By now you've probably heard of Google's latest project VLOGGER – an AI model that can create a sophisticated animated avatar from a still image. While it's still in the early research phase, I'll admit that VLOGGER initially filled me with a little bit of existential dread. But since learning more about the project – consider me converted.
It seems like every day we're bombarded with some doom and gloom news about how AI is replacing artists and set for world domination. While VLOGGER may sound like just another example, research reveals it could revolutionise video content, refining dubbed translation and making laggy Zoom calls a thing of the past.
Google's VLOGGER can create video content from a single still image without the need for prior learning, maintaining an advanced level of photorealism. The AI model then interprets audio recordings of the subject speaking, creating hand and body language that emulates natural movements such as blinking, gestures and facial expressions.
The proposed AI model can edit existing video content, with the ability to change a subject's facial expressions. This is also game-changing for dubbed content, as the AI can match the new speech pattern of the translated audio, avoiding that distracting dubbed video effect. Having technical difficulties in your online meeting? VLOGGER can create a virtual stand-in to narrate your audio (perfect for those days when you're not feeling camera-ready).
As the name suggests, VLOGGER could also change video creation, making the need for physical content creators less of a priority. While I don't see the AI model completely wiping out YouTubers, it could mark an interesting evolution that allows creators to focus more on quality content rather than physical image.
While I'm always cautious at the announcement of new AI technologies, I'm excited to see how VLOGGER evolves – although I think the name is somewhat diminishing. Yes, it could produce a slew of soulless YouTube vloggers, but fundamentally it seems to offer innovative solutions to existing issues with video content. As we've seen with Meta's VR avatars and Sora AI, video content tools are evolving at a rapid rate, and I'm excited to witness this new era of innovation.
