Nvidia’s CGI CEO would not look prepared to interchange the true Jensen Huang
A current keynote deal with by Nvidia CEO Jensen Huang convention has provoked some wild reactions. An organization weblog put up satisfied many observers that the complete presentation was computer-generated.
Nvidia has now up to date the weblog put up to make clear that the CGI Huang spoke for less than 14 seconds of the 108-minute keynote.
The distinction between the faux and actual components is fairly clear to my eyes. You possibly can decide it for your self by watching the video under. The CG footage begins at 1:02:41.
The keynote supplied a showcase for Omniverse, a platform used to create and simulate digital 3D worlds. Nvidia used the system to generate Huang’s digital twin.
The corporate first took 1000’s of images of Huang in a collection of poses. These had been used to create a base for his 3D mannequin.
The staff then generated facial gestures for the mannequin that matched Huang’s speech. To reinforce the realism, they mapped a photograph of Huang to the CG model.
Subsequent, they filmed a motion-capture actor studying previous transcripts of Huang’s keynotes whereas bodily mimicking his performances. The footage was used to coach an AI mannequin to duplicate Huang’s actions.
The efficiency was then loaded onto the digital Huang to create the ultimate CG video.
I don’t need to rain on the digital Jensen’s parade, nevertheless it seems like numerous work for one thing that also doesn’t look photorealistic to me.
The CGI doppelganger was additionally proven from a distance that hid potential defects in its options and gestures.
Nvidia’s AI is undoubtedly spectacular, nevertheless it doesn’t look prepared to interchange the true Huang with a digital duplicate.
HT — PCMag
Greetings Humanoids! Do you know now we have a e-newsletter all about AI? You possibly can subscribe to it proper right here.