1. Story
Paste or type a script. Studio reads it and proposes a structure:- Stages — the high-level beats of the story (opening, turn, resolution).
- Beats inside each stage — the smaller narrative units, the size of “she walks across the room and picks up the photo”.
2. World
A story has a world. The World section is where you describe it:- Characters — name, appearance, attire, voice, and any continuity notes (“always wears the locket”, “left arm in a sling from stage 3 onward”).
- Locations — the places the story happens in, with reference images and notes on what’s distinctive about them.
- Props — objects that matter to the story.
- Factions and relationships — who’s allied with whom, who knows whom, who hates whom.
3. Assets
Where World captures who and what, Assets captures the look. For each character, location, or major prop, you lock a reference:- One or more reference images that define the visual identity.
- Notes on what must stay the same between shots and what is free to vary.
4. Shots
Shots is the workshop. Here you:- Build the shot board. Each beat becomes a row; each shot becomes a card on the row.
- Compose each shot. Pick which assets appear in the shot, describe the camera intent (framing, lens, movement), and generate the opening frame.
- Direct the take. A take is one generated version of the shot. Generate one or more takes from the composed opening frame. If a take captures what you wanted, accept it. If not, refine the composition and generate another take.
- Chain shots into clips. A shot can be a single short clip or a sequence chained together for longer continuous motion.
- Assemble the rough cut. A rough cut is a first assembly of shots in order, before audio polish. Drag the accepted takes into the order you want them to play.
5. Dubbing
Sound is the last layer:- Music. Describe the cue you want — style, mood, tempo, key, duration, and optional lyrics — and Studio composes it. Different cues can be assigned to different ranges of the cut.
- Voice. Provide a reference voice sample and Studio can speak any line in that voice. Multiple languages are supported for voice cloning so you can localise without recasting.
- Sound effects. Studio listens to the video and proposes effects that match what’s on screen — the door creaks when the door opens. You approve, refine, or replace each effect.
Export
Export is in development. Today the primary artifact is a handoff package: a downloadable JSON manifest of the selected takes, with each shot carried as a unit — its story trace, selected take, clip summary, camera intent, sound plan, and readiness. The manifest records, per package, whether an assembled video is available. A single assembled MP4 of the rough cut is available as an early alpha path, produced locally from the selected takes and the temporary audio. When local assembly isn’t available, the package keeps a watchable web preview as the truthful fallback, so you can still review the cut in order. More finished render output is on the roadmap.Concepts
Two ideas come up across every section.Asset-first
Single-prompt video tools take a sentence and try to generate something that matches it; if the next clip doesn’t line up with the last one, the only fix is to roll again. Studio inverts that. You lock the visual identity of each recurring element — character, location, prop — before you start generating shots. Every shot then resolves that identity through the lock, not through a sentence of text. The practical consequence: editing a character changes one reference, not twenty shots. Re-renders are surgical, not whole- project.Story structure vocabulary
Studio uses a deliberate vocabulary for the layers of a project, so the same word means the same thing everywhere in the product:
When something in the product mentions a “beat” or a “shot”, it
means this exactly — there’s no overlap or aliasing between layers.
Next
- Request access to start a project of your own.
- For early-access enquiries write to sales@cyberun.cloud.
