When done right, slow motion footage can be very dramatic and interesting to watch. Some OEMs as now billing the ability to take good slow motion footage as one of the USPs of their handsets and with good reason. It requires substantial processing power and large memories which are primarily found on high-end devices. However, NVIDIA has a new solution which can make it easier for smartphones of the future to capture slow motion video. Its solution involves the use of artificial intelligence.
NVIDIA’s AI is capable of converting 30 frames-per-second video into remarkable 240fps slow motion footage. You can see the difference in the video posted above and it’s really interesting how AI can create a convincing fake, essentially.
It does that by analyzing two different frames and creating intermediary footage which it achieves by tracking the movement of the objects in the footage from one frame to the next. NVIDIA says that it can produce up to seven intermediary frames which leads to accurate results.
The system does have some limitations. The AI first had to be trained so that it could work with certain types of footage so before it could create a convincing slow motion video of a car skidding through water, the AI first needed to be trained on similar footage.
It’s not ready for commercial launch just yet but this is a great proof of concept from NVIDIA which shows that AI can play a major role in improving slow motion video capture on smartphones in the future.