Leonardo.Ai

Leonardo.Ai is an image generation platform started by an Australian team and later acquired by Canva, though it still runs as its own product. It first caught on with game artists and concept designers because it packaged, in one web app, the pieces you would otherwise assemble yourself in Stable Diffusion: model choice, reference-image control, inpainting and upscaling.

If Midjourney is “here is a beautiful image,” Leonardo is closer to “here is a set of tools you can adjust.” It has its own models plus many fine-tuned for particular styles, and you can train a custom style or character on your own images. For anyone producing assets in one consistent style over and over, that control matters more than how striking any single image is.

The trade-off is a busy interface. Models, presets and parameters are all piled together, and first-time users often don’t know where to start. Credit costs also vary by feature and setting, so it takes a while to get a feel for how far your allowance goes.

Key features

• Multi-model text-to-image: Switch between in-house and style-tuned models; the same prompt on a different model gives a very different look.
• Image guidance: Upload references to control style, character or composition, so new images carry specified elements across a series.
• Canvas editing: Inpaint, outpaint and touch up details on a canvas instead of regenerating the whole image.
• Realtime canvas: Sketch and watch the generated image update as you draw — quick for testing compositions.
• Custom training: Train a personal style or element on a set of your own images and call it in later generations.
• Upscaling and motion: Upscale results, or turn a still into a short animated clip.

How to use

1. Go to leonardo.ai and sign up with email or a third-party account; free accounts receive credits each day.
2. Open image generation, choose a model or style preset, then write an English prompt and set ratio and count.
3. To control style or character, add a reference in image guidance and adjust its strength.
4. Send the best result to the canvas for edits or upscaling, then download.

Best for

• Game and concept art: batch exploration of characters, props and environments, where style models and reference control shine.
• Asset series in a fixed style: once a custom style is trained, output stays on-brand.
• SD-level control without self-hosting: reference control and inpainting in the browser.
• Personal projects and practice: the daily free credits cover casual use.

Strengths and limitations

Strengths

• Many models and control methods, far more steerable than prompt-only products.
• Generation, editing, upscaling and training on one platform.
• Free credits that refresh, so it is cheap to try.

Limitations

• A cluttered interface with many models and settings; expect a learning curve.
• Credit costs vary by feature, and advanced options burn through allowance fast.
• Images from free accounts are public by default; private generation requires a paid plan.
• Training data differs across models; commercial rights follow the terms of service, so assess the risk before commercial use.