You're using a guest profile, so some things are limited:
  • Your jobs are deleted 1 day after they finish
  • Up to 25 steps, 1 image per job, 1024×1024 max size, no LoRAs
  • 10 jobs an hour, 30 a day
  • No Flux, Qwen Image, Flux Kontext, Qwen Edit, Relight, Image to 3D
  • No save prompts, save images, share links, zip downloads, the API
Sign up free
Negative prompt
LoRAs
Start image & reference

Add the images on the right. Start image: img2img, the result keeps its layout and colors. Reference: IP-Adapter, the result follows its style and subject.

Person — keep someone's face and body

Pick one of your People: the result keeps their face (Face ID), and with Body strength above 0 also their build, outfit and hair from the photos. Use it instead of a Reference image, with a prompt that shows the face.

ControlNet — follow an outline, pose or drawing

Add a control image on the right, or draw one. The result keeps its edges, depth, pose or lines; the prompt decides the rest.

Hires fix

About twice as slow. The upscaled size counts against your size limit.

Detailer — fix faces & hands

Finds each face / hand and redraws it at full size. To fix an existing image: put it in Start image and set Repaint (start) to 0.

Batch

Prompt variants: write {a|b|c} in the prompt and each option becomes its own job. a {cat|dog} in {snow|rain} = 4 jobs.

Compare models

The same settings on each ticked model. LoRAs only go to models of the picked model's family.

Compare schedulers

The same job on each ticked scheduler. Karras carries over where it applies.

See a batch side by side as a grid: Batch viewer.

Text to Image

Open History

Start image optional

Start image

Reference optional

Reference image

Control optional

Control image

About Text to Image

Type a prompt and generate images with Stable Diffusion 1.5 and SDXL models, including Pony and Illustrious checkpoints. Add LoRAs, start from an image (img2img), follow a reference image, keep a person's face, guide the layout with ControlNet, and upscale with the hires fix.

Is it free?

Yes. This is a free, non-commercial project. Many tools work without an account; a free account unlocks the rest and higher limits.

What happens to my images?

Nothing is shared unless you make a share link. Without an account your results are deleted a day after they finish; with an account they stay in your History until you delete them.

Can I use it from my own code?

Yes. With an account you can create an API key and use the REST API or the MCP endpoint.

Before you start

This site makes AI images, edits and audio, and some models can make adult content. To use it you must be 18 or older and agree to the Terms of Service (and our Privacy Policy). In short:

Leave