Design platform Canva has launched its first image generation model, named Canva Craft, developed by its research team to compete with similar models from Google, OpenAI, ByteDance, and Meta.
How Canva Craft Works
The model operates in two stages:
- Interpretation: It interprets users’ prompts and creates a detailed plan, including spatial relationships of elements, relative scale, visual hierarchy, and information for text generation.
- Image Generation: It generates images up to 2K resolution in 30 different aspect ratios suitable for various platforms.
Canva claims this approach enables the creation of images with accurate details and legible text. Users can also make precise edits without affecting other parts of the image.
Training and Features
Instead of using photographs, Canva trained this model using designs from its library, which includes templates for posters, social media posts, presentations, infographics, menus, and typography-rich layouts. This means that images generated with Canva Craft can be transformed into layered designs using Canva Magic, the company’s image-to-design transformer.
