Is Raytracing a necessity?

Is Raytracing features a necessity for games to have?

  • Yes

    Votes: 135 28.2%
  • No

    Votes: 324 67.8%
  • Cannot decide

    Votes: 15 3.1%
  • Others ( Please elaborate )

    Votes: 11 2.3%

  • Total voters
    478

Soren01

Member
If you’re going to add RT to a 2004 game but fail to make it visually compete with current titles and end up with heavier system requirements, then the tech is pointless
 

PaintTinJr

Member
Because most other efforts don't shoot for RTGI. The ones that have show massive improvements and Cyberpunk is one of them. With path/raytracing, the lighting looks nowhere near as good.
Back before the time of advanced deferred raster GI (2005-ish) RT was a much simpler discussion in which RT interchangeably meant RT/PT as appropriate and even at just 1 ray cast and 1 ray bounce (with an ambient) the benefits as real-time lighting were unquestioned - had performance existed to use it even at 320x240 resolutions.

However, in the time since.. rendering resolutions have risen sharply along with impressive non-RT GI lighting that has became so impressive that - RT/PT terminology aside - the value of RT to real-time rendering on inadequate hardware - that includes xx80/xx90 - comes with extra criteria for things like lighting density - relative to render resolution and geometry complexity - or lighting complexity - bounce counts to go beyond basic physically accurate shadows or reflections on non-planar surfaces - leading to many games that provide RT failing to convey that it is indeed RT and to be easily appreciated by your average gamer without specific knowledge of GI lighting.

In the Quake 2 RT the visuals benefit from the low geometric detail and original target 320x240 software raster resolution relative to higher lighting density, and it is accompanied by higher lighting complexity with higher bounce counts giving a volumetric quality to the dynamic lighting way beyond the volume-less lighting of raster GI. The light at a per pixel level correctly diverges - faintly - with the smallest camera motion that gives a more tangible volumetric quality for every pixel , as though each viewport pixel has its own shaft (or shafts) of light propagating through it, like lighting in lightmap games like quake3 under gang ways, where the RT lightmaps captured this multi-sample complexity and could be appreciated as having volume when viewed at mid range, even though the lighting was static

Which sort of highlights a problem IMO on PC/Console games trying to add RT worth while, because resolutions have been driven high in raster and raster GI techniques over time and set to too high a baseline, to the point that the disconnect between hardware resources for a RT light density and complexity that can give a physical based RT volumetric quality to those resolutions is completely unachievable at 60fps, forcing the use of upscaling and MFG (HL2 RT demo illustrates this) which still results in poor RT density or lighting noise levels relative to the native render resolution and geometry complexity failing to deliver meaningful noise free volume based dynamic lighting that easily puts raster GI techniques in pristine noise free high resolution renders to shame, and in a lower tier.

Cyberpunk was the classic case where even on a 4080 super the lighting density and complexity was too low to render at 1440p60 and with it looking like RT with a volumetric quality, and instead looked closer to raster GI techniques, just with physically accurate shadowing and non-planar reflection as a feature, which barely make a difference IMO.
 
Last edited:

JCK75

Member
Doubtful since I've never enabled it in a single game.. in time it will be something we won't want to live without but it's not quite where it needs to be yet.
 
Top Bottom