Augmented reality home decor: several items in one room

Most writing about AR for home decor covers a single object, such as a lamp placed in a room. That part works reliably. Decorating usually means combining several things, and much less has been written about what happens when a shopper places four items at once to see whether they go together.

Phone AR handles several objects at once worse than most people expect, and the reasons are physical, so better software won't fix them. Knowing where the limits are keeps you from building the wrong thing.

What single-object AR does well

When a phone places one model, it does a handful of things well. It finds the floor or a wall. It shows the model at its true size and anchors it to that spot, so you can walk around it. It also estimates the room's lighting well enough that the object doesn't look pasted on.

For most decor purchases that is enough, because the shopper has one question. Is this rug too big for the space between the sofa and the fireplace? Is this pendant the right diameter for the table? Is the mirror too large for this wall? Each of those takes about fifteen seconds to answer, and the answer is better than any photograph could give.

Where multiple objects start to break

Add a second and a third object and four problems appear.

Drift. Anchors are estimates. After a few minutes of walking around, objects slowly shift relative to the room and to each other. With one object you hardly notice. With four objects arranged together, the arrangement gradually stops matching the one you made.

Occlusion. Most phone AR can't tell that your real sofa stands in front of the virtual side table, so virtual objects are drawn on top of furniture that should hide them. With one object on open floor this rarely matters. In a furnished room with several virtual items it happens all the time, and it looks wrong.

Room mapping. To place things around a whole room, the phone needs a reasonable map of it, so the shopper has to sweep the camera slowly over floors and walls before anything useful happens. That is a lot to ask of someone browsing a product page.

Arm fatigue. Holding a phone up while arranging four virtual objects gets tiring, and dragging an object with one thumb while the camera moves is fiddly. This limit is easy to overlook because demo videos are short.

Multi-object AR still has its uses. It suits a dedicated planning tool where the shopper has time, and it is the wrong fit for a product page where they are deciding on one item.

What to build instead

Use phone AR on the product page for questions about a single product, with no setup. For questions about how items look together, use a 3D room scene that the shopper views on screen instead of through the camera.

A room scene on screen avoids all four problems. Nothing drifts, occlusion is correct, no room mapping is needed and the shopper doesn't have to hold a phone up. They can put a sofa, a rug, a floor lamp and a coffee table into a room and look at them together from any angle, using the same models that power the AR button. It looks less impressive in a demo, and shoppers get more out of it.

The last row is the setup we see working most often. Sweef lets a shopper put together a modular sofa on screen, where the layout is easy to work out, and then sends that exact configuration to the phone to check the fit. MELIMELI does the same for upholstery: the fabric is chosen on screen and AR confirms the scale. Contura uses it for stoves, where placement in the room is most of the decision and there is only ever one object.

The limits of AR lighting estimation

AR lighting estimation is good without being exact. It reads the ambient light in the room and applies an approximation of it to the model. That helps a shopper judge whether a grey sofa looks warm or cool in a north-facing living room. It is not reliable for matching a fabric shade to a piece of furniture they already own, and your marketing shouldn't suggest that it is.

Tell shoppers that AR shows the size and shape in their room reliably and gives a good idea of the colour, and that they should order a swatch if they need an exact colour match.

Where to start

Don't commission a room planner as your first project. It is the expensive option and needs a library of models before it is worth anything, so it only makes sense once you have one.

Start with models for products where one question about a single item decides the sale. In decor that means rugs, lighting, mirrors and anything that hangs on a wall. Put the AR button on those pages. Once you have thirty or forty models, a room scene on screen becomes cheap to build because the assets already exist, and that is the time to do it.

Measure it the usual way, with conversion and return rates on the modelled products compared with the month before. Shopify's published merchant data points the same way, with 94 percent higher conversion on products carrying 3D or AR content in its 2020 figures and a 5 percent return reduction at Gunner Kennels. Your own catalogue's numbers should decide the budget.

We will build your first model free, and you keep it. Send us a product, or see what the full service includes on the solutions page.

Sources

• Google, Scene Viewer documentation.
• Apple, AR Quick Look.
• Shopify, merchant data on 3D and AR content, September 2020.
• Shopify, AR shopping: Rebecca Minkoff and Gunner Kennels results.