Canva launches Craft, its first image generation model, trained on human-designed templates, available to all users, with plans to support third-party chatbots (Ivan Mehta/TechCrunch)

Design platform Canva has released Canva Craft, its first image generation model developed by its research team. The model was trained using human-designed templates from Canva’s library, such as presentations, infographics, menus, and typography-rich layouts, rather than using photographs. According to TechCrunch, the model is available to all users through the Dream Lab tab and the company plans to support third-party chatbots.

The model operates in two stages. First, it interprets user prompts to create a plan regarding spatial relationships, visual hierarchy, relative scale, and text generation information. In the second stage, it generates images with up to 2K resolution in as many as 30 different aspect ratios. TechCrunch reports that this approach enables the creation of legible text and accurate object details, and allows users to make precise edits without altering other parts of an image.

Additionally, images created with Canva Craft can be converted into layered designs using Canva Magic, an image-to-design transformer. This follows a model released by the platform last year that understands designs and layers. TechCrunch notes that the model was developed to compete with similar models from Meta, OpenAI, Google, and ByteDance.

Brief written by urgent.news from Techmeme, TechCrunch — 2 reports on this story. Machine-written — may contain errors; check the original before relying on it.