Skip to content
← Back to News
Launch · Retrace · 3 min read

Retrace: turn any image into a prompt.

Upload a picture and Retrace writes the prompt behind it, in the format of the generator you use. It takes about ten seconds.

Retrace, the live site
Retrace on retrace.qol.one

You see an image you like. You want to make something in that style. The hard part is putting it into words: the light, the lens, the colours, the mood. That is slow to do by hand.

Retrace does it for you. You upload a picture. A vision model reads it and Retrace writes the prompt.

How it works

Your browser resizes the image to 1280 pixels on the long edge and sends that copy. The original never leaves your device. The whole thing takes about ten seconds.

Every image hides a prompt.

Eight formats

One reading, eight formats: GPT Image, Nano Banana, FLUX.2, Midjourney, Stable Diffusion, Ideogram, Recraft and Veo. If you move from one generator to another, you do not start over. Every field is visible, so you can change the light or the palette and keep the rest.

Who it is for

Anyone who works from reference images and writes prompts for an image or video model. Teams turn approved photos into written prompts for client mood boards. Photographers document the look of their own work. Others study how an image they admire breaks down into light, lens, palette and composition, or describe one hero shot for product imagery.

At launch price

Free accounts get 15 analyses a day. Pro is $4.67 a month billed quarterly ($14), a launch price. Monthly is $6 and yearly is $48. Pro gets 40 analyses a day, in all 8 formats.

Every analysis runs a vision model, so each plan has a daily limit. Limits reset at midnight UTC. Pro renews until you cancel, and you can cancel any time from Manage plan. Payments go through Stripe.

The takeawayDrop an image, let Retrace read it, copy the prompt. About ten seconds. Details are on the Retrace page on QOL.

Bring an image.

Get the prompt behind it in about ten seconds.