Discussion about lighting (usage of lighting techniques).

[Silent_Buddha]

OK, I want to first establish what this is and is not. Also, this may or may not be the correct sub-forum for this, but if it can be kept on topic, I believe it can be a good technical discussion about "how lighting is used in games" regardless of the technology behind the lighting used in any specific game.

This should be a discussion about how lighting techniques are used in games. This isn't about RT good or RT bad. This isn't necessarily about the technology behind the lighting being good or bad. This should be about how lighting techniques are used in games.

Why am I starting this? I see too often people get caught up in hardware accelerated RT lighting = good and pre-"hardware acceleerated RT" lighting = bad. Neither of those statements are necessarily true. There are cases where RT lighting (the use of it) is bad and there are equally as many cases where pre-RT lighting is good (again the use of it). Enough that fans of good cinematography will often prefer the lighting in a game with zero hardware accelerated RT or even no RT period versus a game that uses hardware accelerated RT lighting instead of a non-accelerated solution or non-RT solution, IE - older lighting models.

To put this into perspective, it can be good to look at films, TV and even advertisements. Realistic lighting is frankly uninteresting and there are almost no good films, TV shows, or advertisements that use realistic lighting.

What does that mean? It means that good cinematography is all about how to manipulate real world lighting to produce interesting and completely unrealistic lighting conditions. It's about making lighting behave in interesting and unrealistic ways in order to frame images or sequences in interesting, captivating and meaningful ways. It's the difference between a well shot film with good or even mediocre cinematography (unrealistic lighting that behaves realistically) and a home video (100% realistic lighting).

In both cases, home movies (I'd include self made TV advertisements and even most soap operas here) and movies, shows or advertisements which uses cinematogrphic lighting techniques are using "real" lighting. The only difference is that the later is attempting to make lighting work in non-realistic but still belieable ways.

Here's a great example of an advertisement along with a director/cinematographer talking about it (camera framing and he talks a fair bit about the lighting tricks used).

Feel free to browse his other videos where he also talks about camera framing and lighting for game cutscenes, advertisements, cinematics, etc.

If you watch that you can often see cinematographic lighting techniques that older non-RT lighting tech provides for free. Again, note that this doesn't mean that non-RT lighting is better than RT lighting. This is about how the tech is used.

So, this is why when looking at something with what I consider one of the better implmenetations of global RT illumination (Metro: Exodus), I also view it poorly with regards to lighting in games. It implements global illumination really well, but then fails to use any cinematographic lighting techniques to make the actual visuals engaging and enjoyable to look at. IE - people have previously heard me talk about how I much prefer the non-RT lighting of the Metro: Exodus to the enhanced RT lighting in Metro: Exodus EE.

In the former (Metro: Exodus), you can tell they actually spent time to not only attempt to wrestle old lighting techniques into something approximating realistic light behavior but also took time to implement cinematographic lighting techniques.

Whereas in the latter (Metro: Exodus EE) they implemented global illumination and called it a day with almost zero attempts to replicate cinematographic lighting techniques to either make scenes more interesting, more clear or more engaging/suspenseful. IE - lack of facial highlighting makes dialogs less engaging/clear or lightened dark areas becoming less suspenseful.

So, just like there are games that stood out for great use of non-hardware accelerated RT and ones that were bad ... so too are there hardware accelerated games that implement RT well but use it badly (Metro: Exodus EE) and ones that use it well.

IMO, we're still not to a place where the lighting artists are using RT lighting as well in games (even the best ones) as lighting artists have done in some of the better much less the best examples of non-hardware accelerated RT games.

I wonder if it's either a lack of budgeting (we can implement RT lighting cheaper in game than paying lighting artists so lets reduce the lighting budget), lack of experience (how can we manipulate RT lighting to reproduce these effects that are use in good cinematography?), laziness (related to budget as why spend time on lighting if we have RT now?) or a combination of all of that? Alternatively, hardware-accelerated RT is already computationally intensive even with dedicated hardware ... does it become prohibitively expensive to use some of these lighting techniques (real world requires multiple lights and/or bounce lights and/or light occluders and/or light occluders combined with bounce lights, etc.)? Non-RT lighting gives some of these effects basically for free.

Regardless, it goes without saying that properly implemented RT will have far more realistic lighting than non-RT lighting, but games advertising RT lighting still have a ways to go to be as interesting and egaging as some of the best games in the past few decades which have had to lean heavily into using good cinematographics lighting techniques to either hide or take advantage of the limitations of "faked" lighting. After all good cinematography in films, TV and advertisements is all about fake lighting behaving realistically.

