How long does it take to generate one AI product photo?

Quick Answer

There is no defensible universal time for generating one AI product photo. For production planning, measure setup, system wait, human review and rework, and delivery, then stop the timer only when the image meets your acceptance standard and is ready to use. Workflow choice, resolution, service load, and retries can change the result, so time a representative sample of your own products.

What time should you measure for one AI product photo?

The useful number is the total time to an approved deliverable, not the provider's render time alone. A fast first output does not shorten the job if it needs correction, regeneration, upscaling, or another format before use.

Stage	What to record	When the stage ends
Setup	Selecting source photos, product details, creative direction, and output settings	The request is submitted
System wait (queue and render)	Queueing, image processing, built-in checks, and finalizing the output	The requested outputs are available or have failed
Review and rework	Inspecting product identity and image quality, selecting an output, and making any corrections or retries	An output meets the acceptance standard
Delivery	Upscaling when needed, downloading, naming, organizing, or handing off the file	The approved image is ready for its destination

Nightjar calls one submitted image request a Generation. On supported Generations, built-in visual review can retry an eligible obvious failure without charging another Credit. That internal retry belongs inside system wait; a user's later decision to regenerate or edit an output belongs in review and rework.

For repeat work, Nightjar gives you a concrete way to shorten setup: save the item’s photos and factual details as a Product, then save the photographic direction and output settings as a Recipe. Select the Product and apply the Recipe for the next shot instead of rebuilding those choices. This makes Nightjar useful when repeated preparation takes up much of the job; it does not establish a faster render time. Measure the first setup separately from later requests that reuse it.

How do single images, Photoshoots, and batches affect timing?

Single images, Photoshoots, and catalog batches are different work units, so their times should not be compared as if each were one render.

One approved image: Stop when one output passes the defined quality and delivery checks. If the request produced several candidates, include the time spent reviewing them.
Photoshoot: Nightjar's Photoshoot is a cohesive four-image output choice inside Product Photography, not a batch across several products. Measure it as one four-image set and stop when the required images in that set are approved.
Catalog batch: Measure from setup of the first product to approval of the last required deliverable. Record wall-clock time and hands-on labor separately, because overlapping system waits do not remove review work.

A single Generation is not enough evidence for batch throughput. Provider availability, workflow mix, output settings, failures, and rework can affect a larger run differently, so do not multiply one observed render time by the planned image count and present it as a forecast.

How can you measure your own end-to-end generation time?

Choose several products that represent the real workload, including any difficult materials, packaging text, or on-model shots. Keep the acceptance standard, workflow, resolution, and intended destination consistent within each comparison.

For every attempt, log the setup start, submission time, output-ready time, active review minutes, any edit or regeneration, and delivery-ready time. Report the median approved-deliverable time and a slower range separately. For batches, also report approved images per batch and the share that required rework. This produces a planning estimate grounded in the work your catalog actually requires.