Next gen lighting technologies - voxelised, traced, and everything else *spawn*

Nvidia says that Turing and Pascal cards will take two to three times longer, respectively, to render each frame than a Turing RTX card. This difference is particularly acute on Pascal cards. In Turing, the 32-bit integer workload used for raytracing can run concurrently with the 32-bit floating-point workload used for other graphical tasks. That's not the case on Pascal, where the workloads will have to be run consecutively.

That raytracing can be performed on these older chips isn't a big surprise. During DXR's development, Microsoft used a compute-shader-based raytracing algorithm, so clearly the dedicated hardware isn't necessary. However, the substantial performance difference shows that the dedicated hardware is going to be important, at least for now.
https://arstechnica.com/gaming/2019...ctx-raytracing-to-geforce-10-16-series-cards/
 
Tomb Raider runs very well even on an RTX 2060 @1440p. DLSS handles IQ well too.

1440.png



https://www.techpowerup.com/reviews/NVIDIA/Shadow_of_the_Tomb_Raider_RTX_DLSS/4.html
The graph is misleading, rtx number applies to a DLSS scaled-up resolution.
Real impact from RT is more apparent.
I was worried about this when I first heard about DLSS.

Link to real RTX impact.
https://www.techpowerup.com/reviews/NVIDIA/Shadow_of_the_Tomb_Raider_RTX_DLSS/2.html
 
Last edited by a moderator:
So after trying I can sum up the effect of RT shadows to this:

-Adds soft high res shadows to everything, jaggies and low res shadows are 90% gone
-Makes most point and small light cast shadows, like candles, fire crackers, small lamps .. etc, I say most because not all of them cast shadows. And No muzzleflashes don't cast them sadly.
-Increases the number of objects that cast shadows and the number of shadows in any scene.

When held up against my list, 5 points out of 6 are achieved, and one of them is partially achieved. Overall the effect of RT shadows is subtle in the game.

What we need ray traced shadows to do is this:
-Add percentage closer filtering/contact hardening/ to shadows (guaranteed)
Check
-Increase shadow resolution and force small geometry to cast shadows (guaranteed)
Check
-Fix shadows flickering (not 100% guaranteed)
Check, 90% successful, some flickering is noticed every now and then.
-Make every point or small light cast shadows (announced but not guaranteed on every light)
partially checked, several small lights still don't cast shadows.
-Fix the flash light shadows to include every 3d object (not 100% guaranteed?
Check
-Force every dynamic light (fire effects, explosions, flares, gun muzzle flashes) to cast shadows (not 100% guaranteed)
Not checked, none of them cast shadows, which is a missed opportunity, and disappointing to say the least.

So 5 out of 6 points, not bad, still the RT shadows improvements in the game are subtle and not readily apparent, unlike Battlefield's reflections or Metro's lighting.
 
Last edited:
The graph is misleading, rtx number applies to a DLSS scaled-up resolution.
Real impact from RT is more apparent.
I was worried about this when I first heard about DLSS.
It's possible any performance degradation might be due to lack of sufficient RT cores. The move to 7nm should increase the effectiveness of RT cores, but wonder if adding additional RT cores is feasible.

Once we move towards Ultra HD, we can see that the DLSS + RTX Ultra performance effect is much smaller. This is an indication that RTX does not have enough RT cores to keep up with the shader engine (which likes to go faster but is being held back by the RT cores much like what a CPU bottleneck does in a lower resolution opposed towards a high resolution).
index.php



index.php
https://www.guru3d.com/articles-pag...-graphics-performance-benchmark-review,8.html
 
Not checked, none of them cast shadows, which is a missed opportunity, and disappointing to say the least.
Could you guys make some arguments of why you want to see shadows from muzzle flashes? (or similar flashy 'fx' light sources)
I hear this often but i fail to see how or when this could be important.
 