I'm not sure if this will engender enough interest for people to talk about the actual uses of lighting in games versus it devolving into RT = good, non-RT = bad. It also makes me wish some peole like LaaYosh (I think that was their handle) were still around to talk about stuff like this.

Regards,
SB

 

[homerdog]

It'd be a waste of time for artists to implement cinematic lighting techniques using optional features like RT that most users can't (for performance/compatibility reasons) or won't (for performance reasons) enable. Maybe the next console gen will change things. Once features like path tracing are included in the baseline console experience I hope we'll see more thought go into how it's used. But this has been the case for many rendering techniques through the years and is not unique to RT. Though the period of optional/partial adoption of RT seems to be dragging on and on compared to stuff like pixel shaders in the DX8/9 era.

 

[DavidGraham]

Static baked lighting is no longer sufficient for the desired next gen look, it exhibits glaring issues for dynamic objects/characters (as they pass through environments not being affected by light/shadows), corners and distant areas lack indirect light/shadows .. etc, the result is a look that lacks depth and is less convincing. Metro Exodus is a great example of this, in it's baked static version, most of the semi open world areas lack depth as a result, as the baked lighting fails to take into account areas that are lit indirectly.

Also baked lighting often comes with the shortfall of ancient techniques for reflections, as reflections disappear for off screen objects, or lack dynamism entirely, whereas dynamic lighting incorporates modern reflections technique and use them to augment the lighting of the scene for a true cinematic look.

You can absolutely use stylized lighting techniques with ray tracing, either through post processing or through selective tracing for certain lights/shadows (you can turn RT on and off for or reduce the range of certain lights/objects). The Matrix demo is a great example of this, heavily stylized, uses RT for lights and reflections, and selectively uses RT shadows for cut scenes on characters. In the end it's one of the best "cinematic" next gen presentations ever.

In the end, it also comes down to personal taste, some people simply don't care about shadows (yourself included), and can run games just fine without them, others are more attentive to small details and get annoyed by the lack of depth and dynamism when using baked lighting.

 

[trinibwoy]

All things equal artists should have an easier time implementing cinematographic lighting using more flexible and accurate lighting tools. I think homerdog is correct in that artists won’t spend an enormous amount of time on the “art” side of things unless the majority of end users can enjoy it. There really shouldn’t be any debate on the tools side of things.

On the artistic stuff though some lighting tricks are simply harder to pull off in games because the end user has control of the camera. This is true for any lighting technique.

More precise lighting (I.e. RT) doesn’t force you to have realistic or boring lighting. You still control where to cast rays and how to shade what those rays hit. Just need the right engine and tools to expose those controls in an intuitive way to artists.

 

[trinibwoy]

some people simply don't care about shadows (yourself included), and can run games just fine without them, others are more attentive to small details and get annoyed by the lack of depth and dynamism when using baked lighting.

Shadows as a discrete graphics effect will hopefully soon be a relic of the past. The only reason we talk about them separately today is that they’re rendered independently of lighting and only for a handful of lights. Shadows and AO should just be a natural side effect of lighting/GI just like they are in the real world. Hopefully it won’t be possible to turn off shadows in future games.

 

[Dictator]

Shadows as a discrete graphics effect will hopefully soon be a relic of the past. The only reason we talk about them separately today is that they’re rendered independently of lighting and only for a handful of lights. Shadows and AO should just be a natural side effect of lighting/GI just like they are in the real world. Hopefully it won’t be possible to turn off shadows in future games.

I like this view.

 

[JoeJ]

Shadows and AO should just be a natural side effect of lighting/GI just like they are in the real world. Hopefully it won’t be possible to turn off shadows in future games.

The only technique which resolves shadows and GI at once is path tracing, which in it's classic form isn't efficient.
For efficiency, we must use some form of radiance cache, which then is usually at a lower spatial resolution good enough for indirect lighting, but not good enough for HQ direct lighting including its shadows.
However, we can use the cache for all lighting if necessary, accepting blurry shadows. So turning shadow maps or RT shadows off only affects quality of shadows, not necessarily their existence. (May be important to scale down to portable / mobile)
But for high quality results at optimal performance, specific techniques or passes for direct lighting, shadows, and primary reflections won't go away, i'm pretty sure.

I like this view.

I'd like a Ferrari, but this does not make it affordable. : )
Epics VSM helped me a lot to accept there's still a need for those damn shadow maps. Living in the shadow of Nanite and Lumen, the feature would deserve more attention.

