Krea AI has rolled out a new feature that turns regular images into 3D objects that users can manipulate within AI-generated scenes in real time.
Users can resize these 3D objects and position them at any angle within their generated scenes. To access this feature, users need to head to the "Generate" menu and select "Realtime." From there, they can drag an image into the left drawing area, right-click it, and choose "Convert to 3D." After deleting the original image, users can work with the 3D model, adjusting its position and rotation while watching the changes in the preview window on the right.
Video: Justine Moore
Character consistency across images
When asked about using this feature to maintain consistent characters across multiple images, Krea AI mentioned that users could use Krea Train to finetune a flux model and combine that with the new real-time and 3D capabilities.
The platform offers several other tools beyond 3D conversion. Users can create real-time styles, modify image environments with a single click, maintain consistent lighting and colors, generate ultra-slow motion at 120 fps, and merge and animate images. YouTuber Olivio Sarikas has created a detailed video showcasing his experience with the new 3D feature.
Krea AI tends to be tight-lipped about the technical details behind their features, usually only sharing brief updates on social media. They haven't explained exactly how their new 3D capability works. But in general, the field of "novel view synthesis" - creating new perspectives from a single 2D image - is seeing progress recently. Companies like Stability AI are actively working in this area, along with researchers at Tencent and Nvidia.
