"Canva has built its own foundational AI model that generates layered designs users can edit more easily. It's one of several generative AI-related features Canva announced Thursday, alongside expanded access to its AI assistant and content generation capabilities across its app. To date, Canva has partnered with a variety of AI model providers for content generation - Black Forest Labs, Google, and OpenAI among them - and it acquired Leonardo AI last year."
"While these models are adept at image, text or video generation, they lack the ability to produce layered design files that users can then edit themselves, said Jen Thompson, head of product marketing at Canva. "Design is a space that actually hasn't had a fit-for-purpose model yet," Thompson said in an interview with Computerworld ahead of Canva's World Tour event this week."
The company remains tight-lipped on how it uses customer content to train its own AI model, which can generate layered designs that are easier to edit. Canva built a foundational AI model that generates layered, editable design files and expanded AI assistant and content generation capabilities across its app. Canva previously partnered with Black Forest Labs, Google, and OpenAI and acquired Leonardo AI. General-purpose models produce flat, JPEG-style outputs with inseparable elements, limiting editability. The new design model creates layered files that allow AI and human editing in parallel and can be tweaked like templates in the Canva editor. The model promises more control, consistency, integration, and future functionality than mixing general-purpose models.
Read at Computerworld
Unable to calculate read time
Collection
[
|
...
]