Prompt X-ray
See which words
describe which part
Upload an image and get its prompt with an X-ray map: every word is linked to the zone it describes, so you can change one part and keep the rest.
What you can do with the map
See the map
Hover a zone of your image to light up the words that describe it, or hover a word to see its zone.
Change one zone
Select an element and write what it should become. The rest of the prompt, the style and the light stay the same.
Emphasize or tone down
Push one element forward or into the background, written in the weight syntax each model reads.
Remove an element
Take it out of every prompt, and add it to the negative prompt where the model supports one.
Prompt weights, written for each model
Every model has its own way to say "more of this". Prompt X-ray uses the right one.
Midjourney
chrome lettering::2, palm trees::0.5
Flux and Stable Diffusion
(chrome lettering:1.3), (palm trees:0.7)
Nano Banana and ChatGPT
The element moves to the start of the description and gets more detail
Prompt X-ray questions
What is Prompt X-ray?
It shows how your prompt maps onto your image. Each key element of the picture gets a zone, and every word that describes it is linked to that zone, in both directions.
Is this about X-ray images?
No. Prompt X-ray looks inside your prompt, not inside a body. If you want a radiograph style picture, upload one and you will get the prompt to recreate that look.
Does Prompt X-ray cost extra credits?
Seeing the map is included with every image to prompt result. Changing, emphasizing or removing a zone is a remix and uses one credit.
Why not just edit the prompt by hand?
You can, but you have to know which words to touch and how each model writes weights. Prompt X-ray finds the words for you and rewrites all model versions at once.