Panelly editorial exercise — 2026-09-17

Controlled fictional editorial demonstration; starting clutter was requested deliberately; not a customer campaign or a measured success-rate benchmark.
Executed directly through the provider, not a recorded Studio session.

Run 1: perfume-start
Model: gpt-image-2-5-flare-text-to-image
Requested resolution: 1K
Original output: 768 x 1024
Elapsed: 49 seconds
Prompt:
Editorial demonstration for a fictional product, not a real brand. Photorealistic vertical 3:4 advertising image of one amber glass perfume bottle with a short black cylindrical cap and a rectangular ivory front label reading ATELIER. Bottle rests on a warm beige stone plinth. Intentionally busy set: three purple flowers in front of the bottle, a small mirror and a folded fabric behind it. Warm afternoon sunlight from the left, long shadows to the right. Preserve realistic glass and the label. No people, no other lettering or performance claims.

Run 2: perfume-revision
Model: gpt-image-2-5-flare-image-to-image
Requested resolution: 1K
Original output: 1086 x 1448
Elapsed: 79 seconds
Prompt:
Revise the supplied image as a focused editorial demonstration. Keep the single amber glass perfume bottle, short black cylindrical cap, rectangular ivory ATELIER label, original framing, beige stone plinth and warm sunlight from the left. Remove all foreground flowers, the small mirror and the folded fabric. The bottle is the only product and should be the clear focal point. Keep a visible, physically plausible contact shadow where the bottle touches the stone, with the cast shadow extending right. Keep the label unobstructed. Add no props or text. Do not redesign the bottle.

Review: foreground clutter removed; background changed to white and platform became rounded. Recognizable product is not an exact product match. Two outputs do not establish a success rate. No conversion results were measured.