A few months back, I was reading an Etsy thread where a seller had been accused of using AI for her product photos. She hadn't. She'd shot every piece on her kitchen counter with her iPhone and a window. The reason buyers thought it was AI was that her photos looked too consistent. Same lighting, same crop, same background tone across forty SKUs. She'd done what good catalog photography is supposed to do, and it almost cost her her shop.
That thread shaped how I think about FrameOnce.
Here is the line we draw. There is a difference between changing how a real product looks and changing what the product is. The first is enhancement: lighting, background, color balance, framing. The second is generation: replacing the product, reshaping it, inventing details that weren't there. FrameOnce does enhancement. We don't do generation, and we won't build it.
How verification works
Every photo FrameOnce processes is registered in our verification database. This is what makes the promise above operational, not just a policy.
A processing record for every photo
When you process a photo with FrameOnce, we register a verification record: the input we received, the standard adjustments we applied, and when. That record is yours, and it stays private by default. If a buyer, a marketplace review, or a payment dispute ever questions your photos, you produce it as evidence of how your photo was actually made. Sellers who want public transparency can choose to publish a record's detail; nothing is exposed unless you decide it is.
Recognized by our checker, not AI-flagged
When anyone runs an image through our free Check Image tool, we first match it against our registry. A recognized FrameOnce photo is reported as registered and processed by us, and is never run through AI-generation guesswork. Recognition has honest limits, described below, but a recognized photo is never AI-flagged by our checker.
The technical detail, honestly described
We use a combination of an invisible perceptual marker embedded in the output image and a perceptual-fingerprint lookup against our registry. The mechanism is durable for most platform handling (Etsy, Shopify, direct sharing) but heavily cropped or aggressively re-encoded images may not match via fingerprint alone. In those cases, sellers can log in to FrameOnce and verify against their account's full catalog. Verification is registry-backed, not cryptographic. We chose this mechanism because it works without requiring platforms to display content credentials they don't yet support. And one honest boundary: the record documents what FrameOnce did to your photo. It is evidence of our processing, not a certificate about events before the photo reached us.
What FrameOnce actually does
We work with the photos you took. We adjust color, normalize lighting, standardize backgrounds, match shadows so the angle is consistent from product 1 to product 500, crop to the dimensions your marketplace requires, and convert formats for whatever platform you're listing on. We do this in batches because that's the actual problem catalog sellers have: not making one photo perfect, but making two hundred photos consistent with each other.
What we don't do is invent products from text prompts. We don't generate hands holding things you didn't actually photograph. We don't put your jewelry on a synthetic model that doesn't exist. We don't fabricate the room behind your candle. If the input isn't a photograph of something you can pack and ship, FrameOnce isn't the tool for the job.
This isn't a temporary stance while we figure out the market. It's the boundary the product is built around.
Why this matters for your shop
Etsy buyers, in particular, are getting more skeptical. There are subreddits and Facebook groups dedicated to spotting AI listings. Reviews mention it. Customer service tickets mention it. When a buyer suspects a photo was generated, they distrust everything else about the listing too: the materials description, the dimensions, even the seller's good intentions. The AI suspicion alone can sink a sale, even when nothing else was misrepresented.
If you're using FrameOnce, that question has a clean answer. The product in the photo is the product in the box. We changed the lighting and the background. We didn't change the product. You can say that to a customer with a straight face, and you can say it on a marketplace policy form, and both will be true.
You can list FrameOnce-enhanced photos on Etsy, Shopify, Amazon, and anywhere else without crossing the line that platforms are increasingly drawing around fully synthetic product imagery.
The boring stuff we don't refuse to do
Color correction. Background standardization, including swapping a messy desk background for a clean white plate. Exposure fixes for photos shot in changing daylight. Resizing and cropping for different platform requirements. Format conversion. Batch processing so you don't have to do any of the above one image at a time.
None of that crosses the line. None of it changes what the product is. Sellers have been doing all of this in Lightroom and Photoshop for decades. FrameOnce just makes it possible to do across an entire catalog without it taking a week.
The harder stuff we won't do
Generate a product from scratch. Replace a real product with a different-looking version of itself. Add a model when you didn't shoot one. Add hands when there weren't hands. Build a lifestyle scene around a product that was photographed against a wall. Invent texture or detail in places where the original photo didn't have any.
Some of these are technically possible right now. Some are getting easier every month. We're not building any of them.
Built for sellers who actually photograph their products
If you take the photos, FrameOnce makes them consistent. Join the waitlist and we'll let you know when it's ready for your real catalog.