 

[trinibwoy]

I believe it can be a good technical discussion about "how lighting is used in games" regardless of the technology behind the lighting used in any specific game.

For the record I’m struggling to grasp how to have a technical discussion about lighting without contrasting lighting technologies Seems that would then be a discussion on art style and art techniques which gets into some really subjective stuff and is likely different based on game genre and individual tastes.

But any discussion on artistic lighting should start with the artist’s tools. The more elegant and flexible the toolset the easier it will be for artists to flex their creative muscles. Would be great to get some insight from actual artists.

But for high quality results at optimal performance, specific techniques or passes for direct lighting, shadows, and primary reflections won't go away, i'm pretty sure.

Direct lighting + direct shadows are already doable in one pass. Same for indirect lighting + indirect shadows. With a bunch of denoising of course. We don’t need to go all the way to classic path tracing.

 

[JoeJ]

Direct lighting + direct shadows are already doable in one pass. Same for indirect lighting + indirect shadows.

Ah ok. Thought you meant doing it all at once.

So, this is why when looking at something with what I consider one of the better implmenetations of global RT illumination (Metro: Exodus), I also view it poorly with regards to lighting in games. It implements global illumination really well, but then fails to use any cinematographic lighting techniques to make the actual visuals engaging and enjoyable to look at. IE - people have previously heard me talk about how I much prefer the non-RT lighting of the Metro: Exodus to the enhanced RT lighting in Metro: Exodus EE.

I remember they talked about manual work for tweaking lighting to look good, which then became redundant with the move to RT.
Likely, by removing the tweaking, they also lost some benefits from that.

I'm no lighting artist, but for artistic reasons we surely want to tweak lighting a lot, and the move to more accurate and realistic tech does not change this. There's still a need for hero lights in cutscenes, fake lights to illuminate too dark spots, rim lights, etc.
But different techniques give different results, so we'd need to tweak twice if there is support of two different techniques like a RT / non RT branch. Twice the work will ofc. reduce quality at least a bit in practice.
But there is also an advantage in games. We could use huge area lights also in outdoor scenes if desired. For movies this means much more extra work and costs, and often they accept bad lighting conditions in practice to get it done.
We also can do better grading and other post processing due to higher precision and no noise.
But well, ofc. real world still looks much better overall

 

[Shifty Geezer]

For the record I’m struggling to grasp how to have a technical discussion about lighting without contrasting lighting technologies Seems that would then be a discussion on art style and art techniques which gets into some really subjective stuff and is likely different based on game genre and individual tastes.

Yeah, can we have clarification from the OP please? At this point I'm thinking the thread needs moving to Industry or Games. It's certainly possible to have a tech talk, such as "how does Nintendo add all its rim lighting to all its games?" but that's not Silent_Buddha's ask.

As for RT producing results, if it's accomplishable in real life, it's accomplishable, more so, in RT. It's what offline CG has done for decades. We can create fake invisible light sources, shadowless lights, negative lights, and all sorts. RTRT would free up the artists. The reason we don't have that likely at the moment is because RTRT can barely cope with the general lighting requirements of rendering a game. Once you have a robust GI engine, creative lighting will become easy and incredibly versatile.

 

[Davros]

shadowless lights, negative lights,

Confused !!!
Lights don't really produce shadows objects do (do you mean transparent objects in front of the light?) and what is negative light ?

 

[Frenetic Pony]

More precise lighting (I.e. RT) doesn’t force you to have realistic or boring lighting. You still control where to cast rays and how to shade what those rays hit. Just need the right engine and tools to expose those controls in an intuitive way to artists.

Yea, pathtraced animated movies have been around for over a decade (Monsters University) , and still cheat to hell and back for both lighting and camera placement.

Confused !!!
Lights don't really produce shadows objects do (do you mean transparent objects in front of the light?) and what is negative light ?

Light make brighter, anti light make darker.

 

[Silent_Buddha]

For the record I’m struggling to grasp how to have a technical discussion about lighting without contrasting lighting technologies Seems that would then be a discussion on art style and art techniques which gets into some really subjective stuff and is likely different based on game genre and individual tastes.

I'm not saying the use of lighting should be divested from the lighting tech. used to accomplish the lighting effect (hero lights, face lights, backlighting combined with foreground lights, natural lighting combined with overhead lights, multiple light sources with some lights only used to cancel the shadows caused by other lights, etc.). What I'm saying is that the lighting tech. isn't the "be all, end all" of lighting. A game that is fully lit using global RT illumination can and does sometimes (key, as I'm not implying this happens all the time or even potentially most of the time) look worse and/or less interesting than games that don't use any hardware accelerated lighting. It's how the lighting is used that makes for an interesting and engaging visual presentation.

