MyBed: Every Variant, Live on the Page.

Every bed, every fabric, every headboard height: the full catalog matrix, prerendered and served straight to the product page.

The Challenge

MyBed sells upholstered beds for a demanding buyer: dozens of fabric collections, multiple sizes, and a range of headboard heights. Multiply those together and the catalog isn’t a few hundred products, it’s a matrix in the millions of distinct looks.

A camera can’t reach that. Photographing every fabric on every bed in every size and headboard height would cost millions and be out of date the moment a supplier changed a weave. So, like most furniture stores, MyBed showed a handful of photos per bed and let the buyer imagine the rest, on product pages that paid traffic was being sent to

The Turning Point

The first scope put the job at 480,000 visuals. Then the headboards got counted properly: nine height variants per model. The render matrix recomputed to roughly 1.44 million on the spot, and a full-coverage build across every bed, size and collection ran toward numbers no photography budget could touch.

That recompute is the whole point. The problem was never “render some nice images.” It was that the real catalog is far larger than anyone budgets for, and only a system that builds the matrix automatically can actually cover it.

The Solution

Ar-range built the beds as parametric 3D models and put a generation pipeline behind them, not a photoshoot and not a 3D engine bolted onto the product page.

Parametric models mattered for the cost. Adding nine headboard heights added compute, not modeling work: the headboard is a parameter, not nine separate models to build by hand. MyBed chose full automation, so the system generates the entire matrix from the models and the catalog data, rather than someone rendering images one at a time.

Every visual runs on Ar-range Content Hub, the single catalog spine. The same source of truth that holds the fabrics and the beds feeds the visuals, and later fed the configurator and the AI fabric search built in the same engagement. The store calls an image by its product and option IDs; the right render comes back, page-speed fast, with no render engine sitting on the product page.

The Process

1. Parametric models. Each bed and headboard built as a parametric model, so size and headboard height become parameters, not new modeling jobs.
2. The coverage matrix. Fabric × size × headboard height computed as a grid, so “every variant” is a checked claim before a single render runs.
3. The automation pipeline. A generation system renders the matrix from the models and the Content Hub catalog data, with consistent light, angle and staging across the whole catalog.
4. Delivery. Renders served as static images, addressed by the store’s own product and option IDs, so the live product page swaps the full image set instantly and never waits on a render engine.

Results

• ~1 million product visuals delivered, the full sellable matrix of fabric, size and headboard height.
• Every combination a buyer can pick is live on the product page, so paid traffic lands on a page that shows the exact bed, not “shown in grey, other fabrics available.”
• A new fabric line goes live as a re-render, not a photoshoot cycle.
• One catalog spine, Content Hub, feeds the visuals, the configurator and the AI fabric search, so nothing drifts between channels.