Case study · The Children’s Place photo studio · 2024 to 2025
AI in the photo studio
Evaluating AI-generated and 3D product imagery, and defining what production-ready had to mean before anything shipped.
The opportunity
The studio shot every style on a model, for every brand, every season. Two technologies promised to change that. Browzwear offered 3D apparel development: digital garments and virtual samples that could support earlier creative planning and reduce reliance on physical samples. FlixStock offered AI-generated on-model imagery that could expand product coverage without a separate model shoot for every style.
For the studio, the prize was capacity: faster content delivery and photographers freed for the work that needed hands-on execution.
My role
As the creative lead over the photo studio, I led the cross-functional discovery with both companies. My focus was the studio: how the technology would improve the images we already made, and which product imagery could move to generated or 3D rendering without the customer noticing anything but a better picture.

Sequins: the first fidelity test. Generated imagery tends to smear them into texture. This product is live today with an AI-generated model.

Iridescent finishes: shimmer that was never easy to capture in photography, and harder still to generate honestly.
The fidelity bar
Not convincing images. Production-ready content that tells the truth about the product.
01
→
Pick the hardest merchandise
Instead of testing on a plain tee, we tested on the product most likely to break the technology: the details a parent pays for.
02
→
Sequins, metallics, sheer
Sequins and metallic finishes have to shimmer, not smear. Sheer and see-through fabrics have to read as sheer, not solid.
03
→
Where the holes are the design
Pointelle knits, open-stitch and mesh, intentional cutouts. Generated imagery fills them in or blurs them. This is where 3D looked like the stronger route: a true digital garment keeps the structure instead of guessing at it.
04
Text and glow
Printed text graphics had to stay legible and correct. Glow-in-the-dark features had to be shown as a feature, which photography already struggled with.
Why the bar mattered
A tool that passes the easy product and fails the hard product does not save the studio anything. The hard product still needs a shoot, and now the team has two pipelines instead of one. Worse, a convincing image of the wrong garment is a returns problem and a trust problem. So the question was never “can it generate a nice picture.” It was “can it be trusted with the merchandise a parent is about to pay for.”
What we found looking at the whole supply chain
The evaluation started as a photography question and widened. A true digital garment is an asset that feeds more than the studio. It feeds the PIM system, reducing the need for physical samples at all. And it feeds the copywriters, who wrote product copy from physical samples: with a realistic rendering, they could start writing features and details before the sample ever landed. Nobody asked for that at the start. It came from looking at the content supply chain end to end instead of the studio alone.
The fidelity checklist
• Sequins and metallic finishes: shimmer, not texture
• Sheer and see-through fabrics: read as sheer
• Pointelle, open-stitch, mesh: the holes stay the design
• Intentional cutouts: preserved, not filled
• Text graphics: legible and correct
• Glow-in-the-dark: shown as a feature
• Motion features like loose glitter: video where stills fail
• One standard for photography, 3D and generated imagery alike
The outcome
We did not pick a winner. We defined what winning meant.
Where it landed
Neither vendor clearly cleared the bar during the evaluation, and the decision was still open when I left in 2025. Adobe Firefly was in the conversation then, though not the leading option at the time. What the work produced was a standard: a definition of production-ready that any vendor, model or workflow could be measured against, and a map of where a digital garment adds value beyond the studio.
What is live today
The Children’s Place now labels product imagery on its site as “AI-Generated Model,” including on sequin dresses and glow-in-the-dark pajamas: the exact categories the fidelity bar was built around. The direction became production.
Why it matters now
This is the first thing I would do with any brand adopting generative production: pick the hardest real use case, define what production-ready means in that brand’s own merchandise, and hold the tooling to it. A workflow that passes that test will scale. One that only passes the easy cases stalls the day the creative team catches it.

The shake bag: loose glitter and confetti that moves. The studio added video so a shopper could see it work, a feature stills could not carry.

Glow-in-the-dark pajamas on the live site, November 2026, labeled “AI-Generated Model.”
Let's build something that lasts.
Hire me for a senior creative or design role, or bring Rire Design in on a project or retainer.
