started · updated
Canva launches Canva Craft AI image generation model
Canva has introduced Canva Craft, its first proprietary image generation model developed by Canva Research. Unlike many existing AI models that generate pixels directly from text, Canva Craft utilizes a two-step process: a planning model that analyzes user prompts to determine composition, scale, and text placement, followed by a rendering model that produces high-resolution images up to 2K.
The model was trained on millions of design templates—such as posters, social media posts, and infographics—rather than just photographs. This training approach allows the AI to better understand layout, visual hierarchy, and typography. The resulting images support over 30 aspect ratios and are designed to work with Canva’s Magic Layers, allowing users to edit specific elements or text within a generated image.
Canva Craft is available to users via the Dream Lab feature and to developers through a public API. The tool supports text generation in 55 languages.