One job state per product
Selected, submitting, queued, generating, ready, and failed states remain visible for each file. A failed model request does not pretend the full batch succeeded.
A visible queue instead of an opaque bulk button
A Batch Product Image Generator You Can Review SKU by SKU
Batch generation becomes useful only when failures stay isolated and filenames survive the trip. Product Ad Pack processes each product as a separate job, records its source name, and lets you approve or reject results without restarting the whole batch. Paid credits make cost visible per completed product pack.
- No card for the first preview
- Editable text, not AI lettering
- Human product review required
1:1Feed
4:5Portrait
9:16Story
16:9Wide
Free daily preview · No card required
Build a reviewable product-image batch
Choose up to 20 product photos. Each file becomes an isolated job with its own status, preview choices, retry, and commercial export.
Per-product queue
Review every SKU independently
A failure or rejection affects only that product. Ready directions can be opened below without restarting completed jobs.
Preservation review
Compare the source with each direction
Reject any preview that changes packaging, shape, count, color, controls, material, or important label details.
Four placements
Review the complete ad pack
Change the headline or brand color above and the canvases update without another AI generation.
Free downloads are for evaluation. A paid credit is required for commercial ZIP delivery.
What the workflow controls
From source photo to reviewed campaign family
Stable SKU-oriented filenames
The name you provide becomes the base for every ratio, making downstream upload, review, and handoff less error-prone.
Retry only the failed item
System failures preserve the job record and paid-credit refund path. You can retry the affected product instead of spending again on successful results.
A short, reviewable flow
How it works
- 1
Prepare clear source images
Use one product per file and name the files with the SKU or product handle before upload. Consistent inputs reduce manual cleanup later.
- 2
Apply one campaign brief
Set the shared scene, brand color, and campaign headline, then adjust an individual product only when it genuinely needs a different treatment.
- 3
Review the queue and export
Open products individually, retry isolated failures, and export a commercial per-SKU ZIP only after reviewing the selected direction.
Clear limits before checkout
Frequently asked questions
How many products can I process in one batch?
The MVP limits a browser batch to 20 source files so every result remains reviewable. Available paid credits still control how many new model jobs can be submitted.
Does a failed batch item consume a paid credit?
A model job reserves a credit. A confirmed system failure releases that reservation through the ledger. A completed but creatively rejected image is not treated as a system failure.
Can I upload a CSV?
The initial workflow uses source filenames and a shared campaign brief. CSV mapping is introduced only when real users need different copy per SKU; it is not required to get the first batch result.
Try the real workflow
Prove one product first, then scale the queue
Use the free preview to validate preservation before buying batch credits.
Create a free pack