# AI Interior Design Glossary: 20+ Terms Explained Simply > An A-to-Z glossary of AI interior design terms — image-to-image, diffusion, tokens, virtual staging, design intent — each defined in one plain sentence. - Canonical: https://roomredesign.co/en/blog/ai-interior-design-glossary/ - Published: 2026-07-28 · Language: en AI interior design comes with its own vocabulary — image-to-image, diffusion, tokens, design intent — and nearly all of it describes one of three things: how the AI makes a picture, what it changes in your room, or how you pay for it. This glossary defines 20+ of the terms you'll meet in these apps, each in a single plain sentence you can quote, with a little context underneath. Skim it once and every app in the category will make more sense. Terms are listed A to Z. ## The terms, A to Z ### Before/after slider **A before/after slider is an interactive control that wipes between your original photo and the AI redesign so you can compare them on the exact same frame.** It's the honest way to judge a result: drag the handle and see precisely what changed. If an app hides the "before," you can't tell whether the AI kept your room or replaced it. ### Declutter mode **Declutter mode is a one-tap edit that removes mess and clutter from your room photo while keeping the furniture and layout in place.** It shows you a tidy version of the space you already have — great for motivation or selling prep. We cover it alongside repaint and relight in [AI home makeover modes](/en/blog/ai-home-makeover-modes-declutter-repaint-relight). ### Design intent **Design intent is what you want the AI to *do* to your room, usually one of three modes: transform, preserve, or rearrange.** *Transform* restyles everything into a new look; *preserve* keeps your furniture and changes finishes like walls or flooring; *rearrange* keeps your furniture but moves it into a new layout. Knowing which intent you want is the fastest way to get a result that matches your goal instead of fighting it. ### Diffusion model **A diffusion model is the type of AI behind almost every interior design app — it turns random visual noise into a coherent image, one denoising step at a time.** Trained on millions of images, it "knows" what rooms and furniture look like and cleans up static toward your instructions. It's the engine; the app is the steering wheel. ### Digitally-altered disclosure **A digitally-altered disclosure is a note on a listing photo stating that the image was virtually staged or edited rather than photographed as-is.** Under 2026 MLS rules, virtually staged real-estate photos generally must carry this label (this is general guidance, not legal advice). The ethic is simple: never let an edited photo imply a property looks different than it really does. ### Empty room **Empty room is an edit that strips a space down to bare walls, floor, and windows by removing all the furniture.** It gives you a blank canvas for planning a fresh layout or imagining the room from scratch. It's the "restart" counterpart to declutter's "tidy up." ### Geometry preservation **Geometry preservation is an app's ability to keep your room's real structure — walls, windows, doorways, ceiling height, camera angle — unchanged while restyling everything else.** It's the single biggest quality signal in AI interior design. When it fails, you get the warping complaints — extra doors, melted furniture, impossible scale — that dominate negative reviews. ### Hex color **A hex color is a six-character code (like `#8A9A5B`) that names an exact color with no ambiguity.** It's how you test a *specific* paint shade instead of a vague "sage green." If you can find a brand color's hex value, you can apply that precise color to your wall — the core of [testing paint colors on a photo](/en/blog/test-paint-colors-on-a-photo-of-your-wall). ### Image-to-image **Image-to-image is when the AI transforms the actual photo you uploaded, using your image as the starting point it edits.** This is what keeps a redesign looking like *your* room, because your pixels anchor the result. It's the opposite of text-to-image, and choosing image-to-image tools is how you avoid "that's not my room" surprises. You can watch it work in the [iPhone and Android app](/en/download/). ### Iteration **Iteration is generating the same room multiple times to find the best result, rather than judging the first render.** Because diffusion is partly random, each attempt varies slightly — so the third or fourth try is often the keeper. Reviewers who generate once and quit are working against how the tool behaves. ### Palette picker **A palette picker is a tool that lets you choose exact colors — often by hex code — and apply them authoritatively to your redesign.** It ensures the app uses *your* colors instead of drifting toward whatever a generic prompt suggests. In Architectural AI it's how you test real paint and decor colors on a photo of your actual space before buying anything. ### Photorealism **Photorealism is how convincingly an AI result looks like a genuine photograph of a real room rather than an obvious render.