Skip to content
AspirecoStart
All systems

03App

Room Visualizer

on-device room visualiser with embeddable catalogue SDK

Customers buy what they can picture. This shows them, with the room read on their own device.

Status
Live
Period
2026
Relationship
In-house venture
Site
Private
Available as
  • Build for you
  • License
  • White-label

The pipeline

  1. 01 CapturePhoto or live camera, read on the device
  2. 02 SegmentSix surfaces isolated client-side, no server
  3. 03 ApplyProcedural materials matched to the room light
  4. 04 MeasurePerspective homography derives real area
  5. 05 CostArea becomes an installed price estimate
  6. 06 EmbedOne iframe tag; tenant catalogue auto-ingested
  • 6
    surfaces independently refinishable
  • 16
    procedural floor materials
  • 100%
    segmentation runs in the browser
  • 1
    iframe tag to embed it

The problem

Flooring and paint are bought on imagination. Sample chips and lightboxes lose the sale because nobody can picture the finished room — so the visualiser has to work from the snapshot already on the customer's phone, and it has to do that without their home being uploaded anywhere.

What we built

  1. 01

    Ran semantic segmentation entirely client-side — SegFormer through transformers.js, WebGPU where available and WASM where not — so the room is analysed in the browser, with no vision-inference tier to scale or secure.

  2. 02

    Built sixteen procedural floor materials plus wall, cabinet, counter, bath and door finishes — every surface with brightness, room-light matching and shadow controls, the five painted finishes adding sheen, and the floor adding tile scale and rotation.

  3. 03

    Derived real-world area from a perspective homography: trace one reference line of known length and the photo yields square metres, square feet and a materials cost estimate with a 10% waste factor.

  4. 04

    Added live camera mode with continuous floor detection, an A|B split comparison, and saved projects held in local storage.

  5. 05

    Packaged it as an embed SDK with white-label tenants that auto-ingest a catalogue from any Shopify, WooCommerce or plain JSON feed.

Stack

  • React
  • Vite
  • TypeScript
  • React Three Fiber
  • transformers.js
  • SegFormer
  • Zustand
  • Railway

Ideal for

Flooring, paint, tile and cabinetry sellers who lose the sale at the sample chip — and suppliers who want the tool under their own brand.

Next system

POD Engine

print-on-demand ecommerce system and automation

Scope a build like this