Thank you so much for this video, there were so many extremely enlighting infos in it :) Regarding rasterization and fields I also think that both don't fit together. However, if rasterization is something that is preferable, the radiation should maybe not be a field but objects that are treated as radiators. In raytracing, the idea has always been to shoot viewing-rays like pythagoras did. What if an object was a neural network that I can ask to radiate the object onto a point in space? (a nice side-effect would be, that a neural network could learn that during training with photos that have been shot in real-world physical constraints, the point is traveling propertional to the distance of the radiation-source)
Great talk!
Thank you so much for this video, there were so many extremely enlighting infos in it :)
Regarding rasterization and fields I also think that both don't fit together. However, if rasterization is something that is preferable, the radiation should maybe not be a field but objects that are treated as radiators. In raytracing, the idea has always been to shoot viewing-rays like pythagoras did. What if an object was a neural network that I can ask to radiate the object onto a point in space?
(a nice side-effect would be, that a neural network could learn that during training with photos that have been shot in real-world physical constraints, the point is traveling propertional to the distance of the radiation-source)