Before commissioning hundreds of product renders, an international marketing team should ask its supplier a surprisingly small question: which combinations does the catalogue actually offer in this market?
A 3D model might accept any shell colour, trim finish and wheel assembly. That does not make every resulting suitcase a product the brand sells. A convincing image can turn a technically possible configuration into an unintended sales promise. For brands working with Chinese manufacturers and production partners, the first deliverable should therefore be an approved selection of configurations—not the largest possible image batch.
ONCE proposes starting each production request with an identifiable catalogue record, then resolving its visual parameters. This is a workflow recommendation, not a report of a completed client implementation or measured savings.
The missing asset is the relationship between options
Adobe's current Digital twins product page describes a marketing digital twin as including geometry, materials and variant information, along with business logic defining valid combinations and constraints. That is a useful distinction: having individual components is different from knowing which combinations belong in the product range.
The description does not establish that every rendering tool enforces those constraints. Nor does it establish that a brand's ERP, inventory or market catalogue has already been integrated. Those are separate implementation questions to settle with the provider.
Google's Variant option [variant_option] documentation provides another reason to keep the relationships explicit. The optional attribute describes variant-identifying properties and is used to show variants in conversational AI and traditional Search. The guidance calls for consistent property names within an item group, unique combinations of property values, and matching landing-page information. Google's Item group ID [item_group_id] guidance also requires variant details—including images and availability—to match the submitted information.
These are documentation facts checked on 18 October 2026, not claims of a launch on that date. They do not guarantee search exposure or establish that platforms automatically detect visual inaccuracies. ONCE's inference is narrower: an image library becomes easier to govern when its configuration records agree with the catalogue and destination page.
Use three selection states before creating the batch
Consider a fictional luggage range. A champagne shell and a black trim might each be approved options, yet their combination might not be offered in the destination market. Alternatively, the factory's spreadsheet might simply be incomplete. Those situations should not produce the same decision.
ONCE's proposed Three-State Selection makes the distinction explicit. Each decision is scoped to a named market and an approved catalogue version.
| Selection state | What is known | Production response |
|---|---|---|
| Confirmed configuration | An identifiable product record confirms the combination in scope | Admit it to the production candidate list; visual approval is still required |
| Confirmed exclusion | The product owner explicitly says the combination is not offered in scope | Exclude it from sales-SKU image requests |
| Unresolved mapping | Records are incomplete, contradictory or not mapped to model components | Hold that request and resolve it with the product owner |
“Not found” is not enough to declare a configuration nonexistent. Equally, a renderer accepting the parameters is not enough to approve it. The luggage example is hypothetical, not an ONCE customer case. The cover is an AI-generated fictional chair concept illustrating material selection, not a photographed or advertised product range.
This method does not require a live database integration. If the production software cannot interpret configuration rules, the brand can supply an approved allowlist for a producer to follow manually. The automation can evolve later; responsibility for deciding which products exist cannot be left implicit.
Keep stock status out of the configuration definition
A real SKU that is temporarily unavailable is still a real configuration. Removing its approved material mapping or source asset would confuse a changing commercial condition with the definition of the product itself.
Maintain a separate use-time check: is this image appropriate for the intended market, campaign or product page now, and does the destination reflect the right variant and availability? The merchant should answer that question before use. Whether an out-of-stock image remains visible depends on the actual page, placement and applicable platform rules; this is not a blanket instruction to remove all such imagery.
There is also a legitimate place for configurations that are not sold: clearly identified design exploration. Keep that purpose separate from sales content. An attractive concept render should not silently acquire the status of an approved purchasable item when a file changes hands.
Specify traceable outputs, then check the routing
For an overseas team briefing a China-based production partner, the request should carry the product or approved configuration identifier, destination market, catalogue version, defining attributes, approved reference and intended use. The studio can then map that record to the source model, material files and product markings. It should not invent an official-looking sales SKU to fill a gap in the brand's records.
Ask for an image-to-configuration mapping alongside the pictures. The brand owns confirmation of the assortment and changes to it; the production partner owns faithful visual mapping and disclosure of uncertainty. This division makes a later market adaptation less dependent on someone remembering how an old image was assembled.
Before scaling, propose a small routing check with three records: one known-valid configuration, one explicit exclusion, and one unresolved combination. The expected responses are admission to the candidate list, exclusion from sales-image production, and a hold for clarification. This checks selection behaviour before generation. It is not an image-quality benchmark, a conversion test or proof that an automated integration works. A valid record still needs the normal product-accuracy review.
For teams planning product and advertising production with ONCE, bring those three records into the initial brief. They expose a more useful question than “How many colourways can you render?”: can the proposed workflow distinguish a product you sell from a combination that merely looks possible?
The durable asset is the approved relationship between the image and the product record, including who confirmed it and which catalogue version applied. ONCE expects that traceability to support reuse, but the maintenance burden and commercial benefit must be evaluated for each range. No cost reduction, customer outcome or industry-wide adoption rate is claimed here.
