Stable Diffusion detector
Stable Diffusion is the most self-documenting generator in common use. Automatic1111, ComfyUI and most other front-ends write the full prompt, negative prompt, sampler, CFG scale, seed and model name straight into the PNG as a text chunk, and almost nobody removes it.
When that block is present the answer is not really a detection problem at all — the file is telling you how it was made. This tool finds and reports it. Failing that, it looks for the standard SDXL output dimensions and runs the classifiers.
Drag a file here, or
JPEG, PNG, WebP, AVIF, PDF or .txt · up to 10 MB · you can also paste a screenshot
By uploading you agree to our Terms and Privacy Policy. Your file, IP address, approximate location and device details are stored.
What this checks
- PNG tEXt, zTXt and iTXt chunks, including compressed ones
- The 'parameters' block containing prompt, sampler, CFG scale and seed
- ComfyUI workflow JSON embedded in the same way
- Standard SD and SDXL output dimensions such as 512×512, 1024×1024 and 1216×832
- Image classifiers as a fallback for stripped files
Frequently asked questions
Where does Stable Diffusion store the prompt?
In a PNG text chunk, usually keyed 'parameters'. It is plain text sitting inside the file, which is why saving as JPEG or taking a screenshot destroys it — JPEG has no equivalent chunk.
Will this show me the original prompt?
When the chunk survives, the result quotes what it found, including the sampler and seed. That is often enough to reproduce the image.
What if the image was saved as a JPEG?
The generation block is lost, and the check falls back to dimensions and classifiers. Expect a much wider confidence range.