Integration · Salesforce Commerce Cloud

Generate on-model photos for your Salesforce Commerce Cloud catalog

Generate Product page set views with your assistant, then publish through a built B2C catalog-import adapter that maps images to products and color variations.

Get a demoSet up Uwear MCP
Model wearing a polka-dot tie-neck blouse, full-body front on the Pure White studio, generated with nano-banana-pro through the Product page set template
Front hero from the Product page set template: polka-dot tie-neck blouse on the Pure White studio, generated with nano-banana-pro through the Uwear MCP.

Use cases

What can you build?

  • Generate catalog sets by color variation

    Create front, upper-body and back white-studio photos, then map each set to the master product, color values and existing view types through catalog XML.[5][21]

  • Build virtual try-on for your storefront

    Identify the storefront version first: extend an SFRA cartridge, the PWA Kit React product page or Storefront Next components. Prototype with Uwear MCP and run shopper jobs through your backend and Uwear API.[2][3][34][35]

Model wearing the same polka-dot tie-neck blouse, full-body back camera edit on the Pure White studio, generated with nano-banana-pro through the Product page set template
Back view from the same Product page set run: a camera edit of the front hero, so the model, light, and studio carry over. Generated with nano-banana-pro through the Uwear MCP.

Assistant workflow

How do you run this from your assistant?

Proposed integration, not a preinstalled connection. Use Uwear MCP plus authorized Salesforce API or file operations and an explicit catalog-import adapter. The shopper MCP cannot publish these images.[7]

  • B2C Commerce MCP Shopper Service

    Reads products and images and writes carts. No product-media write tool is listed; shopper access does not authorize catalog imports.[7]

    Pilot with account-executive onboarding and a SLAS token with sfcc.shopper-mcpagent. The guide says Agentforce connection is not currently supported.[7]

  • Merchant Agent for B2C Commerce

    Works in Business Manager on sorting, boost-and-bury and category configuration. Its inspected tasks do not document image upload or assignment.[6][8][9]

    GA; connected org, SSO, enabled features and Business Manager permissions are required. Confirm entitlement.[6][8][9]

  • B2C DX MCP Server

    Supplies schemas, documentation and developer tools. File upload and job CLI commands are separate operations, not built-in MCP image publishing.[10][11][12][19][23][36]

    Local developer tooling with preview tools. Configure merchant API and WebDAV credentials separately.[10][11][12][19][23][36]

  1. Connect developer context and merchant access

    Connect B2C DX MCP Server and Uwear MCP. Configure authorized merchant API and WebDAV operations. The shopper MCP pilot can select online products where enabled; do not use its shopper token for catalog writes.[7][10][11][12]

  2. Read products and image groups

    Use SCAPI POST /product/products/v1/organizations/{organizationId}/product-search and GET /product/products/v1/organizations/{organizationId}/products/{productId}. Keep master IDs, color values and image assignments; imageGroups is read-only in the product API.[15][21]

  3. Approve the Product page set brief

    Load Product page set on its Pure White studio at 2:3 with the AI reviewer on. Review the costed brief in credits before generating. QA is automatic and toggleable: when on, an AI reviewer compares results with source photos and retries flagged results.

  4. Upload files and import assignments

    Discover the catalog image directory before WebDAV upload. Build XML image groups per color variation with path, alt and title; upload the XML to IMPEX/src and invoke the configured ImportCatalog job. Test merge behavior first.[19][20][21][22][38]

  5. Wait for the import and verify the storefront

    Under https://{instance}/s/-/dw/data/{version}, POST /jobs/{job_id}/executions and poll GET /jobs/{job_id}/executions/{id}. Read images through SCAPI and check the selected variation. Confirm the tenant’s staging-to-production publication process.[14][17][18][23]

Run it from your assistant

Run this photoshoot from your AI assistant with a Uwear template.

Connect the Uwear MCP server to ChatGPT, Claude, or Codex and ask for the Product page set template. It is a batch workflow that runs once per garment: front, upper-body, and back product-page shots on a white studio background, plus a 360-degree video, from one garment photo. Any Uwear account can load it by name, and the assistant shows a costed brief in credits before generation starts.

Copy this prompt into your assistant

Connect B2C DX MCP Server for developer context, the Uwear MCP server and authorized Salesforce API/file operations with an explicit catalog-import adapter. Read my B2C products and image groups and keep master product IDs, color variation values and view types; use B2C Commerce MCP Shopper Service only for supported reads if pilot access is enabled. Load "Product page set" on its Pure White studio setup at 2:3, turn the AI reviewer on, and show me the costed brief in credits before generating. After I approve, propose the image assignments, upload files to the discovered Catalogs/{catalogId} image directory over WebDAV, upload catalog XML to IMPEX/src and run the configured ImportCatalog job with reviewed merge settings. Poll OCAPI job execution, wait for readiness, read images back through SCAPI and verify the storefront and tenant publication steps before reporting.

Set up the MCP server

What one run does, per garment

  1. Generate the front hero

    One full-body front photo of the garment on the template model, in the Basic white photoshoot art direction on the Pure White studio location.

  2. Edit to the upper-body crop

    A camera edit that reframes the front result to an upper-body front frame instead of generating a new image, so the crop comes from the same session as the hero.

  3. Edit to the back view

    A second camera edit of the front result to a full-body back frame, so the product page gets the back of the garment from the same session.

  4. Render the 360 turn

    A 5-second Kling 3 Pro video from the front photo: the model turns a slow, full 360 degrees in place, garment readable throughout.

