CopernicAI: Pioneering the Generative AI Worlds

Generative AI has witnessed astonishing growth in recent times. The quality of text and imagery generated by AI has been continuously impressing us. And now, the focus is shifting towards generating 3D worlds.

The 2+1D Bubble World Approach The first step in creating generative worlds is the 2+1D bubble world approach. Here, generated 2D imagery is enhanced with 1D depth information. Thanks to recent research published in premier deep learning conferences, we can already achieve generative 2+1D bubble worlds as we progress towards fully AI-generated 3D worlds.

AI Generated 360° Panoramas (Gen360) Gen360 allows you to convert text to 360° panoramas. The alpha release shows major improvements compared to the pre-alpha version, with more enhancements on the way. You can freely use the generated 360° images for both commercial and non-commercial purposes, following the Creative ML OpenRAIL-M license.

There's also a free Gen360 API available for a limited time. Whether you're a terminal user, an IDE enthusiast, or prefer using Postman, you can access this API. Once you sign up, you'll receive access instructions.

AI Generated 2+1D Bubble Worlds in VR You can turn your generated 360° image into a 2+1D bubble world (panoramic + depth) using copernic360. This enables you to explore these unique worlds in VR.

The Roadmap of Panoramic AI

• Efficient Panoramic AI (Published in ICLR 2021): It introduced efficient techniques that significantly improved computational complexity and achieved state-of-the-art performance on low-resolution tasks, though still limited to low-resolution.
• Partially Scalable Panoramic AI (Published in ICLR 2022): It could handle high-resolution input data but only produced low-resolution output data, suitable for classical AI tasks with low-resolution outputs like image classification, object detection, and tracking.
• Scalable Panoramic AI (Published in ICLR 2023): This achieved high-resolution input and output data, reaching state-of-the-art for dense-prediction tasks such as semantic segmentation and depth estimation. It forms the building blocks for Generative Panoramic AI.

Generative Panoramic AI The early alpha of Generative AI panoramas and videos using the building blocks of Scalable Panoramic AI is already available, with major improvements expected soon.

In conclusion, CopernicAI is at the forefront of generative AI, especially in the realm of creating immersive worlds, whether it's 2+1D bubble worlds or the upcoming full 3D worlds. Stay tuned for more updates by signing up to receive announcements and invitations to closed alpha releases.