How to Use Leonardo AI: A Beginner’s Guide to Images and Video (2026)

Last updated: 13 November 2026

What is Leonardo AI?

Leonardo AI started as a tool for generating game assets and concept art, then grew into a general creative platform used by marketers, designers, illustrators and content creators. Canva acquired Leonardo in 2024, and Leonardo’s technology now powers some of Canva’s AI image features, while Leonardo itself continues as a standalone platform for people who want more control.

What sets Leonardo apart is choice. Instead of one model with a single look, you can choose from Leonardo’s own foundation models, specialised models for styles like anime, community fine-tunes, and third-party image and video models, all billed through one token system.

Leonardo AI plans and tokens

Annual billing is cheaper. Token cost per image depends on the model, resolution, number of images and features used; video and premium third-party models use far more tokens than a basic image. Prices and allowances change, so confirm on Leonardo’s pricing page.

Choosing a model

If you are unsure, start with Phoenix or Lucid Origin. Try the same prompt on two or three models to see which look suits your brand.

How to create your first image

1. Sign up at leonardo.ai with Google, Apple, Microsoft or email.
2. Open Image Creation from the left menu.
3. Choose a model and preset style, such as photography, illustration, 3D render or cinematic.
4. Set the aspect ratio and number of images. Fewer images per generation saves tokens.
5. Write your prompt. Describe subject, setting, lighting, mood, style and composition.
6. Add a negative prompt if supported, listing things to avoid, such as “blurry, extra fingers, text, watermark”.
7. Generate, then compare results.
8. Refine with image-to-image, Canvas or upscaling, and download.

Writing prompts that work in Leonardo

Leonardo’s models respond best to descriptive, organised prompts. A useful structure is:

Subject + details + setting + lighting + camera or art style + mood + colour palette.

Avoid stacking many unrelated styles (“photorealistic anime oil painting”), which confuses models. Pick one clear direction.

Image-to-image and references

Image-to-image lets you upload an image and use it as a starting point. You control how strongly the result follows the original with a strength setting: low strength keeps the composition close to your upload, high strength gives the model more freedom.

Practical uses:

• Restyle a product photo, placing your real product into new backgrounds and moods.
• Turn a sketch into a finished illustration.
• Keep a character consistent across a series by using the same reference image.
• Match a brand look by using an existing campaign image as a style reference.

When placing a real product into new scenes, check that labels, shapes and colours stay accurate, because models sometimes alter small details.

Editing with Canvas

Leonardo’s Canvas is an editor for fixing and extending images:

• Inpainting: select an area, such as a hand, a sign or a background object, and regenerate only that part with a new prompt.
• Outpainting: extend the image beyond its borders, useful for turning a square image into a wide banner.
• Sketch to image: draw rough shapes and let the model turn them into a detailed scene.

Canvas is often where good images become usable ones: fixing an odd hand, removing an unwanted object, or adding space for a headline.

Making videos with Leonardo

Leonardo can animate still images with its Motion models and offers selected third-party video models within the same token system. A typical workflow is to create the perfect still image first, then animate it with a short motion prompt, such as “slow zoom in, steam rising from the cup, gentle camera drift”. Video uses many more tokens than images, so test on short clips and lower settings first. For more advanced AI video, compare with dedicated tools in our guide to the best Sora alternatives and how to use Google Flow.

Understanding the main settings

Saving tokens: practical tips

• Generate fewer images per batch while exploring, then increase once you have a direction.
• Use lower resolution for drafts and upscale only final picks.
• Turn off extras like alchemy or ultra modes when testing ideas, if your model offers them.
• Reuse good seeds or references to get closer results faster.
• Use relaxed mode on paid plans when you are not in a hurry.
• Check the token estimate shown before you generate, especially for video and third-party models.

Leonardo AI for marketing teams

Leonardo works well for teams that need a steady flow of visuals:

Teams on Canva can also use Leonardo-powered features inside Canva, which is convenient for non-designers. Leonardo itself is better when you want detailed control over models and settings.

Leonardo vs other image tools

Common mistakes

• Using public free creations for client work. Free images are public; use a paid plan for confidential projects.
• Too many style words. Pick one style and describe it well.
• Ignoring negative prompts. They reduce common artefacts like extra limbs or unwanted text.
• Not checking hands, faces and text. Zoom in before publishing.
• Burning tokens on video too early. Perfect the still image first.
• Generating real people or brands. Avoid likenesses and trademarks without permission.

A beginner’s 30-minute practice session

If you are new to Leonardo, this short exercise teaches the main features quickly:

1. Minutes 1 to 5: Generate the same simple prompt, such as “a cup of masala chai on a wooden table by a rainy window”, with Phoenix and Lucid Origin. Compare the results.
2. Minutes 5 to 10: Change one element at a time: lighting (“golden hour”), angle (“overhead shot”), then style (“watercolour illustration”). Notice what each word does.
3. Minutes 10 to 15: Add a negative prompt and see whether unwanted details disappear.
4. Minutes 15 to 20: Upload a photo from your phone and try image-to-image at low and high strength.
5. Minutes 20 to 25: Open your best image in Canvas, remove one object with inpainting and extend the image sideways with outpainting.
6. Minutes 25 to 30: Animate the final image with a simple motion prompt and download both versions.

