This is true of literally any technology. There are so many things that can be improved in the early stages that progress seems very fast. Over time, the industry finds most of the optimal ways of doing things and starts hitting diminishing returns on research & development.
The only way to break out of this cycle is to discover a paradigm shift that changes the overall structure of the industry and forces a rethinking of existing solutions.
The automobile is a very mature technology and is thus a great example of these trends. Cars have achieved optimal design and slowed to incremental progress multiple times, only to have the cycle broken by paradigm shifts. The most recent one is electrification.
path tracing is a paradigm shift, a completely different way of showing a scene to that normally done, it’s just a slow and expensive one (that has existed for many years but only started to become possible in real time recently due to advancing gpu hardware)
Yes, usually the improvement is minimal. That is because games are designed around rasterization and have path tracing as an afterthought. The quality of path tracing still isn’t great because a bunch of tricks are currently needed to make it run faster.
You could say the same about EVs actually, they have existed since like the 1920s but only are becoming useful for actual driving because of advancing battery technology.
This is true of literally any technology. There are so many things that can be improved in the early stages that progress seems very fast. Over time, the industry finds most of the optimal ways of doing things and starts hitting diminishing returns on research & development.
The only way to break out of this cycle is to discover a paradigm shift that changes the overall structure of the industry and forces a rethinking of existing solutions.
The automobile is a very mature technology and is thus a great example of these trends. Cars have achieved optimal design and slowed to incremental progress multiple times, only to have the cycle broken by paradigm shifts. The most recent one is electrification.
Okay then why are they arbitrarily requiring new GPUs? It’s not just about the diminishing returns of “next gen graphics”.
path tracing is a paradigm shift, a completely different way of showing a scene to that normally done, it’s just a slow and expensive one (that has existed for many years but only started to become possible in real time recently due to advancing gpu hardware)
Yes, usually the improvement is minimal. That is because games are designed around rasterization and have path tracing as an afterthought. The quality of path tracing still isn’t great because a bunch of tricks are currently needed to make it run faster.
You could say the same about EVs actually, they have existed since like the 1920s but only are becoming useful for actual driving because of advancing battery technology.
That’s exactly why. Diminishing returns means exponentially more processing power for minimal visual improvement.
I think my real question is what point do we stop trying until researchers make another breakthrough?