VISIONPROMPT.STUDIO
Direction, not chance.
We're building a system to understand and control how AI visuals are created — from camera and light logic to structured prompts and reproducible results.
Not another generator. A production system.
Layout Boxes
Result
Place every person in the frame. A visible box guides FLUX.3 on position and size, for up to ten people in one image.
Kling 4.0 is in the studio. Take a finished shot and turn it into a moving one, without leaving your scene.
Built for campaigns — the same people, outfits and light across every shot, model and format.
Build your cast once — every shot draws on it.
Fix what has to stay the same, change only the rest.
Turn one setup into a full campaign.
The result: a campaign that looks like one shoot — not twenty separate generations.
Chapter 01 — Compose
Camera, light and cast are choices you make, not words you hope for.
The same people and looks carry a whole campaign.
From the first composition to the final clip in one place.
Camera, lens and light are objects you move on the 2.5D canvas. Set the angle you want and generate the same setup from there. One scene, written for every model.
The control room of every shot. You direct, the agents prepare, the optimizer writes — and nothing runs without you seeing what it costs. Guide Maps from your own images steer the result.
Start from one image and choose new perspectives in a full 3D viewport. As the camera moves, the background changes with it, so every angle is a new shot.
Chapter 02 — Cast
Build a cast, dress it and place it once. Every shot after that starts from the same people.
Define once. Reuse in every shot and every model.
Your people — recognizable in every shot.
Looks you assign per person and scene.
The same place, every time.
Start from a photo, a preset or your own hands on the figure. You decide how much you set yourself.
Every result keeps its DNA — the full recipe behind it. Come back weeks later, restore any version and keep producing in the same look.
Chapter 03 — Finish
Change the light, edit by chat or find a new angle, right next to the original.
Don't just generate. Finish. Describe a change in chat, and the result opens next to the original in the viewport — compare, accept or go back. The paid result is never thrown away.
Original
Relight
Multi-scene story creation for AI video generation.
From versioned storyboards to Seedance 2.5, Kling 4.0 and Veo 3.1.
Campaign-grade looks, generated in VPS.
One core studio, plus a full production toolkit.
Camera, lens and light as objects — one scene for every model.
Your cast, their looks and places — defined once, reused everywhere.
From fast drafts to high-end image models.
Every result keeps its recipe — restore any version.
Every image and clip stored and organized.
Orbit the scene and render any angle.
Take a pose from a photo and use it in your scene.
Move the light on a finished image, keep the composition.
Chat edits, masks and precise local edits.
Up to 4× for print, plus video upscaling.
Kling 4.0, Seedance 2.5 and Veo 3.1 — scene by scene.
Agents, prompt optimizer and cost control in one place.
Depth, pose and edges from your images steer the result.
A character record with identity, age range, body, face and a reference turnaround. Use it in any scene; its references keep the person recognizable.
Your wardrobe. Build looks from garments and assign one look per person and scene.
The location and surroundings of a scene, reusable across shots.
Where you compose: camera, lens, lights and people as movable objects. The 3D mode orbits the scene to render new angles.
Locks freeze parts of a scene, so a new take only changes what you allow.
Depth, pose, edge and segmentation maps extracted from an image. Pinned maps steer the next generation on models that support them.
The Composer is where you write, ask or edit with support from agents.
Prompt DNA is the versioned recipe behind every result. Projects group shots, versions and pinned results, so a campaign can be restored and continued later.
Post-production on finished images: move the light, make local edits by chat or mask, and upscale for print.
Multi-scene projects with versioned storyboards that hand each scene to the video model that fits it.
VisionPrompt Studio comes from a production perspective.
AI tools today are powerful. They generate strong visuals and open new possibilities. What we focus on is how these visuals can be controlled, structured and used across real workflows.
Our background is in post-production, film and creative work — environments where details like lighting, composition and consistency matter. That mindset shapes how we approach AI.
VisionPrompt Studio is built around that idea. It connects camera logic, lighting structure and prompt design into a system that makes results more consistent and easier to work with.
The focus is simple: make AI outputs not only look good, but actually work in production.
Made in Berlin with LOVE 🇩🇪
Where Post-Production Meets AI Innovation
Beyond the Frame
VisionPrompt Studio is in development and coming soon as a SaaS. The alpha is invitation-only, with no entitlement to access.