leonardo-ai capabilities

Leonardo.Ai's Canvas Editor is a browser-based image editor that supports inpainting (masking and regenerating specific regions) and outpainting (extending the canvas beyond original image boundaries). It is categorised as a legacy feature in current documentation.

All Leonardo.Ai users (free and paid) may use generated images commercially. The key distinction is ownership and privacy: paid subscribers retain full copyright to all content they generate, with the option to keep generations private; free users' generations are public and other users receive a non-exclusive license to reuse them.

Elements are user-trained Low-Rank Adaptations (LoRAs) that encode a specific visual style, character, or product for consistent reuse across generations. Flux Dev is the current default base model for new Elements; legacy fine-tuned models (SD 1.5/2.1) are deprecated.

Leonardo.Ai exposes 13 ControlNet-based image guidance types that steer generation using a reference image. Supported modes include Style Reference, Character Reference, Content Reference, Depth to Image, Pose to Image, Edge to Image, Sketch to Image, Line Art, Normal Map, Pattern to Image, QR Code to Image, and Text Image Input.

Leonardo.Ai uses tokens as its web app currency (consumed per generation, deducted by computational cost) and a separate API credit system for the REST API. Four web plans are offered: Free, Apprentice, Artisan, and Maestro.

Leonardo.Ai exposes a REST API at cloud.leonardo.ai/api/rest/v1/ covering text-to-image, image-to-image, video generation, upscaling, 3D texture generation, Elements training, and Realtime Canvas (LCM). API credits are purchased separately from web app subscriptions and billed in dollars with no expiry.

Realtime Canvas is an interactive drawing surface that converts brushstrokes into AI-generated imagery in near real-time as the user draws, enabling rapid iterative ideation. It also includes a Realtime Gen mode for adding elements via text prompt without drawing.

Leonardo.Ai generates high-fidelity images from text prompts using a curated suite of first-party and third-party models. The platform offers both foundational models (Phoenix 0.9/1.0) and third-party integrations (FLUX, Ideogram, Seedream) selectable per generation.

Universal Upscaler enlarges images to higher resolutions while generatively adding detail. It offers two modes: Ultra (with Artistic and Realistic style sub-options) and Legacy (Cinematic and other presets), with a hard cap at 20 megapixels output.

Leonardo.Ai supports short-form AI video generation via multiple models including Veo 3/3.1 (Google), Kling Video 3.0, and Leonardo Motion 2.0. Both text-to-video and image-to-video workflows are supported. Output is MP4 only.