Bulk product photos: approve the right item before scaling the batch
Test a small catalogue group in Sowar, map each result to its product and variant, then review and verify the published Salla page before scaling up.

How do you prepare a large catalogue without mixing up images?
Start with a small group you can review completely. Match every source and result to a product identifier and variant. Expand only after checking the actual item, then verify the product page after publishing. Generated images and sale-ready images are different counts.
This workflow uses Sowar alongside your existing review record. It does not mean the app automatically manages SKU filenames or your team’s approvals. A simple worksheet is enough to begin; you do not need another catalogue system.
Group products by what the buyer needs to inspect
Two mugs may suit the same background, but each mug’s colour, size and handle still need to match its reference. Separate products with different review needs: a food label, a device’s ports and a fabric pattern are not interchangeable details.
Choose a scene and aspect ratio for each group’s intended placement. Test a representative item first and inspect its complete shape, edges and details. A scene that works for a mug may not work for a transparent jar or reflective device.
| Product and variant | Review decision | Next step |
|---|---|---|
| MUG-CREAM / cream | Accepted: colour and handle match | Match the approved file to this variant’s page |
| JAR-HONEY / one jar | Rejected: label text changed | Keep the original and revise the scene before regenerating |
| HUB-USB / exact model | Needs review: a port is hidden | Use a real detail photo that shows the ports |
Record the mapping before generating
Keep the product identifier, variant, source filename and intended use together. Add the result filename and review status when it is ready. Do not use download order alone to identify products. If a ZIP contains several images, open it and verify every file before handing it over or publishing.
Use the batch size and generation settings available in your account, and check the displayed cost before starting. There is no universal ideal batch size: begin with what you can inspect, then expand when the results are usable. This is a work-planning choice, not an extra Sowar plan entitlement.
Open the per-image review checklist
- Is this the intended product and variant, not a similar item?
- Do the colour, shape, label and item count match the source?
- Has the scene added an accessory a buyer could mistake for an included item?
- Are important details visible rather than cropped or hidden?
- Does a clear filename and review record connect this result to the correct product?
Fix the cause, not the whole batch
Mark each result accepted, needs correction or unsuitable for this use. If a label changes or a detail disappears, do not repeat the same brief across the entire catalogue. Review the source image and scene, then test one change on the affected item. Check the regeneration cost first; do not assume retries are free.
A real source photo may be better for a detail the generated scene cannot preserve. Use it to explain that detail rather than publishing an attractive but inaccurate image. Keep accepted results instead of recreating them without a reason.
Decide whether to expand from your reviewed sample
Keep three counts: results generated, images accepted and product pages verified as updated. Add the actual attempt cost and review time from your own trial. Image volume alone does not establish cost savings or higher sales.
Start with one product in Sowar and compare the usable result with the effort and cost before expanding. Follow the Salla quick-start guide for publishing, or the approved-image handoff guide when another designer will finish the work.