Could you guys make some arguments of why you want to see shadows from muzzle flashes? (or similar flashy 'fx' light sources)
I hear this often but i fail to see how or when this could be important.

lol but honestly. I had a chaotic crazy shootout with my squad in Division 2 in night time in the fog where we walked into an ambushed open area. It was tremendously awesome to be in such chaos, to have muzzle flashes that light up the environment and shadow would have sealed the deal.
 
Immersion for dark environments. In daytime, the brightness of muzzle flashes may not be sufficient though.
 
Ah, thanks - i see it makes a lot of sense in the dark. (Never noticed myself - too busy with hating emissive materials that only cause bloom to the environment :) )

I don't know if denoising can work with such high frequency, or if noise would be just acceptable?
But likely the effect is equally important for enemies as it is for the players gun fire. And we want omnidirectional lights, so with shadow maps this is expensive.
Maybe a shadowed spotlight along the shot, plus a dimmed unshadowed point light could do the trick well enough.
 
Why are results from the same people far lower than the OptiX 5 results?

Hello, I'm the guy behind those benchmark scores. Fermat 2.0 introduced support for more complex materials and better samplers so one can't compare results with our previous results with Fermat 1.0. The newer edition requires less samples but produces less noise as well.
512 passes is enough to stress the GPUs, each iteration gather one sample per pixel.
 
Ah, thanks - i see it makes a lot of sense in the dark. (Never noticed myself - too busy with hating emissive materials that only cause bloom to the environment :) )

I don't know if denoising can work with such high frequency, or if noise would be just acceptable?
But likely the effect is equally important for enemies as it is for the players gun fire. And we want omnidirectional lights, so with shadow maps this is expensive.
Maybe a shadowed spotlight along the shot, plus a dimmed unshadowed point light could do the trick well enough.
Ultra high quality is not necessary. Somewhat wrong shadows > no shadows.

Killzone 2 used spotlights to save on resources:

 
Cool indeed, technologically speaking, but that execution is not very convincing.
They simply shouldn't have used it for leaves, it looks wrong.
If they want light leaking through leaves properly they would need to handle it as light bounce and how it scatters out from leaf. (So handle shadow as if it is fully dark as light is diverted from it's original path and leak light trough with GI sampling with proper BRDF from leaf to get directionality to light.)

Otherwise shadows look quite good.
 
Last edited:
Originally the artist decided in this setup it makes no sense for sharp shadows. With RTX it makes no sense, but it shows RTX can shadow map:

Capturefdsfds.JPG

Lot of objects failing to cast shadow as in Metro:

Capturesdfdsfds.JPG

Again it makes no sense for sharp shadows, but RTX needs to show to people it can do what shadow mapping can do:

Capturetrtrtr.JPG

About this - it is not necessarily an improvement:

Capturerewrwerewrew.JPG

This is just the same bad artifact as here:

Capturegfdgfd.JPG

Source: https://www.techpowerup.com/reviews/NVIDIA/Shadow_of_the_Tomb_Raider_RTX_DLSS/2.html

An improvement can be seen here:

upload_2019-3-21_11-50-3.png

I admit the game improved much in peter panning.

Sometimes RTX makes it worse:

upload_2019-3-21_11-44-10.png

upload_2019-3-21_11-44-50.png

The shadows of the leaves are mixing sharp shadows with blured shadows for RTX Off. Which is nearer to real life than RTX On version.

Capturerewrew.JPG

Transparent shadows of leaves are not an improvement.

I personally take decisions only based on the way the final render looks. Visuals are mandatory as normal humans are not able to tell if complex reflections of reflections are correct or not. Most of people would not be able to tell which is more correct - env mapping or RT reflections.

 
Sometimes RTX makes it worse:

View attachment 2944

View attachment 2945

The shadows of the leaves are mixing sharp shadows with blured shadows for RTX Off. Which is nearer to real life than RTX On version.

View attachment 2946
Pretty sure in that scene there are plants near ground which cast those sharp shadows, so the RTX one is closer to the correct solution.
Transparent shadows of leaves are not an improvement.
Happily transparent leaves are only on with the Ultra setting, so use high to have proper shadows.
 
Back
Top