What would make you trust an AI-edited product photo?
I help build Mazee.ai, and I’ve been testing what a background editor should show a seller before they use an image in a listing. The prompt said to change the scene while keeping the bottle and label intact. I wondered how much of that instruction would hold on its own.
I ran the same AI-generated perfume bottle and prompt three times at 2K. All four lines on the high-contrast label were still readable in every result. The bottle outline stayed close to the reference, but the glass reflections and paper grain shifted. One output also changed the edges of some letters. It was easy to miss those changes when I looked at each image by itself.
That suggests two small design ideas: give the seller a side-by-side crop at the same zoom, and keep an original-product cutout available for jobs where the packaging must match. I tested the latter by placing the original cutout over a generated scene and adding a separate contact shadow. It made the label easier to preserve, although the glass edges still needed a close look.
View the original reference, label close-ups, and three 2K results
If you build or use an image editor, which would catch more mistakes for you before export: a side-by-side comparison, a locked product layer, or a warning when the text appears to change?
p/runevr
p/reso
p/producthunt
p/google-cloud-platform
p/many-displays
p/slotify-2
p/airuncode
p/ai-context-flow
p/kodlo
p/aperture-7
p/raycast
p/lasso-11
p/dino-kids
p/kuc-kuc
p/supabase
p/jev
p/offiai
p/murror
p/dex-by-exmergo
p/htmldrag
Replies
Be the first to reply
Have a question or a thought to share? Add a comment above to start the conversation.