3D for e-commerce5 min read
3D SaaS vs agency vs in house for product models
What an agency project, an in-house 3D team and a subscription service cost, where each one falls short, and which kind of catalogue each one fits.
By CharpstAR · 21 Mar 2023 · Updated 9 Oct 2026
Part of 3D models for e-commerce: the complete guide
There are three ways to get 3D models of your products. You can commission an agency, hire your own artists or subscribe to a service. They differ less in price per model than in what happens after delivery.
An agency project
You brief a studio, they quote per model or per batch and deliver the files. You pay, and the project closes.
You get exactly what you commission and have no commitment beyond the invoice. For a launch, a campaign or a single hero product that has to look exceptional, an agency is often the right choice. Good studios have people who will spend days on the stitching of one cushion.
The weak point is everything that happens later. Revisions have to be negotiated. Every new fabric, finish or product means a new scope and a new purchase order, and after a few rounds the procurement work costs more than the modelling. Six months on you may need a small change to a file and find that the person who built it has moved on. The spend is also capital expenditure, so in the budget round it competes with things like new warehouse racking.
An agency fits a catalogue of fewer than about thirty products, a stable range, or a one-off campaign.
An in-house 3D team
You hire one or more 3D artists, buy the software and build a pipeline.
Changes are quick, since a product manager can walk over and ask for one. Your artist learns your materials, your lighting and the quirks of your range, and after a year the models come faster and look more consistent. If 3D feeds your product development as well as your marketing, this is usually the option that pays off.
Before any of that you have to hire. A single artist takes holidays and will leave one day, and the knowledge of the pipeline leaves with them. Two artists are a significant salary cost, plus software and hardware that can render, plus management time to keep a small specialist team busy. Catalogue work arrives in bursts around product launches, so an artist who is fully loaded in March is underused in July. Scope also grows. Once there's a 3D artist in the building, everyone has a request, and product page models end up queued behind the trade fair animation.
An in-house team fits large catalogues, continuous new product development, or a business where 3D goes into design and manufacturing too.
A subscription service
You pay a monthly fee per product or per model, and the provider builds, maintains, hosts and updates the assets.
The cost is operating expenditure, so it goes through a different approval and can be signed off at a lower level. Maintenance is included. Platform requirements change, formats get stricter and a fabric supplier changes a weave, and the provider keeps the files current. Capacity follows your launch calendar without a change in headcount.
You commit to a term. Our own Basic plan starts at 10 dollars per product per month on a 12-month term with a minimum of 100 variants, and Value starts at 20 dollars per model per month on a 24-month term, with Enterprise quoted. Over a long enough period a subscription costs more than buying the same models once; the fee pays for maintenance, hosting and the freedom to change things later. A catalogue that never changes doesn't need any of that, and buying once is cheaper.
You also depend on the provider. Ask for the source files and the delivery formats in writing before you sign, so that you can leave with your models.
A subscription fits a hundred products or more, ongoing variant churn, and a team that does commerce and has no wish to run a studio.
The three options side by side
| Agency project | In-house team | Subscription | |
|---|---|---|---|
| Cost shape | Capital, per project | Salaries plus tools, fixed | Operating, per product per month |
| Approval path | Business case | Headcount approval | Departmental budget |
| Speed for the first ten models | Fast | Slow, hiring first | Fast |
| Cost of a small change in month seven | New scope | Same day | Included |
| Format and platform maintenance | Yours | Yours | Provider's |
| Capacity when launches cluster | Buy more | Bottleneck | Flexes |
| Capacity when nothing launches | Pay nothing | Paying anyway | Paying anyway |
| Quality ceiling | Highest | High and rising | High and consistent |
| Risk | Scope creep | Key-person dependency | Provider dependency |
Maintenance after delivery
A product model is never finished, because the platforms keep changing their requirements. Shopify documents GLB and USDZ as the supported formats, automatic optimisation above 15 megabytes and roughly 4 megabytes as a sensible target. Google's Scene Viewer takes glTF 2.0 or GLB and lists a recommended ceiling of 100,000 triangles and 10 materials. Apple's AR Quick Look takes USDZ, and Apple has added material variants and an HTML `model` element in Safari on visionOS 26.
Someone has to keep up with each of those: an agency charges for the update, an in-house artist fits it between other jobs, and a subscription includes it.
How to choose between them
Count your products and your variants, and estimate how many will change in the next two years. A small catalogue where very little changes points to an agency. Hiring makes sense when the range changes constantly and 3D also feeds product development. A large catalogue with steady variant churn, run by a commerce team with no appetite for a studio, is where a subscription fits.
Before you commit to any of the three, have one model made and look at the result on your own page. Send us one product and we'll build it free of charge, and the file stays with you whatever you decide.
Sources
- Shopify, 3D e-commerce.
- Google, Scene Viewer documentation.
- Apple, AR Quick Look.
