Zalando required images per article: 3, 2 or 1, then a batch plan

Zalando's required image count depends on the category: 3 for apparel (including kids), accessories and shoes, 2 for home, and 1 for hosiery, equipment, beauty, toys and electronics, with a catalogue view in every case. Jewellery, per its own guide, needs a primary packshot plus two further compliant images. These come from the image guidelines, updated October 1, 2026, and the category guides (read 2026-10-02). For a catalogue, that turns into a list of views per SKU, and one Sume job per view.

How many images does each category need?

Duplicates do not count, and a multipack must include a primary packshot that shows all items in the pack. Electronics and toys need one image showing the CE label on the device or packaging, so that one is a real photo of a real label, not something to generate.

Which views can I derive from one photo?

That depends on what the photo shows. A reference-guided edit can change a background or remove a mannequin, but a back view or a detail of a fastener needs to be visible in the source or photographed. Zalando's jewellery guide asks for a back view where the fasteners can be seen, and its shoes guide asks for sole and detail views as additional views. A model cannot invent a clasp it has never seen without risking a product that looks different from the real one.

So split the plan in two: views you can edit from existing photos, and views you must shoot.

How do I batch it on Sume?

Build a list of (SKU, view) pairs, send one POST /v1/images per pair with input_references pointing at the right source photo, image_size of 2000x2880, and output_format of jpeg, and give every request its own Idempotency-Key such as sku-1042-front. The docs recommend an idempotency key on every paid submit that may be retried (Generation admission).

Each call either returns 200 with the image or 202 with a job to poll. Store the job id with the SKU so a reviewer can trace each file. Prices differ per model, so read them from GET /v1/images/models and multiply by your view count before you start.

What should I do next?

Here is the short version, in the order to do it.

• List the category of each SKU and its required count.
• Mark which views exist in photos already.
• Generate only the derivable ones, one job per view.
• Review each output against the real product before upload.