So would this kind of lighting and reflection be conceptually impossible without using ray casting? How do people implement lighting/reflection/shadows without it, and what are the differences? I would assume that it would somehow just be less accurate in some way.
barath
I was wondering from the beginning why Ray tracing has gained so much popularity if Rasterization is pretty much similar to ray tracing.
Are the difference above the brief reason why Ray tracing is hyped up so much right now? Or did Nvidia figure out how to efficiently compute ray tracing stuff with RTX series though the concept was available for a longer period of time.
Also, Prof. Keenan, Did you expect Ray tracing to gain popularity in the future beforehand, like before all the hype, knowing how Ray tracing could actually make a difference?
rgrao
^ what NVIDIA says is that it was not possible to do "real-time" raytracing until they released their latest RTX GPUs. I think it was a combination of several factors that enabled them to finally do this, including Moore's Law that allowed them to shrink the transistors and cram billions of them per GPU, implementing the BVH in hardware, and also some deep learning which lets them use things like Deep Learning Super Sampling (DLSS) to do approximate techniques like Monte Carlo rendering and then fill in the incomplete areas with deep learning. Would love to hear Prof. Keenan's thoughts on this as well.
So would this kind of lighting and reflection be conceptually impossible without using ray casting? How do people implement lighting/reflection/shadows without it, and what are the differences? I would assume that it would somehow just be less accurate in some way.
I was wondering from the beginning why Ray tracing has gained so much popularity if Rasterization is pretty much similar to ray tracing.
Are the difference above the brief reason why Ray tracing is hyped up so much right now? Or did Nvidia figure out how to efficiently compute ray tracing stuff with RTX series though the concept was available for a longer period of time.
Also, Prof. Keenan, Did you expect Ray tracing to gain popularity in the future beforehand, like before all the hype, knowing how Ray tracing could actually make a difference?
^ what NVIDIA says is that it was not possible to do "real-time" raytracing until they released their latest RTX GPUs. I think it was a combination of several factors that enabled them to finally do this, including Moore's Law that allowed them to shrink the transistors and cram billions of them per GPU, implementing the BVH in hardware, and also some deep learning which lets them use things like Deep Learning Super Sampling (DLSS) to do approximate techniques like Monte Carlo rendering and then fill in the incomplete areas with deep learning. Would love to hear Prof. Keenan's thoughts on this as well.