As for RT producing results, if it's accomplishable in real life, it's accomplishable, more so, in RT. It's what offline CG has done for decades. We can create fake invisible light sources, shadowless lights, negative lights, and all sorts. RTRT would free up the artists. The reason we don't have that likely at the moment is because RTRT can barely cope with the general lighting requirements of rendering a game. Once you have a robust GI engine, creative lighting will become easy and incredibly versatile.

This is true, and I'm by all means not claiming otherwise. However, I believe that for games, it's, IMO, more difficult to accomplish this than with tradition media (films, advertisements, TV, etc.). As someone mentioned before, in traditional media the camera is meticulously controlled and directed by the cinematographer and thus it's easier to place all the numerous light sources, occluders, light bouncers, etc. without them interfering with the camera or interfering with each other in undesirable ways as can happen if someone other than the director had free control of the camera.

In other words, if someone were given free rein of the camera in the specific sets/scenes used for a live films or even say a Pixar film the end result would likely be lighting that looks irregular versus all the lighting tricks coming together to present a coherent image/scene as the director/cinematographer intended. Even in a CG film, you'd want some light sources to follow the camera, some to remain stationary and some to potentially move independently of the camera (perhaps set to follow another character). Now, how to make that all happen with a user controllable camera without making it really obvious that there are other light sources that have no real world analogue (IE - no real world or in scene light source)?

For non-RT traditional lighting tricks it easy as they aren't actually emitting lights/rays and some don't even have a "source light". How to accomplish this with RT lights?

So, some things become easier with RT versus lighting "hacks". But you also lose some of the benefits of those more limited lighting models. For example, a point light only lighting a characters face but not the surroundings in a game isn't realistic but it achieves a cinematographic effect that is much desired in films and often requires multiple lights combined with shadow canceling lights to achieve. But in tech (games) its rather limited nature basically gives it for free.

This reminds me of how some people think say, the latest Forza or Halo: Infinite look drab/uninteresting/unimpressive at times. It's a side effect of global illumination more closely resembling real world lighting conditions. Cinematographers prefer to shoot as close to sunrise or evening as they can because the longer shadows gives a more interesting, engaging and interesting visual presentation. Simultaneously they will attempt to avoid filming when the sun is directly overhead or close to it as then shadows are less visible/smaller and leads to a less interesting/less appealing/less engaging visual presentation. Those games look interesting at certain times of the day and at other times of the day look worse than the older games.

In other words, many discussions lately focus too much on tech and many people assume that better tech = better visuals. What's important and likely even more so as RT lighting becomes more commonplace is how lighting is used and what tricks are used to make the RT lighting less realistic while still behaving realistically.

Right now as I see it, we have the old way where good games try to make unrealistic lighting look as realistic as they can and they often get cinematically interesting lighting for free. On the other hand we have developers dipping their toes into a more "correct" representation of how light works by using RT, but not many developers are taking the time to then make the lighting interesting or "look good". As such the impact is often lessened or non-existent beyond, hey this RT lighting is behaving more realistically but then you have people commenting that it doesn't "look good". As cinematographers have learned over the past hundred+ years, realistic lighting isn't engaging or interesting ... hence almost no movie has natural realistic lighting but instead use lighting tricks to make unrealistic lighting behave realistically.

Confused !!!
Lights don't really produce shadows objects do (do you mean transparent objects in front of the light?) and what is negative light ?

Negative light is probably refering to the use of shadow canceling lights used by cinematographers because good lighting requires multiple, often VERY strong lights. Those lights would lead to multiple distinct shadows that ruins the illusion of light behaving realistically in a scene if they weren't counteracted by addition lights placed specifically to "erase" or cancel those shadows.

The goal is to have multiple strategically placed lights to emphasize mood, tension, feeling, dialog, etc. while still having it not appear "wrong" to the viewer. Without those additional lights films would appear flat and lose much of their intended emotional impact.

Regards,
SB

 

[Davros]

So it's not a light that produces darkness then as was suggested ?

