Documentation
Everything InteriorAID can do — from organizing projects to AI image edits, video, 3D scenes, versioning, and sharing.
How InteriorAID is organized
InteriorAID keeps your work organized in a simple, nested structure. Understanding it makes everything else click into place.
Project
Top-level container for one job — an apartment, a house, a client.
Space (Room)
A room or area inside the project — kitchen, bedroom, office…
View
A single camera angle within a space, started from a photo, sketch, or render.
Version
Every AI result, saved and linked to its parent so you can branch and step back.
Projects
A Project is the top-level container for everything related to one job. Create one from your dashboard (the New Project button) or from the workspace sidebar.
Give it a name and an optional description. Newly created projects open automatically so you can start adding rooms right away. Your projects list is sorted by most recently opened.
Spaces (Rooms)
A Space is a room or area inside a project. Add one with the + next to “Spaces (Rooms)” in the workspace sidebar, name it, and pick a room type — living room, kitchen, bedroom, bathroom, office, dining room, hallway, outdoor, or other.
Each space keeps its own views and version history, so a single project can cover an entire home, room by room.
Views (Photos / Sketches / Renders)
A View is a single camera angle or composition within a space. Use Add view (photo / sketch / render) to upload a starting image — a real photo, a hand sketch, or a 3D render.
That uploaded image becomes the root of the view's version history. A space can hold several views (for example two angles of the same kitchen), each with its own independent set of AI results.
Image operations
Image operations are the core AI editing tools, found in the right-hand Image Actions panel of the studio. Select an image, choose an operation, set its options, and generate — each run saves a new version you can keep refining.
Restyle Room
Change the design style — modern, minimalist, industrial, Scandinavian — while preserving the room's layout, camera angle, and architecture. Generate 1–5 variants at once.
Photorealize
Turn sketches, 3D renders, or line drawings into photorealistic images. Tune the lighting and style.
Edit Area
Paint a mask over a specific region and change only that area (swap a sofa, repaint a wall) while the rest of the image stays untouched.
Furnish Empty Room
Fill an empty or unfurnished room with realistic furniture, lighting, and decor in the style you pick — the walls, windows, flooring, and camera angle stay exactly as they are.
Declutter Room
Remove clutter and mess for a clean, tidy version of the same room, or target specific items to delete. Detected objects from the image analysis appear as one-tap chips to remove.
Add / Replace Furniture
Add a new piece of furniture or swap an existing one for something else — placed to match the room's perspective, lighting, and scale while everything else stays put.
Match Moodboard
Apply the colors, materials, and atmosphere of one or more reference images to your room, keeping its structure.
Generate Variants
Create several alternative versions of the current design at once — vary materials, colors, furniture, or lighting.
Continue Editing
Keep refining the current version with a new prompt, extending the same branch step by step.
Upscale
Increase resolution while preserving detail — ideal for final exports and presentations.
Tip: operations that support variants let you generate several results in one run — pick the one you like and branch from it. See prompting tips for getting the best output.
AI Assistant
The AI Assistant lives in the right-hand panel of the studio (the AI Assistant tab). Instead of clicking through the Image Actions panel, you describe what you want in plain language and it proposes a plan— an ordered list of steps you review and run. It always works in the context of your project's brief, style guide, and current structure.
Organize & import
Attach or drop room images and ask to add them — it creates the rooms and views for you. Give it several sketches at once and it groups them into rooms.
Restyle & photorealize
Restyle a room in any style, or photorealize a sketch. With no target named, it works on the image you're currently viewing.
Options & iterations
Ask for 3 style directions, generate variants, or refine the current result — “warmer lighting, add a rug”, “try a different idea”.
Video & 3D
Turn a photorealistic image into a short walkthrough video or a navigable 3D scene.
Rooms & organization
Create, rename, or delete rooms and views, and move images between rooms — all from chat.
Brief & style guide
Set the project brief and a structured style guide (styles, palette, materials) so every result stays coherent.
How it works
- Plan & run. The assistant proposes steps with editable settings (style, output count, etc.). Run each one, or Run all. Every generative step shows its token cost and an input → output preview.
- Manual vs Automatic (toggle under the composer): Manual waits for you to run each step; Automatic runs the safe steps for you, up to a spend limit. Anything that deletes always asks first.
- Attach images. Upload, take a photo, browse your project library, paste, or drag-and-drop into the chat. Each image is analyzed; the send button waits until analysis finishes.
- It knows what you're looking at. Unnamed requests target the image on screen, and you can reference earlier results (“the last one”, “those options”) or branch from a previous version to try a fresh direction.
Example prompts
Tip: the assistant can occasionally misread a request or propose the wrong step. Review plans before running — especially anything that deletes a room, view, or version.
Video generation
Turn any photorealistic image into a short cinematic clip with the Generate Video operation (under Advanced). It works best on high-quality, photorealistic source images.
Configure:
- Motion type — Cinematic (recommended), Orbit, Walkthrough, or Reveal.
- Duration — 3s, 5s, 10s, or 15s. Longer clips take longer to process.
- Motion strength — Subtle, Moderate, or Dynamic.
- Motion instructions (optional) — describe the camera move, e.g. “slowly pan toward the window.”
Generation runs in the background; the finished video is saved as a new version you can download or share.
3D Gaussian Splat scenes
The Generate 3D Scene operation builds a navigable 3D Gaussian Splatting scene from an image or video using World Labs. The result is an interactive environment you can move through right in the browser.
Pick a reconstruction quality— higher quality produces a richer scene but costs more tokens and takes longer (from a couple of minutes up to half an hour). You can add optional scene instructions to guide what's generated beyond the frame.
Finished scenes open in the built-in 3D viewer and can be shared with a public viewer link so clients can explore the space themselves.
Versioning & branching
Every operation saves a new Version linked to the one it came from. Nothing is ever overwritten — open the View History tray at the bottom of the studio to step back through every result.
Run a new operation from an earlier version and you create a branch: a separate design direction that lives alongside the original. Branches are labeled (for example “Modern” and “Rustic”) so you can develop multiple ideas in parallel and switch between them from the branch switcher.
Variants generated together are grouped, so a single run that produced five options stays tidy in the timeline while remaining fully accessible.
Share your work with a public link — viewers don't need an account. Mark individual versions as shared, then use Copy link in the sidebar to share the whole project; the public page shows every shared result grouped by room, with full-screen viewing and downloads.
3D scenes have their own interactive viewer link. Use Preview to open the public page exactly as a viewer will see it before you send it on.
Writing good prompts
Prompts steer image, video, and 3D operations. A few habits make a big difference.
Structure beats keywords
Lead with the subject, then style, then context. Connect elements into one coherent scene rather than listing tags.
❌ Weak (keyword list)
“Kitchen, white, marble, modern, bright, clean, minimal”
✅ Strong (structured)
“Modern kitchen with white marble countertops and clean lines, bright natural lighting, minimalist aesthetic”
Describe what you want
Focus on the result, not what to avoid. Instead of “no clutter, no dark colors,” write “clean, organized living room with light colors and minimal decor.”
Match length to complexity
Simple ideas, fast iteration.
Most scenes — balanced control.
Complex scenes and technical requirements.