Outputs per garment

  • Full-body front photo, 2K, 2:3
  • Upper-body front photo, 2K, 2:3
  • Full-body back photo, 2K, 2:3
  • 5-second 360-degree video

The setup the template pins

Template kind
Batch workflow: runs once per garment row
System key
demo_full_pdp_test_white_studio
Art direction
Basic white photoshoot (system art direction)
Location
Pure White (system studio location)
Model
The Uwear system model by default; the assistant can swap in one of your saved models when it proposes the brief
Image model
nano-banana-pro, 2K, 2:3
Video model
Kling 3 Pro, 5 seconds
QA
Optional. With the AI reviewer on, each result is checked against the source product photo and flagged results are retried
Approval
The assistant shows a costed brief in credits before anything generates

Pure automation

How does API delivery work?

Build a catalog-import adapter. Upload images with WebDAV PUT under https://{instance}/on/demandware.servlet/webdav/Sites/Catalogs/{catalogId}/ into the discovered image directory. Upload catalog XML under Sites/Impex/src/{workingFolder}/{file}.xml. Assign image-group view types, color variation values, image path, alt and title, then run an ImportCatalog job.[19][20][21][22][38]

  1. Read eligible merchant products

    A worker reads changed product records through the authorized merchant product API and stores master, variation and result mappings. Keep image assignments separate from general product updates.[15][17]

  2. Run generation without an assistant

    Submit Uwear REST API jobs and receive completion webhooks or poll. A published Production Automation can also start when a garment is processed, a tag is added or an outfit is created. These are Uwear workspace triggers; platform delivery still needs an adapter.

  3. Run the catalog-import adapter

    Use the same WebDAV, catalog XML and ImportCatalog path, then poll job execution and read images back. Deduplicate callbacks and protect unrelated image groups; these are adapter responsibilities.[14][17][19][21][22]

Platform requirements

What image specifications apply?

Salesforce Commerce Cloud: What image specifications apply?
SpecificationRequirement or guidance
DIS dimension ceilingSalesforce Help’s indexed text states a maximum transformable size of 3,000 × 3,000 pixels. This is a transformation ceiling, not a required square ratio; the direct Help page returned a shell.[24]
DIS file thresholdImages over 6 MB are not transformed. This is not an upload rejection rule; confirm the Help guidance in the target deployment.[24]
Image-matrix sizeThe product-image JSON structure must not exceed 10 MB. This limit is not a limit on each image file.[4]
Delivery formatsCurrent responsive-image guidance recommends AVIF or WebP with JPEG or PNG fallback. Check the storefront’s delivery configuration.[25]
Image assignmentsUse the merchant’s view types and variation values with image path, localized alt and title. Test merge behavior before changing an existing color group.[21]
VideoThe catalog image-group schema does not establish video attachment. Use a separately confirmed player, DAM or content component for the 360 clip.[21]

Existing options

Which AI image tools are already available?

Documented features from the linked sources. These are not independent quality tests.

  • Cloudinary B2C Commerce Cartridge; Cloudinary for Page Designer

    A fit for media management, transformation and delivery after generation. The certified cartridges do not by themselves establish that every Cloudinary AI operation is available in Business Manager.[27]

  • Amplience Salesforce Commerce Cloud integration

    A fit for CMS/DAM content planning and publishing, with an integration certified for SiteGenesis and SFRA. Current Salesforce-specific pricing and installation or review counts were not verified.[30]

  • Amplience Image Studio

    A fit for AI product and campaign imagery within Amplience. The product material does not establish a ready-made garment-to-model action inside Salesforce.[31]

  • Amplience Studios + Salesforce Einstein Copilot preview

    A 2024 preview of custom creative actions, a proposed fit for creative updates in a commerce interface. Future-tense launch language is not evidence of a released Agentforce image feature; current packaging and price remain unverified.[32]

Frequently asked questions

Yes, through Uwear MCP and authorized API/file operations backed by a built catalog-import adapter. The shopper MCP is a pilot for product reads and cart writes; Merchant Agent does not document image uploads, and B2C DX MCP is developer tooling.[7][8][10][11][19][21][22]

SCAPI marks imageGroups read-only; the inspected OCAPI product updates do not accept image assignments. Upload files through WebDAV and assign image groups in catalog XML with the correct color variation and view type, then run ImportCatalog.[15][17][19][21][22]

No. Salesforce Help’s indexed text describes a DIS transformation ceiling of 3,000 × 3,000 pixels, not a required ratio. The direct Help body was not accessible; confirm limits and gallery crops in your deployment.[24]

Identify the storefront version first. Use SFRA cartridges, PWA Kit React components or Storefront Next extension points for an API-built try-on interface, prototyped with MCP. A 360 video needs its own player or content path; it is not an image-group entry.[2][3][21][34][35]

Uwear uses credits at $0.10 per credit; model rates are on /models. The assistant shows a costed brief before generation. QA is automatic and toggleable: when on, an AI reviewer compares results with source photos and retries flagged results.

Generate in Uwear Studio and download the stills. Use the merchant’s configured B2C image-storage and product-assignment workflow, then verify the selected variation in the storefront.[4][5]

Research

Sources and access dates

  1. [2]
  2. [3]
  3. [4]
  4. [5]
  5. [6]
  6. [7]
  7. [8]
  8. [9]
  9. [10]
  10. [11]
  11. [12]
  12. [14]
  13. [15]
  14. [17]
  15. [18]
  16. [19]
  17. [20]
  18. [21]
  19. [22]
  20. [23]
  21. [24]
  22. [25]
  23. [27]
  24. [30]
  25. [31]
  26. [32]
  27. [34]
  28. [35]
  29. [36]
  30. [38]

Next step

Connect your catalog to Uwear.

Start with an assistant-led test, then build the API delivery your catalog needs.

Get a demoSet up Uwear MCP