DLSS doubt


Hi everyone! Here my first post on this forum :D

I’am currently enrolled in an Artificial Intelligence subject and have been assigned to present on an AI.

I am having difficulty finding detailed information about the AI techniques that DLSS employs (apart from DeepLearning) and how they are applied.

If some insights are provided or resources where I could learn more, I’d be so grateful.

Hi there @yiqiyt and welcome to the NVIDIA developer forums!

I hope you will understand that we don’t share the exact models that are used for DLSS. But if you check out for example NVIDIA DLSS Overview & Game Integrations | NVIDIA On-Demand during the first 2 to 8 minutes, Andrew describes in some detail how the Deep Learning trainig works. It basically is a combination of traditional convolutional networks with back-propagation and an autoencoder.

If you are looking for details on the “hot off the press” Ray Reconstruction as part of DLSS, I recommend this video:

I hope this helps!