News Watch AMD's DirectX Raytracing Demo on Future RDNA 2 GPU

[digitalgriffin]

You can't say that for all light sources and surface sizes. If my primary light source is a candle, and my secondary light source is sunlight or an explosion, then even tertiary bounces could overpower direct lighting from the primary.

Likewise, if my secondary illumination is bouncing off a planet I'm orbiting, then I could be miles away from it and it could still be the dominant illumination.


I don't know what you mean by "full rays". Do you mean a uniform sampling grid, relative to the incident ray?

I also don't follow what you mean by selectively rendering objects in multiple passes. Are these screen-space passes, in which case it would only apply to the primary rays?

By primary and secondary lighting i mean same light source. I mean primary incident and secondary incident. Primary incident is direct lighting from a surface. Secondary is reflected light lighting from the same source.

By full rays i mean a surface point by surface point sampling. The greater the incident count (numer of bounces an ray is allowed to gave to create something like an infinity mirror) the more computationally expensive it gets and quickly.

GI uses a somewhat random sampling of an area and then uses a temporal filter to denoise. With GI you can get away with this as secondary reflection lighting tends to be defuse in nature.

Seus uses a similar technique in ptgi seus shader minecraft. Unfortunately this can have side effects with sharp light transitions. Thus if a light source quickly changes you have to temporarily turn off temporal aa.

 

[cryoburner]

Go back and watch Nvidia's Star Wars trailer. That's a demo with reflections literally everywhere, but it just wouldn't have that life-like quality without GI.

It probably also helps that they had Epic Games and Industrial Light & Magic working on that demo, whereas AMD probably only had the guy who used to do graphics card box art working out of a broom closet. : 3

Seriously, guys, did you not see the benchmarks of the different RTX effects? Are you really saying that GI is no big deal, performance-wise?

To be fair, a lot of that likely comes down to the fact that the global illumination in Metro is being performed on everything on-screen, whereas the per-pixel reflections in Battlefield are typically only being performed on a small portion of the screen. I believe less-reflective surfaces have less sampling performed on them in that game, and at reduced RTX settings, only certain objects get reflections. In this demo though, nearly everything has been given a mirrored surface, so rays need to be cast for most pixels. In any case, it's hard to tell much about performance from what's being shown here.

 

[bit_user]

the per-pixel reflections in Battlefield are typically only being performed on a small portion of the screen.

It needs to be fast enough that you can walk up to a (reflective) window, filling your screen, without your framerate dropping precipitously.

 

[bit_user]

GI uses a somewhat random sampling of an area and then uses a temporal filter to denoise.

I'm not saying they don't use temporal filtering, but it's certainly not only temporal filtering. They highlighted their use of deep learning, to filter GI light maps.

 

[joeblowsmynose]

...


I think you're exaggerating. It does use fewer samples and maybe things like TAA. But one thing people like about ray tracing is that it involves way fewer hacks and optimizations than traditional rendering.

I also don't really see how it's relevant to the discussion, since I was never comparing professional renderers, in the first place.

You woudn't understand how heavy the optimizations are unless you understood how professional renderers actually work.

There was a fair bit in all that (tour post) that made literally no sense to me ... but I'll focus on this part.

I have been a hobby 3D artist for a while now (~20 yrs), and unless you understand the difference between what is done for say, movies, and what is done for games, you would be forgiven for not understanding the optimizations that first the artists use, then the game engine uses to portray that, then and then the GPU drivers optimize that further yet.

The difference between professional GPUs (quadro / WX) and gaming ones is QA. A game GPU can be very inaccurate, and because games are well, just games, they can get away with a frame here or there not being accurate. You won't notice it. But as a pro, you need accuracy. Just like a Pro musician isn't going to record a master track as MP3.

This is what drives the optimizations behind game graphics vs 3D rendered graphics. In fact until very recent advances in APIs, GPUs were actually useless at photorealistic rendering, because A) they weren't designed for such and B) the simplistic code couldn't handle shaders that needed high levels of ram. Things have gotten better in that area but that means nothing toward the fact that game engines and game specific driver attributes, are optimizing for the highest FPS, at the expense of making things look "good enough" vs fully accurate.

In fact if this hadn't been the case, games would be looking pretty shi ... er, crappy these days.