Canva has launched its first in-house AI image-generation model, designed to create visuals with better layouts, sharper text and editable elements.
Generating an image with AI is one thing. Getting the text, layout and placement of every element right is another. Canva reckons it has a solution with Canva Craft, its first image-generation model developed and trained from scratch by its in-house AI research team.
Craft is built specifically for design rather than just generating attractive images. It can create visuals at up to 2K resolution, supports more than 30 aspect ratios and can render text in 55 languages. That means everything from social media posts and marketing banners to posters and presentation visuals is on the menu.
Also Read: Oxford University Press and Canva Partner to Bring Design Tools to Indian Classrooms
What makes Craft interesting is how it approaches image generation. Instead of jumping straight from a prompt to pixels, it uses a two-stage process. First, an image-planning model works out the composition, including object placement, visual hierarchy and text positioning. A separate rendering model then turns that plan into the final image.
The idea is to give users more control over the result, particularly when a design needs precise layouts and readable text. Craft is also built to work with Canva’s Magic Layers, which can turn generated images into editable design elements.
Canva says Craft ranked fourth among image-generation models on DesignArena at launch. It also ranked first in the People & Portraits and Landscapes & Nature categories.
The model is available globally through Canva’s Dream Lab and can also be accessed by developers through a public API. Faster versions, called Craft-fast and Craft-ultrafast, are in development.
For now, Craft’s biggest promise isn’t simply making AI-generated images look better. It’s making them more usable. Because sometimes, getting the picture right is only half the design job.