How many products can one bulk run handle?

Up to 1,000 products per run from a single CSV or XLSX file. Larger catalogs split naturally into multiple runs — which is good practice anyway, since running one category at a time lets each batch carry its own model and pose configuration.

What happens when a generation fails mid-run?

Bulk Studio retries failed generations automatically, and credits for failures are refunded automatically, so you only pay for delivered images. For a faulty image that renders but misrepresents the garment, the credit-refund policy covers it within about 24 hours.

Which plan includes Bulk Studio?

The Business plan at $60/month (~SAR 225), which carries 3,000 credits (~100 images) per month, up to 8 poses per garment, 4K output, a priority queue, and 30-day credit rollover. Smaller stores generating product by product are usually better served by Pro.

Do I have to review every generated image?

No — sample roughly 1 in 10 images per category and zoom into the fidelity basics: necklines, buttons, print alignment, logos, and colorway accuracy. If drift concentrates in one category, re-run that slice with stronger detail references instead of fixing images one by one.

How do I keep 500 products looking like one brand?

Configure the model and pose set once per category and let the run apply it uniformly — that is the consistency advantage bulk has over months of hand-configured sessions. The check is the grid test: screenshot your category page and ask whether it reads like a single shoot.

From CSV to Full Catalog: Bulk AI Photography for Large Stores

How do you generate product photos for an entire catalog at once?

With a bulk catalog mode instead of one-by-one uploads. In Rokon's Bulk Studio (a Business-plan feature), you upload a CSV or XLSX file of your catalog and the system generates on-model imagery for up to 1,000 products in a single run, with automatic retries and automatic credit refunds when a generation fails. Generation itself is the easy part — this guide covers what actually determines whether a bulk run succeeds: source-photo prep, file structure, credit budgeting, and QA sampling.

Scale is a different problem than quality

One good image is a solved problem: upload a flat lay, configure a model, wait about 30 seconds. At 500 products, the problem changes shape:

Queueing. One-at-a-time means hundreds of upload-configure-download cycles — weeks of clicking that nobody budgets for.
Failures. At catalog scale, some generations will fail or come out unusable. Without automation, failure handling becomes a spreadsheet of its own.
Consistency. Products configured by hand across many sessions drift: a different model here, a different pose set there, and the category grid stops looking like one brand.
Credit tracking. Without a budget set before the run, you discover mid-catalog that the credits ran out.

Bulk tooling exists to industrialize these four, not to make any single image better.

For context on why this is the problem worth solving: covering a 500-product catalog with a freelance photographer at $300–$500 per product prices out at $150,000–$250,000 plus months of lead time — which is why large catalogs historically stayed half-photographed. Generated imagery on the Business plan works out to about $0.60 per standard image ($60 for ~100 images), so the constraint stops being money and becomes operations. Operations is exactly what a bulk workflow manages.

What Bulk Studio actually does

Takes your catalog as a CSV or XLSX file.
Processes up to 1,000 products per run.
Retries failed generations automatically.
Refunds credits automatically on failures — you do not pay for images that never arrived.
Ships with the Business plan, which also carries the priority queue, up to 8 poses per garment, and 4K output.

That is the whole feature surface, and it maps one-to-one onto the four scale problems above.

The run is only as good as the source photos

A bulk run amplifies whatever you feed it — including sloppiness. Before touching a spreadsheet, standardize capture: flat lays shot in daylight against the same plain background, a front view plus a detail shot for every product, and clean per-SKU file naming so each row maps to its images without archaeology. Multi-image reference (front, back, detail, logo shots) is what preserves each actual garment — stitching, prints, silhouette — while only the model and pose are generated, so the detail shots are not optional at scale; they are your fidelity insurance.

Structure the catalog file by category

Keep the file boring and predictable: one row per product, mapped to its source photos. The decision that matters is one category per run. A model configuration and pose set that flatters abayas — full-length, movement poses — is wrong for knitwear close-crops, so running abayas, thobes, and knitwear as separate batches lets each run carry one correct configuration instead of a compromise. Keep colorway variants adjacent in the file so you can spot-check a color run as a group.

The credit math before you press start

A standard image costs 30 credits; a 4K image costs 60. The Business plan grants 3,000 credits per month (~100 images) with a 30-day rollover.

Run size	Images (1 per product)	Credits at standard	Credits at 4K
100 products	100	3,000	6,000
500 products	500	15,000	30,000
1,000 products	1,000	30,000	60,000

Poses multiply the total: 3 poses per product triples the credit line. So a full-catalog project is a budgeting exercise — monthly grant plus rollover, topped up for heavy months — and often a sequencing exercise: many stores roll the catalog out over two or three monthly cycles rather than one push. The refund mechanics help here: failed generations return their credits automatically, so the budget only pays for delivered images.

QA by sampling, not exhaustive review

Reviewing 1,000 images one by one recreates the labor the bulk run just removed. Sample instead: inspect roughly 1 in 10 images per category, zooming into the fidelity basics — neckline shape, button count, print alignment, logo integrity, colorway accuracy against the real piece. If the sample shows drift concentrated in one category, re-run that slice with stronger detail references rather than hand-fixing outputs. For a faulty image that slipped through, Rokon's credit-refund policy covers it within about 24 hours.

Roll out in sequence, not alphabetically

Your top 20% of SKUs produce most of the revenue, so they earn imagery first — publish best-sellers, then work category by category down the long tail. Time runs against the calendar: imagery for White Friday needs to land in October, and a Ramadan refresh needs the catalog finished before the season starts, not during it. And keep one model configuration per category across runs so the storefront grid reads as one brand, not a bazaar.

Where bulk honestly is not the answer

Bulk does not remove the pre-publish check — it concentrates it into sampling. Heavily embellished pieces — sequins, dense embroidery, metallic threadwork — deserve individual runs with extra detail references, and your true hero pieces may still justify a physical shoot. And Bulk Studio is deliberately a Business-plan tool: a 30-product boutique gets to a finished catalog faster by generating one product at a time on Pro. Bulk earns its keep from a few hundred SKUs upward — stores and agencies operating at catalog scale.

A bulk pipeline starts with a single good source photo. Test the fidelity on your fastest-moving product first — 5 free generations (150 credits), no credit card — and scale the workflow only after your own zoom test passes.