NVIDIA thus presented its innovations in the field of graphics, where it is about AI, ray tracing, scene lighting, and especially in the case of the named Neural Radiance Caching, we have to prepare for a “breakthrough”.
–
We have already dealt with the Path Tracing technique in the case of the test RTX verze Quake II, so in short it can be said that it is a more demanding technique, where more monitored rays are used for each pixel, thanks to which the results can be much better. Therefore, path tracing has been used in real time only in graphically simple games, such as the mentioned Quake II or Minecraft, and even in their case it is a very demanding technique, where the card’s power must be used for additional noise removal, or it is possible to reduce noise by constantly rendering the the scene itself and averaging the results, but of course you can no longer talk about rendering graphics in real time. –
Real-Time Path Tracing in a scene such as the theater space shown above, there has so far been such a small science fiction in terms of the capabilities of today’s hardware. So what has changed? Is NVIDIA presenting a new generation of hardware that will be much more powerful in ray tracing than Ampere? The time has not yet come to it, and in fact it is a new path tracing algorithm, which is to somehow prioritize such rays and their reflections, which should contribute as much as possible to the rendering of the resulting image.
In short and well, there is a new algorithm for path tracing, which is according to NVIDIA about 100 times faster than the old one. –
And then we have the breakthrough Neural Radiance Caching, which is to serve as a technique for global enlightenment of scenes. Both RT and Tensor cores are used here, while the graphics card dynamically creates a tiny neural network directly when rendering the image. This network then learns how the light is to spread in the scene, and we can see this in the video on the example of a walking tiger, in whose fur a very subtle play of light and shadows takes place. NVIDIA specifically used the RTX 3090 card here, ie today’s generation.
Prices of related products:
–