What does Shopify's Catalog API media filter change for beauty?

Shopify's Catalog API media filter narrows product search by the type of media attached to a product. It accepts `image`, `video` and `model_3d`; when `video` and `model_3d` are requested together, the entries combine with AND logic and only products carrying both types qualify (Shopify Changelog (2026)).

The release also expands three Catalog API tool responses. `search_catalog`, `get_product` and `lookup_catalog` can return video and 3D assets in a product's media array, which previously carried images only (Shopify Changelog (2026)). The filter applies only to `search_catalog`, while the asset URLs can appear in the other two tools even when no filter is used.

The distinction matters: this filter makes existing media discoverable as a search condition; it does not create the media. Shopify's product-media infrastructure already supports images, Shopify-hosted video, external video and 3D models (Shopify Developer Docs). The release moves those assets into responses that agents and tools can act on.

Shopify's Catalog API video and 3D filter moves the old fetch-inspect-discard loop into the search layer. Shopify says clients previously had to request products, inspect each one for usable media and discard products without it; the new filter replaces that pattern (Shopify Changelog (2026)). Because Global Catalog MCP is designed for AI agents to search and discover products across the Shopify ecosystem, that selection layer is not confined to one product page (Shopify Developer Docs).

• For a video-only experience, add `video` to `catalog.filters.media`.
• If video and a 3D model are both mandatory, include both values; only products meeting both conditions remain.
• Video and 3D records in a matched product's media array carry a `sources` array with entries for available file formats.
• The requested media can exist while the specific format your client needs does not, so format support still needs a separate check.

To receive video and 3D responses, the client must declare the draft version of `dev.shopify.catalog.global` in its agent profile. Without that draft declaration, the media field continues to return images only; Shopify says the schema will move to stable in the next UCP release (Shopify Changelog (2026)). There is no migration for existing queries, and no action is required when an application does not need video or 3D.

Shopify's Catalog API media filter lets a beauty brand feed a camera or rich-product flow only with products that carry a suitable asset. A lipstick application video, a 3D fragrance bottle or a multi-angle package asset can become a prerequisite instead of something the application discovers after selecting the product.

That is relevant to virtual try-on and camera-first discovery. Shopify explicitly frames the change for teams building AR try-on, 3D product viewers and camera-first shopping (Shopify Changelog (2026)). Google's 2025 shopping announcement separately shows U.S. users virtually trying beauty looks made from multiple makeup products (Google (2025)). It is a separate platform example demonstrating that camera-based beauty shopping is already a real product category, not only a hypothetical use case.

Do not treat `model_3d` as the same thing as face-based makeup try-on. Shopify's 3D product models support GLB and USDZ and can be used for AR on supported devices (Shopify Developer Docs). Putting foundation or lipstick on a face still requires face tracking, colour matching and a dedicated experience layer. The new filter finds a renderable asset; it does not build that layer.

A beauty catalogue should prepare for Shopify's media filter by cleaning the relationship between product, variant and use case—not by maximizing file count. The search filter proves that a media type exists on the product; it does not prove the asset depicts the right shade, package or application method.

• Write down the purpose of every asset: application technique, packaging, texture, finish or 3D inspection.
• Associate each video with the correct product and test that the returned format can play in the target client.
• Check both GLB and USDZ outputs for 3D models; Shopify serves them for the web and iOS AR respectively (Shopify Developer Docs).
• Audit the association separately when media belongs to a variant. Shopify says media must exist on the product before a variant can reference it (Shopify Developer Docs).
• Do not substitute the media-type filter for shade, availability or shipping checks; those remain separate catalogue conditions.

This preparation extends the store's underlying product and variant architecture. If those platform decisions are still open, our Shopify setup guide for cosmetics brands establishes the data foundation first. For agents searching a catalogue and operating a cart inside the storefront, our Shopify WebMCP article covers that separate mechanism.

For Shopify's Catalog API media filter, the useful next step is a media inventory followed by one small end-to-end query test. If no agent or shopping application needs video or 3D, Shopify says no action is required and existing queries continue to return the same products (Shopify Changelog (2026)).

• List products carrying video or 3D media and record which variant each asset actually represents.
• Declare the draft `dev.shopify.catalog.global` extension only for clients that need these richer media responses.
• Run three test searches: `video`, `model_3d`, and both values together.
• Open the returned media `sources` on target devices and log formats that exist but cannot be rendered.
• Evaluate the result alongside product, shade and inventory accuracy; never report “media exists” as “the experience is ready.”

The strongest use of this release is not adding a 3D model to every product in a rush. It is defining which asset a specific shopping experience genuinely requires. The filter reduces selection cost; content production, product accuracy and experience design remain the brand's work.