Image to Stable Diffusion prompt
Stable Diffusion works best with weighted keywords and a solid negative prompt. Upload a picture and get both, with CFG and step values to paste into your favorite interface.
How your Stable Diffusion prompt is written
Positive and negative
A keyword prompt with quality tokens, and a full negative prompt for the artifacts you do not want.
Weights in SD syntax
Prompt X-ray can emphasize or tone down one element with the (word:1.3) syntax that Automatic1111 and ComfyUI read.
Batch your references
Run up to 20 images at once and export every prompt as CSV or JSON for your training or testing set.
Example Stable Diffusion prompt
A watercolor fox in a forest, described the way Stable Diffusion reads it.
watercolor illustration of a red fox sitting in a misty birch forest, loose wet brush strokes, soft paper texture, muted greens and burnt orange, gentle morning light, storybook style, highly detailed, masterpiece
How to get a Stable Diffusion prompt from an image
- 01
Upload your image
JPEG, PNG or WebP. It is resized in your browser first.
- 02
We read the image
Subject, style, lighting, lens, palette and visible text are pulled out.
- 03
Copy the Stable Diffusion prompt
The Stable Diffusion tab opens first. Copy it and paste it into Stable Diffusion.
Stable Diffusion prompt questions
Does it work with SDXL and SD 3?
Yes. The prompt and negative prompt are plain text and work with SD 1.5, SDXL, SD 3 and the fine tuned checkpoints built on them.
Do I need an extension like CLIP Interrogator?
No. PromptXray runs in your browser, so there is nothing to install, and it describes lighting, lens and palette in more detail than a caption model.
Where do I paste the CFG and steps?
In the matching fields of your interface. In Automatic1111 and Forge they sit under the prompt boxes. In ComfyUI they live on the KSampler node.