How to Create Product Photos for a Marketplace: Step-by-Step Workflow

Use even light: a window with diffusion, softbox, or two side sources to avoid deep shadows under the product. The item should fill most of the frame with a readable silhouette and no clipped details. Remove unrelated objects, other-brand packaging, and watermarks that trigger moderation issues. The cleaner the source, the less AI invents shape, color, and texture.

For clothing, lay the garment flat or hang it without folds that hide pockets or prints. For shoes, photograph the pair with matching angle so buyers see sole and profile. Save multiple angles—even if only one AI frame becomes the main image, others anchor QA. Never use supplier catalog images without commercial usage rights.

For Kaspi and other channels, tie the card to real warehouse inventory. Store originals in an SKU folder—evidence when buyers dispute color or contents. Open current platform image help before upload; requirements update without a separate notice to Studio.

Main listing images usually need exact product card or white/neutral background modes. Social scenes can be more creative, but the product must not change: same color, print, and bundle. Apparel catalogs sometimes benefit from on-model frames—use a separate mode and review fit carefully. Video and Reels from photo are in development; do not publish video if the feature is unavailable in your version.

Generate at least two variants—do not pick the first preview without comparing to the original. Complex categories—jewelry, glass, small electronics—need extra review time. Exact product card mode is usually more conservative: less creative drift, better detail retention.

Define who may send a frame to the cabinet after QA so rush errors do not leak into the catalog. When switching marketplaces, re-check file format and aspect ratio. Keep the source beside the AI export for dispute resolution.

Compare shape, color, pattern, seams, hardware, logos, stones, labels, and bundle contents to the source. Open images on a monitor or tablet—phone screens hide distortion. If a detail drifts, reject the variant; regeneration is cheaper than a return and negative review. AI may remove scratches or add shine—that is still misleading if reality differs.

Inspect cutout edges: straps, tassels, and transparent parts fail often. Shadows should look natural and must not imply a different size or hidden stand. For sets, confirm the photo matches description and warehouse inventory.

Archive a before/after screenshot per SKU. During bulk upload, spot-check every fifth SKU for systematic model errors. When in doubt, delay upload rather than risk a complaint.

Open current Kaspi, Wildberries, or Ozon help for images, video, and prohibited content. Rules change—what passed last season may fail today. Vitrina AI does not guarantee moderation and is not an official partner. Final decisions and liability stay with the seller.

Verify minimum resolution, aspect ratio, text bans on main images, and background requirements. Apparel and intimates have extra rules for models and adult-only content—read category notes. If a card is rejected, compare the reason to your checklist—often the issue is product mismatch, not AI itself.

Log rejections: SKU, reason, date, fix owner—teams learn faster. Do not copy visuals from another marketplace without checking local rules. When unsure, choose a more neutral background and simple composition.

Publishing the first variant without source comparison is the top cause of mismatch returns. Aggressive lifestyle backgrounds hide size and material cues. Third-party photos and logos risk account-level issues, not only one rejection. Ignoring differences between main image and gallery angles confuses buyers.

Poor lighting forces AI to invent details that are not on the product. Bulk generation without spot checks spreads one systematic error across the catalog. Listing copy promises items not visible in frames—a separate complaint category.

Schedule review time like copywriting and pricing. One hour of QA beats a cluster of returns on a hot SKU. Train the marketplace manager on the checklist, not only the designer. Refresh platform image help before each major upload batch.

Build a pack: marketplace main, detail, Instagram lifestyle, and 9:16 when needed. All frames must show one SKU, one color, one bundle. AI pays off as a pack, not a single file. Align style with ads so banners and cards stay consistent.

For Kaspi start with an honest product shot; for Wildberries unify catalog style; for Ozon add extra angles. Keep a high-resolution master; upload per-channel exports. When scaling assortment, copy workflow and checklists—not lucky settings without per-SKU verification.

Monthly, review top-return SKUs—visual issues often beat price as the driver. Update photos when suppliers or batches change color and hardware. Document which source matches which warehouse batch.