"We started by creating flat images with diffusion models. Omni models have taken that a step further, where you're able to edit those flat images with a lot of sophistication through prompting. But the tools have made you prompt your way to the final result, which, for a visual medium, is challenging,"
"What we've found is that people . . . want the ability to start with a prompt and get far, but also be able to iterate directly themselves."
"Earlier this year, the company unveiled an AI assistant called Canva AI that has a chat-like interface for generating visual content from text prompts. The platform is now making that assistant available throughout the interface, including the design and elements tabs. Users can also tag (@mention) the bot in comments to get text or media suggestions while working on a project with others. Plus, the AI tool can now generate 3D objects and allows users to copy the art style of any design."
Canva launched a foundational design model trained on its design elements that generates editable layered designs and objects instead of flat images. The model works across formats including social media posts, presentations, whiteboards, and websites, enabling direct iteration on layers and objects. Canva integrated its Canva AI assistant across the interface and elements tabs, added @mention support in comments, enabled 3D object generation, and added the ability to copy art styles. The platform introduced a spreadsheet product and a mini-app creation feature, and connected them so users can build data-driven widgets from spreadsheet data. The company also acquired ad analytics firm MagicBrief.
Read at TechCrunch
Unable to calculate read time
Collection
[
|
...
]