'

Nvidia showcases AI Slow Motion Video Conversion Tech

Turning 30FPS footage into 240FPS footage

Nvidia showcases AI Slow Motion Video Conversion Tech

Nvidia showcases AI Slow Motion Video Conversion Tech

One of Nvidia's fastest growing markets is in the machine learning category, where their Tesla V100 series GPUs are utilising their purpose-built tensor cores to accelerate a large number of deep learning workloads, accelerating AI training and development to create compelling new technologies. 

Nvidia's research arm has created a deep learning system which can create high-quality slow-motion videos using standard 30FPS footage, creating a reasonably accurate approximation of real-like slow-motion video without the need for specialised equipment. 

Using their AI-focused Tesla V100 GPUs and a  cuDNN-accelerated PyTorch deep learning framework, Nvidia trained their system to predict extra frames for video content, allowing low framerate footage to be given extra fluidity and for slow-motion footage to be made even slower, up to four times slower. 

This technology can create multiple intermediate frames, allowing footage to become several times more fluid when played back side-by-side. Nvidia has stated that they used 11,000 240FPS videos to train their AI, allowing it to create approximate extra detail in low framerate footage for extra fluidity.   

 

While high-framerate cameras are becoming increasingly popular in modern smartphones, it is undeniable that such equipment is both power and memory intensive and comes only on high-cost devices, making this AI-driven technique extremely useful for post-recording high-framerate enhancements. 

Nvidia's technique seems to be both faster and more accurate than existing slow-motion interpolation techniques, with Nvidia being able to offer up to seven intermediary frames to create a potential 8x slow-motion effect while retaining the same levels of fluidity as before. 

You can join the discussion on Nvidia's AI-driven slow-motion video conversion technology on the OC3D Forums

«Prev 1 Next»

Most Recent Comments

19-06-2018, 13:09:27

NeverBackDown
Impressive stuff. This is what happens when 2 companies exist but 1 has all the R&D.Quote
Reply
x

Register for the OC3D Newsletter

Subscribing to the OC3D newsletter will keep you up-to-date on the latest technology reviews, competitions and goings-on at Overclock3D. We won't share your email address with ANYONE, and we will only email you with updates on site news, reviews, and competitions and you can unsubscribe easily at any time.

Simply enter your name and email address into the box below and be sure to click on the links in the confirmation emails that will arrive in your e-mail shortly after to complete the registration.

If you run into any problems, just drop us a message on the forums.