Deepmind's new video AI, Transframer, can handle a whole range of image and video tasks – and dream up 30-second videos from a single frame. Generative AI systems have moved from research labs to industrial and consumer applications in recent years, kicked off by OpenAI's large-scale language model GPT-3. Then last April, the company introduced the DALL-E 2 imaging system, which indirectly spawned alternatives such as Midjourney and Stable Diffusion. Google sister Deepmind is now showing Transframer, an AI model that could offer a glimpse of the next generation of generative AI models. Deepmind's Transframer is a visual prediction framework that can solve eight image modeling and processing tasks at once, such as depth estimation, instance segmentation, object recognition or video prediction.

In 2010, I attended the IEEE (Institute of Electrical and Electronics Engineers) CVPR (Computer Vision and Pattern Recognition) conference at the Hyatt in downtown San Francisco. I didn't expect the conference to be as large as it was, but it had more than 1,500 in attendance, to the best of my recollection. The conference reminded me of the size of the conferences held at the same hotel when the industry was arguing over different standards for Wi-Fi, with multi-billion dollar markets at stake. However, unlike the practical approach of implementing the maturing Wi-Fi technology, where presentations were mainly made by engineers working for companies competing over their ability to assert their intellectual rights into the standards, the CVPR presentations were mainly made by university researchers, and researchers from "deep-research" arms of some of the world's largest technology companies, who didn't expect the fruit of their research to reach maturity anytime soon. One of the presentations I sat through struck a chord with me. The presenter showed a 30-second video taken from a dash camera.