Leonardo

What is Leonardo AI?

Leonardo AI is an AI image generation platform founded in 2022, built on extensive custom development based on Stable Diffusion. It offers its own proprietary models (such as Leonardo Phoenix and Kino XL), supports community-uploaded models, and allows users to train their own LoRA models.

The product’s core differentiators include:

1. Game and digital asset generation
2. Character consistency (maintaining the same character’s appearance across different scenes)
3. Localized image editing (Canvas feature)
4. Custom model training (Dataset Training)

Core Features

Image Generation

Text-to-image is the foundational feature, supporting various aspect ratios and resolutions. Leonardo offers several proprietary models designed for specific scenarios:

• Leonardo Phoenix: The general-purpose flagship model with the highest overall quality and more accurate prompt understanding.
• Kino XL: Leans toward cinematic lighting and dramatic effects, suitable for movie concept art and game promotional images.
• Anime XL: Designed for anime-style illustrations.
• Leonardo Lightning XL: A fast-generation model; while slightly lower in quality, it is ideal for quickly producing large volumes of reference images.

In addition to proprietary models, Leonardo hosts a vast library of community models (similar to Civitai), covering various styles and genres.

Character Reference

This is one of Leonardo’s key features. A common issue with many AI image tools is that the same character can look drastically different in each generation, with variations in face, hairstyle, and body shape, making it difficult to maintain character continuity.

Leonardo’s Character Reference feature allows you to upload a reference image of a character. Subsequent generations will keep this character’s appearance consistent—maintaining the same face across different backgrounds, outfits, and poses. For comic artists, storyboard creators, and game character designers, this feature saves significant time.

Canvas (Canvas Editing)

Similar to Photoshop’s AI canvas capabilities, you can:

• Draw masks over existing images to have AI regenerate specific areas (local inpainting)
• Expand the image canvas to let AI fill in extended content (outpainting)
• Composite scenes across multiple images
• Manually sketch rough drafts and have AI generate refined versions based on those sketches (Sketch to Image)

The Sketch to Image feature is particularly useful for artists and illustrators. By drawing a rough line art sketch and describing the desired style, AI renders it into a polished illustration. This significantly accelerates the process from "idea to visual."

Custom Model Training (Dataset Training)

This is Leonardo’s professional-grade feature. You can upload a batch of your own images (dozens of them) to train a dedicated LoRA model. You can then use this model to generate images that maintain your specific style.

For creators with fixed styles (illustrators, brand visual designers), game companies seeking unified art styles, or commercial photography post-production teams needing consistent product image styles, this feature transforms AI into a tool that truly "understands your style."

Realtime Generation

In real-time generation mode, you type descriptions in the left text box while the right side displays the generated effect live. Every word change updates the image instantly.

This makes exploring different directions highly intuitive. Without waiting for generation times, you can rapidly iterate on prompts to find the desired style and composition. It is ideal for quickly exploring directions during the early stages of creation.

Who Should Use Leonardo?

Game Developers: Games require large volumes of consistent art assets—characters, props, environments, and UI icons. Leonardo’s character consistency features and asset generation models are specifically designed for this scenario, with a strong user base in the game art community.

Illustrators and Digital Artists: Features like Sketch to Image, Canvas local editing, and custom model training make Leonardo an auxiliary tool in the creative workflow, rather than just a "black box" that generates images for you.

Indie Game Developers: Independent developers without dedicated art staff can use Leonardo to generate game icons, NPC designs, and background scenes, significantly reducing art costs.

Creative Content Creators (YouTubers, Social Media): Those needing large volumes of stylistically consistent images—whether for cover art, illustrations, or storyboards—can rely on Leonardo’s batch generation and style consistency to maintain visual unity.

Comparison with Major Competitors

vs Midjourney: Midjourney remains stronger in terms of raw image aesthetics, particularly in artistic style and photographic texture. Leonardo’s advantages lie in its more comprehensive feature set (Canvas, character consistency, model training), free tier availability, and a direct web interface rather than reliance on Discord.

vs Stable Diffusion (Local): Local SD is more flexible and completely free, allowing you to run any model. However, it requires environment setup and model file management, presenting a high technical barrier. Leonardo offers a cloud-based SD experience, eliminating technical configuration but imposing usage limits.

vs DALL-E 3: DALL-E 3 is better suited for simple, intuitive image generation with superior prompt understanding and clearer commercial copyright terms. Leonardo excels in fine-grained control and professional features.

Pricing

Leonardo offers a relatively generous free tier, considered quite fair among AI image tools:

• Free Tier: 150 tokens per day (each image consumes between 5–20 tokens depending on settings); full access to basic features.
• Apprentice ($12/month): 8,500 tokens/month, Canvas features, higher priority processing.
• Artisan ($30/month): 25,000 tokens/month, real-time generation, high priority processing.
• Maestro ($60/month): 60,000 tokens/month, highest priority processing, commercial usage license.

The free tier is sufficient for light usage; 150 tokens per day can generate approximately 10–30 images.

Usage Tips

Take time to choose the right model. Leonardo has many models with noticeable quality differences. Browse example images from different models first, select one that matches your desired style, and then start generating.

Use Negative Prompts to exclude unwanted elements. Adding "do not want" elements to your description—such as low quality, distortions, extra fingers, or watermarks—can improve results. Common exclusion terms: worst quality, low quality, deformed, extra fingers, watermark, text.

Use Realtime Generation to explore directions, and Standard Mode for final outputs. The real-time mode is faster but slightly lower in quality. Once you find a satisfactory direction and prompt, switch to standard mode to generate high-quality versions.

Upload reference images to control style. Leonardo supports uploading images as style references (Style Reference). Generated images will lean toward the style of the reference image. This is one of the most reliable ways to control output style.

Leonardo AI is currently one of the AI image platforms with the most generous free tier and the most complete set of professional features. For game art and digital creation, its customization capabilities make it a primary tool for many professionals; for general users, its generous free tier makes it a worthwhile option to try alongside Midjourney.