured JSON with complete scene descriptions
- 02
Character generation
Nano Banana 2 or Nano Banana Pro
Consistent characters, generated fast or photorealistic depending on the job.
- Nano Banana 2: 1 token per image, for fast iterations
- Nano Banana Pro: 4 tokens per image, photorealistic and cinematic
- Reference image support with Nano Banana Pro
- Character IDs for project-wide reuse
- Facial feature consistency across scenes
- Cost
- 1–4 tokens per image
- Output
- High-resolution character images with unique IDs
- 03
Frame generation
Nano Banana 2 (text + image → image)
The first and last frame of every scene, drawn before anything is filmed.
- The AI describes the opening and closing shot
- Nano Banana 2 generates the frame images
- Character images injected as references
- Scene composition preview
- Continuity validated before video generation
- Cost
- 1–4 tokens per image
- Output
- First frame + last frame images per scene
- 04
Veo 3.1 video generation
Google Veo 3.1 (Quality / Fast)
Eight-second 1080p clips with synchronised audio already in them.
- Quality mode: 60 tokens per 8s clip
- Fast mode: 30 tokens per 8s clip
- First frame used as the conditioning image
- Last frame used for scene continuity
- Synchronised audio generated automatically
- Cost
- Quality: 60 tokens/clip, Fast: 30 tokens/clip
- Output
- 8-second 1080p Veo 3.1 video with audio
- 05
Continuity system
Text + image continuity
What keeps the visual and narrative flow intact between scenes.
- Text continuity: closingAction → openingAction
- Image continuity: previous last frame → next first frame
- Character tracking across scenes
- Automatic scene-to-scene transitions
- Timeline synchronisation
- Cost
- Free, and automatic
- Output
- Seamless scene transitions
- 06
Movie maker
Drag-and-drop timeline
Clips stitched into a full movie of any length, ready to hand over.
- Drag-and-drop interface
- Trim and reorder clips
- Preview full sequences
- Export in multiple formats
- Commercial licensing included
- Cost
- Free, no export fees
- Output
- Complete video files ready for distribution
Technical specifications
The whole spec sheet, on one page.
AI models used
- The AI Director
- Script and scene planning
- Nano Banana 2
- Fast character and frame stills
- Nano Banana Pro
- Photorealistic, cinematic stills
- Google Veo 3.1
- Video, in two modes
Video specifications
- Resolution
- 1080p (1920×1080)
- Aspect ratios
- 16:9, 9:16, 1:1
- Duration
- 8 seconds per clip
- Audio
- Generated automatically
- Format
- MP4
Limits and constraints
- Scenes per project
- 99 max
- Clip duration
- 8s, to 148s per scene
- Character consistency
- Via Nano Banana 2
- Token top-ups
- Any time
- AI script writing
- Included
A clip is eight seconds because that is the native Veo 3.1 shot. Continuity frames are what stretch a single scene to 148 seconds, and a production holds up to 99 shots stitched into one film.
Why this architecture
Built in stages on purpose.
One model doing everything would be simpler to explain and worse to use. Splitting the pipeline is what makes the cheap steps cheap and the expensive step optional.
The trade
- Planning
- Free
- Stills
- 1–4 tokens
- Video
- 30–60 tokens
- Re-render
- Per scene
Best model for each task
The AI Director handles script writing. Nano Banana Pro delivers cinematic images. Veo 3.1 is the state-of-the-art video generator. No single model is asked to do all three.
Cost optimisation
Script writing is included with the subscription. Images cost 1–4 tokens, so you preview cheaply before spending tokens on video.
Quality control
Preview frames before generating expensive video. Problems get caught early in the pipeline, on the cheap step.
Maximum flexibility
Full control at every stage: adjust the script, regenerate frames, choose Quality or Fast mode per scene.
Ready to try it
Run the pipeline yourself.
The 7-day trial starts you with 50 tokens, which is enough to write a production, cast it, draw the frames and film a shot.
7-day trial · 50 tokens · no credit card