Here is the afternoon plan for a catalog of forty items. We have run it, with a clock, and the times below are honest for a brand that has its garment photos ready. If the photos are not ready, the afternoon starts with a tripod and a wall, and the plan begins tomorrow.
Before 13:00, the photos
Every item flat or on a ghost mannequin, plain background, front view, even light. The batch is only as good as the worst photo in it, and try-on cannot invent a print it cannot see. If a photo has a coloured background or a fold across the print, reshoot it now; it costs less than regenerating the look three times later.
13:00, models
Register three models from full-body photos, or pick them from the global roster. Three is enough for variety without making the catalog look like a crowd. Your own models stay private to your account. If the brand has a face, a saved character from Film Studio also works as a model, and it will look the same in the film later.
13:30, garments
Upload all forty garment photos into the wardrobe with clean labels. Labels become search terms later, and they become the names in the ledger, so "linen shirt, sand, SS" beats "IMG_4471". Assign each garment to its category while you are here; the multi-garment mode chains one piece per category and it needs to know which is which.
14:00, batch
Use the batch tool: choose the look configuration, assign garments to models, and read the estimate before running. Forty items on the standard configuration is a predictable number, and the estimate shows it with your plan multiplier already applied. Compare it with your allowance; if you are short, a pack is one-time and its credits stay valid for 365 days from purchase. Then run. The jobs queue and you can watch them complete in the strip at the bottom while you set up the shop links.
15:30, review
Scan the results in the catalog view. Look at the drape, the print, the shoulders, the hem. Regenerate the few that need a style note: "tucked in", "sleeves rolled", "front view". Most first passes are right; the ones that are wrong are usually wrong for a reason you can name in three words. Attach shop links; the catalog supports shoppable items out of the box, so a link pasted now is a link the customer clicks tonight.
16:30, films
Pick five hero items and send them to short fashion films for social. Six seconds each, vertical. Brief the film tool with the motion, not the garment; the garment is already in the image. A turn, a fabric swing, a step toward camera. If a piece needs a voice, the UGC script generator writes one you can record or synthesise.
17:00, the ledger
By five you have a catalog, a provenance ledger and five clips. Open the ledger once and look at it: every image records its source model, garment and job. That is your answer when a marketplace asks whether the photo is real, and your answer to yourself next season when you cannot remember which photo of the sand shirt was the good one. The photographer is not offended; you never had one.
What slows the afternoon down
- Bad photos. The single biggest cause of regenerations. Fix at the source.
- Too many models. Five models on forty items is two hundred images to review. Three is plenty.
- Multi-garment everywhere. Full outfits are for lookbooks. Product pages want one garment per image, and it costs one operation instead of up to five.
- Skipping the estimate. The batch tool shows the cost before running for a reason. Read it.
The next afternoon
Turn on the size advisor for the ten items with the worst return rate, enter the real measurements, and check analytics in a month. Then run the batch again for the new colourways, which need no new photos at all.
Frequently asked
Can I do this from a script instead of the app? Yes. The CLI uploads assets and runs try-ons with JSON output, so a nightly job can process a folder of garment photos into a folder of looks.
What if a try-on fails halfway through the batch? That single operation is refunded to your wallet automatically; the rest of the batch continues. Rerun the failed item afterwards.
Do regenerated images replace the originals in the ledger? No. Each generation is its own entry with its own job. You choose which image goes to the catalog; the ledger keeps both.