** High photorealism means believable light, materials, and proportions; low photorealism looks plasticky or dreamlike. It's a big part of why one app feels usable for real decisions and another feels like a toy. ### Prompt **A prompt is the written instruction that tells the AI what to create or change.** In most consumer apps you rarely type one directly — the style presets and quick edits load expert-written prompts for you behind the scenes. When you do add custom text, a clear, specific prompt beats a vague one. ### Quick edit **A quick edit is a tightly scoped, one-tap action that changes a single thing — like wall color, flooring, or clutter — while leaving the rest of the room untouched.** That narrow scope is exactly what makes quick edits reliable for real decisions. They're the tool for "just change this one thing," as opposed to a full restyle. ### Render **A render is a single finished image the AI produces from your photo and instructions.** Each render is one output you can keep, compare, or iterate on. "Generate a render" and "generate a design" mean the same thing in most apps. ### Style preset **A style preset is a saved, expert-tuned look — like Scandinavian, Japandi, or industrial — that applies a coherent aesthetic in one tap.** Under the hood it's a carefully written prompt, so you get a polished result without art-directing in words. Browsing a deep library of [styles](/en/styles) is usually more reliable than typing your own descriptions. ### Text-to-image **Text-to-image is when the AI invents a brand-new picture from a written description alone, with no obligation to your actual room.** It's fantastic for pure inspiration but risky for redesigning your space, because the result may look nothing like your photo. Most "the AI gave me a random room" complaints trace back to a tool leaning too hard on text-to-image. ### Themed world **A themed world is a curated, pre-composed aesthetic — often cinematic or fandom-inspired — that bundles a whole mood into one selection.** Where a style is a single look, a world is a finished vibe with coordinated details. Explore the range under [worlds](/en/worlds). ### Token / credit **A token (or credit) is the unit you spend to generate one redesign, roughly mapping to the GPU compute that image costs.** Because every render runs on rented graphics hardware, tokens are how apps cover that per-image cost — which is why nothing here is truly unlimited-free. See how it's structured in [pricing](/en/pricing). ### Virtual staging **Virtual staging is digitally adding furniture and decor to a photo of an empty or sparsely furnished room, usually to help a property sell or rent.** Because the furniture isn't physically there, listings that use it generally must carry a digitally-altered disclosure. It's a marketing tool, distinct from redesigning a room you actually live in. ### Virtual try-on for rooms **Virtual try-on for rooms is previewing a style, color, or piece of furniture on a photo of your real space before you buy or commit.** It's the room-scale version of trying clothes on — see it in *your* context first, avoid expensive mistakes. It's the single most-praised use case across AI design app reviews. ## The bottom line Most AI interior design jargon collapses into a few ideas: diffusion models make the pictures, image-to-image keeps them looking like your room, quick edits and design intents control what changes, and tokens cover the compute. Keep this page handy, and the next app you try — or the next review you read — will read like plain English. For a broader tour of the category, see our [best AI interior design apps guide](/en/blog/best-ai-interior-design-apps-2026). Ready to put the vocabulary to work? [Open AI Room Redesign](https://apps.apple.com/us/app/room-design-ai-home-interior/id6762158999) and try image-to-image editing, quick edits, and the palette picker on a photo of your own room. ## FAQ **What is image-to-image in AI interior design?** Image-to-image is when the AI transforms the actual photo you uploaded rather than inventing a new room from text. It's what keeps the redesign looking like your real space, because your photo is the starting point the model edits. **What's the difference between a themed world and a style?** A style is a single design look, like Scandinavian or industrial, while a themed world is a curated, pre-composed aesthetic — often cinematic or fandom-inspired — that bundles a whole mood in one tap. Styles are building blocks; worlds are finished vibes. **What are tokens or credits?** Tokens (or credits) are the units you spend to generate a redesign, each roughly mapping to the GPU compute one image costs. Apps hand out free starter tokens and sell more because every generation has a real hardware cost behind it. **What is virtual staging?** Virtual staging is digitally adding furniture and decor to a photo of an empty or sparse room, usually to help a property sell or rent. Because the furniture isn't real, listings that use it typically must disclose that the photo is digitally altered.