Object vs. Surface AR, Explained (for Retailers)

Augmented reality on product pages is no longer a novelty budget line. The data behind it is now hard to argue with: Shopify reports products with 3D or AR content convert at roughly +94% versus those without. Wayfair's "View in Room" has driven +92% conversion, +28% AOV and -43% returns; IKEA Place has reported +189% conversion, -20% returns and 98% size accuracy. A 2025 Snap/Publicis study (n=4,028) found 80% of AR shoppers feel more confident in a purchase and 66% are less likely to return it. Independent estimates put AR-driven returns reduction at up to -40%, and the AR-commerce market at roughly $6.62B in 2024, projected toward ~$139B by 2034.

But "add AR" is not one decision. There are two fundamentally different AR experiences, they suit different products, and they carry different fidelity promises. Getting object vs surface AR right per SKU is the difference between AR that lifts conversion and AR that quietly disappoints. Here's the practical breakdown.

Object AR: put the thing in the room

Object AR takes a 3D model of your product and drops it into the shopper's actual space at true scale. They point their phone at the floor, the model anchors, and they can walk around it, view it from any angle, and check whether the sofa clears the bay window.

Critically, this is a solved problem. On iOS it runs through Apple's AR Quick Look; on Android through Google's Scene Viewer. These are native, OS-level AR frameworks — the same technology that powers Apple's and Google's own shopping demos. Tracking is stable, scale is accurate, occlusion and lighting are handled by the platform. From the shopper's side it feels premium and reliable because the heavy lifting happens in hardware and OS, not in a fragile web hack.

Object AR is the right mode for discrete, movable items with a defined footprint:

1. Furniture — sofas, tables, chairs, beds, shelving
2. Lighting — floor and table lamps, pendants
3. Decor — planters, vases, mirrors, sculptures
4. Rugs, appliances, and other freestanding goods

The shopper question object AR answers is "how big is it, and where does it go?" That question maps directly onto the two biggest problems in furniture ecommerce: fit anxiety and returns. Furniture returns alone are a roughly $30B/year problem in the US, much of it driven by "it was bigger than I thought" and "it looked different in my room." Letting someone see the piece at scale, in their space, before they buy attacks that directly.

The historical catch with object AR has been asset production. High-quality 3D models were built manually, per SKU, by specialist studios — the model behind incumbents like Cylindo and Threekit. That's accurate and beautiful, but it's slow and expensive, which is why manual 3D never scaled past hero products. The shift that makes object AR viable catalogue-wide is auto-generating the 3D model from a standard product photo — cheap enough to run across the whole catalogue instead of a curated dozen.

Surface AR: retexture the wall the shopper already has

Surface AR is a different job. Instead of adding an object, it changes a surface the shopper already has. Point the camera at a wall and it repaints it in your chosen colour; point at the floor and it retiles it; hold it up and it hangs your wallpaper pattern live in the frame.

This is the right mode for materials sold by area that cover a plane:

1. Paint and wall colour
2. Wallpaper and wall coverings
3. Tile and flooring
4. Panelling and cladding

The shopper question here is "what would this look like across my whole wall or floor?" — a question object AR literally cannot answer, because there's no discrete object to place. This is the category Roomvo built its business on, and it's genuinely valuable: colour and pattern choices are high-anxiety and high-return.

But surface AR is technically harder than object AR, and honesty about fidelity matters. Retexturing a live camera feed means detecting the wall or floor plane, respecting real-world lighting, and cleanly handling edges around trim, sockets, corners and furniture — all in real time, on the web, without a native framework doing the heavy lifting.

So here's the fidelity promise we hold to:

1. On the web, surface AR is preview-grade. Colour and pattern are accurate and true to the product — which is exactly what most material decisions turn on. Lighting response and edge precision are below native quality. The right words to a shopper are "accurate colour preview," not "photorealistic."
2. Native-grade surface AR runs through a premium iOS App Clip tier. An App Clip is a lightweight, install-free slice of a native app that launches on demand, giving surface AR access to the same native rendering that makes object AR feel so solid. This is the path to near-native surface fidelity.

Promising photorealism on the web and delivering preview-grade is how you erode trust. Promising an accurate colour preview and delivering exactly that is how you build it.

Mapping SKUs — and pricing tiers — to the two modes

The clean way to think about your catalogue:

1. Does the shopper place it, or spread it? Discrete objects with a footprint → object AR. Materials that cover a surface → surface AR. This one question resolves the vast majority of SKUs.
2. What fidelity can you truthfully promise? Object AR: native, high-quality, place-in-room, on both iOS and Android. Surface AR on web: accurate colour and pattern, preview-grade lighting. Surface AR via the App Clip tier: near-native.

The two modes also map naturally onto pricing tiers, because they represent different capability levels. Object AR and web surface AR cover the broad, catalogue-wide case at accessible cost. The native iOS App Clip surface tier is a premium step-up for retailers whose materials business justifies the highest fidelity. You buy the mode your products actually need rather than one monolithic package.

The reason TARDIS exists is that most retailers sell both kinds of product — a furniture catalogue that also lists rugs and wall art, or a flooring brand that also sells trim and fixtures — and historically that meant stitching together a furniture-3D vendor and a separate surfaces vendor, each with per-SKU manual work. TARDIS auto-generates the asset from a product photo and covers both object and surface AR from a single embed — one `