I’ve been doing a lot of stuff with procedural textures recently. I think that might be the way everything will have to go in the future.
3D graphics have gotten progressively better, more detailed, more reactive and interactive. Light bounces around in realistic ways, or if you like, unrealistic ways. Graphics cards can dynamically increase and decrease the resolution of models. We don’t really use that feature very often, but it exists and works fairly well.
One of the last frontiers of dynamic, resolution independent graphics, is the texture.
If you want to change the color of a surface in a 3D game or movie, one of the easiest ways is to apply a texture to it. It’s like an incredibly versatile sticker. It can be sliced up, moved around, rotated, adjusted, and animated. Everything that you can think to do to an image on a computer, you can probably do to a texture. Usually in real time.
The thing you can’t do is zoom in forever. There is a maximum number of pixels that you can store in a texture. If the texture you use to wrap a 3D surface is of a lower resolution than that image takes up on screen, you will see those pixels. If a player gets real close to a surface, they are going to see those pixels.
I don’t think this is really a problem, and most people who play games seem to agree, but there will come a time that being able to see the pixels won’t cut it.
There are two paths to go here. Bigger and bigger textures, or procedurally created materials that scale as they need to. The first one takes memory, the second one takes processing power.
For a long time in graphics, memory, or the time it took for a machine to swap images in and out of memory, was cheaper than processing. With textures already being made regularly at 4096x4096 the options for just making them bigger are sort of running out. Not only will we run out of memory to store huge textures, but we will run out of the ability for a person to meaningfully use that higher resolution.
I think the path over the next few years will be to use more small images in procedurally blended ways. Three, relatively tiny, grass textures mixed together to cover thousands of meters of plains terrain. Metal constructed surfaces made of a miniscule handful of textures covering huge surfaces with no visible repeats.
These materials are more complex to assemble, but they can take less memory while taking advantage of the massive amounts of processing power graphics cards have on tap.
A lot of these techniques are being used already in non-real time applications, but more and more this will be the path forward for games and VR content too.
If PBR setups were the revolution of the last decade, procedural materials will be the art creation workflow for the next decade, at least.