Keep notes on what worked. Within a week of short sessions like this, most people develop a reliable personal prompting style.

Example: a festive campaign in an afternoon

As an illustrative example, imagine a small sweet shop preparing Diwali social posts. The owner writes a style line (“warm golden light, rich reds and golds, brass and marigold details, close-up food photography”), then generates hero images of kaju katli, laddoos and gift boxes with Phoenix. One image has an odd reflection, fixed in Canvas. The best image is extended with outpainting to create a wide website banner, and animated into a short clip with steam and twinkling lights for Instagram. Finally, offer text and prices are added in Canva, where Hindi and English text can be placed accurately. The result is a coherent set of visuals for feed, Stories, website and WhatsApp, created without a photoshoot, though real photos of the actual sweets are still used on product pages so customers know exactly what they will receive.

Building a consistent brand look

AI images can easily look random across a feed or website. To keep a consistent brand style in Leonardo:

• Write a brand style line you add to every prompt, such as “warm natural light, earthy tones with orange accents, minimal composition”.
• Stick to one or two models for most work.
• Use style references from approved images.
• Save successful prompts in a shared document with example outputs.
• Review images together before publishing to catch off-brand results.

Licensing, privacy and responsible use

Leonardo’s terms generally allow commercial use of generated images, but there are details to watch. Free creations are public in the community feed, third-party models may have their own terms, and you remain responsible for not infringing others’ rights. Avoid generating real people without consent, other brands’ logos or copyrighted characters. Label realistic AI images where platforms require it, and keep original photos for anything that must show your real product accurately. For image detection and watermarking, see our SynthID detector guide.

Image SEO for Leonardo images

If you use Leonardo images on your website, optimise them like any other image: descriptive file names, meaningful alt text, compression to WebP, and placement near related text. Original images can also help your pages stand out from competitors using the same stock photos. Our visual search optimization guide explains more.

Real-time generation and other handy features

Leonardo includes several tools that make experimentation faster:

• Real-time generation: results update as you type or adjust the prompt, which is a fast way to explore ideas before committing tokens to full-quality images.
• Upscaler: increase resolution and sharpness for print, large banners or detailed product shots.
• Background removal: isolate subjects for use in other designs; compare with the best background removers.
• Collections: organise images by project or client.
• Prompt history: revisit what you generated and reuse settings that worked.
• API: developers can generate images programmatically for apps and automated workflows, with separate API pricing.

Leonardo on mobile

Leonardo offers mobile apps for iOS and Android, which are useful for generating quick ideas, reviewing results and sharing images on the go. For detailed Canvas editing and managing many settings, the desktop web version is more comfortable. Your creations sync across devices.

Who should use Leonardo AI?

Leonardo is a great fit for creators, small marketing teams, game and app designers, illustrators and agencies who want control and variety. It is less ideal if you want the simplest possible experience; in that case, Canva’s AI tools or the image features inside ChatGPT and Gemini may be quicker. For text-heavy graphics, Ideogram is still the specialist.

Our take

Leonardo AI rewards curiosity. Its range of models, Canvas editing and video features make it one of the most flexible creative platforms available, and the daily free tokens are generous enough to learn properly before paying. Start with Phoenix or Lucid Origin, build a brand style line, use Canvas to fix details, and keep a human eye on accuracy and rights. Used that way, it can replace a lot of generic stock imagery with visuals that actually fit your brand.

Leonardo AI checklist for better results

Before you generate, run through this short checklist. It prevents most of the wasted tokens and disappointing results beginners experience.

• Purpose: Do you know where this image will be used and at what size and aspect ratio?
• Model: Have you chosen a model that fits the style, such as Phoenix for accurate marketing visuals or a specialist model for anime or illustration?
• Prompt structure: Does your prompt cover subject, setting, lighting, style, mood and colours in clear language?
• Negative prompt: Have you listed common artefacts to avoid, such as extra fingers, blurry details, unwanted text or watermarks?
• References: If consistency matters, have you added a reference image or style guide image?
• Batch size: Are you generating only one or two images while exploring?
• Privacy: If this is client work, are you on a plan that keeps creations private?
• Rights: Does the image avoid real people, logos and characters you do not have permission to use?
• Finishing: Will you need Canvas edits, upscaling or text added in a design tool afterwards?

After a few projects, this checklist becomes second nature, and you will spend far more of your tokens on images you actually use. Keep a small library of your best prompts and outputs so teammates can learn from what worked and match your brand style quickly.

Bottom line

Leonardo AI is one of the most flexible creative platforms available, and its daily free tokens make it easy to learn properly before paying. Start with one model, write structured prompts, use negative prompts and references, fix details in Canvas, and keep client work private on a paid plan. With a consistent brand style line and a library of proven prompts, it can replace much of the generic stock imagery on your website and social channels with visuals that genuinely fit your brand.

Frequently asked questions