Canva launches Canva Craft, its first in-house image generator

In short: Canva has released its first in-house image generation model, called Canva Craft, and made it available to all users.

What happened

Canva, the design app many people use for social posts and presentations, says it has built its own system for generating images from text prompts. It is called Canva Craft, and Canva says it was developed by its research team.

Canva says the tool works in two steps. First, it reads your prompt and plans the layout, like where objects go, how big they are, and what text should look like (similar to sketching a draft before coloring it in). Second, it creates the final image, up to 2K resolution, and in many different shapes and sizes so it can fit different platforms.

A key detail is how Canva says it trained the model. Instead of training mostly on photos, Canva says it trained on human-made design templates from its own library, like posters, menus, infographics, and slides with lots of text.

Canva Craft is available to all users inside Canva under a “Dream Lab” tab. Canva also plans to offer it through an API (a way for other apps to connect to Canva), and it says it wants to support image generation inside third-party chatbots like ChatGPT and Gemini through Canva’s MCP.

Why it matters

For everyday users, this could mean AI images in Canva that keep text readable and make small edits easier without changing the whole picture. For Canva, using its own model may also reduce how much it pays other companies when it generates images, which the company has pointed to as a cost concern.

Source: TechCrunch AI