As for RT producing results, if it's accomplishable in real life, it's accomplishable, more so, in RT

While it's true that any behaviour can be programmed in I certainly believe I know of a behaviour (possibly several) that no RT implementation models
Talk me through what happens when you reduce the intensity of a light in RT
If your going to say the light gets dimmer then your only partially correct. It will only get dimmer to a point until you reach a single photon then what happens is the light will flash and the more you reduce the intensity the the longer the duration between flashes.
That end's todays lesson in quantum mechanics....
Tomorrows lesson : How the example of Schrödinger's Cat was used to create the enemies in Dead Island...

 

[Shifty Geezer]

So it's not a light that produces darkness then as was suggested ?

Yes, it was. Once you have a maths model, you can screw with it however you want. RT can trace a ray to a lightsource of brightness 5.0 and calculate the amount of the light added to a surface. So then set the lightsource to brightness -5.0, and subtract light from the surface!

Back when I made Realsoft 3D art with its programmable RT pipeline, I could do all sorts of things. It's just maths - doesn't need to follow the laws of physics.

Now you need a versatile model for that. Current RT models are trying to match reality and so might not play well with concepts beyond the scope of that remit. The energy conservation models might be including various assumptions in the RT algorithm such as light sources can only have positive values. R3D brute-force-traced rays, a basic Monte Carlo algorithm, where you could add and manipulate any value 'passed' along a ray. You could access every step of the tracing - ray intersect, surface evaluation, secondary ray cast - and add whatever shader you wanted at that step, manipulating a primary 'illumination' value that was resolved down to the final pixel value. You could add something like "On ray termination, multiply surface illumination by 1.2 times recursion depth" and make surfaces brighter based on more recursions.

There's nothing actually stopping that being added to a game if you wanted (other than potential workload slowing everything down too much).

While it's true that any behaviour can be programmed in I certainly believe I know of a behaviour (possibly several) that no RT implementation models
Talk me through what happens when you reduce the intensity of a light in RT
If your going to say the light gets dimmer then your only partially correct. It will only get dimmer to a point until you reach a single photon then what happens is the light will flash and the more you reduce the intensity the the longer the duration between flashes.
That end's todays lesson in quantum mechanics....

To be fair, someone could do that. The problem with RTRT is approximating bazillions of photons. Tracing only one at a time is a cinch.

Negative light is probably refering to the use of shadow canceling lights used by cinematographers because good lighting requires multiple, often VERY strong lights. Those lights would lead to multiple distinct shadows that ruins the illusion of light behaving realistically in a scene if they weren't counteracted by addition lights placed specifically to "erase" or cancel those shadows.

The goal is to have multiple strategically placed lights to emphasize mood, tension, feeling, dialog, etc. while still having it not appear "wrong" to the viewer. Without those additional lights films would appear flat and lose much of their intended emotional impact.

That's what we have to do in real life. In a computer model, we can very easily have lights that don't cast shadows (in Realsoft objects had a checkbox for "cast shadows" and "receive shadows" just like a standard game engine). And we can have 'lights' that have a negative energy and remove photon energy from the system - negative lights that shine darkness. As an artist, if you want a corner of a room to be dark and foreboding but the GI model makes it very normally room-lit from the window, you could set up an antilight and suck light out of that corner (metaphorically 'suck', you'd add negative light energy in tracing a source).

 

[Silent_Buddha]

That's what we have to do in real life. In a computer model, we can very easily have lights that don't cast shadows (in Realsoft objects had a checkbox for "cast shadows" and "receive shadows" just like a standard game engine). And we can have 'lights' that have a negative energy and remove photon energy from the system - negative lights that shine darkness. As an artist, if you want a corner of a room to be dark and foreboding but the GI model makes it very normally room-lit from the window, you could set up an antilight and suck light out of that corner (metaphorically 'suck', you'd add negative light energy in tracing a source).

That is good to hear (read? ). So, then it's possibly likely the case that it's either that, for budget reasons, games implementing RT just implement RT and call it a day without going through the extra work of having a lighting artist actually go through and manipulate the lighting to make the lighting look good, impactful, emotional, clear (for dialog), etc. Instead relying on the "wow" factor of RT and light behaving more realistically and calling it a day to save money. Or alternatively, maybe the cost in terms of rendering time to actually do all that extra lighting (or light removal ) work in hardware is too cost prohibitive (for a realtime game) for the lighting artists to do much beyond just implementing RT lighting.

Anyway, the main purpose for this thread, or at least my hopes for it was to move the discussion on lighting beyond just hardware RT = good/non-hardware RT = bad as I've yet to see many, if any instances where the RT lighting is used to really good effect other than to present a more coherent lighting model. The potential it represents is obviously leaps and bounds beyond traditional lighting "hacks", but I've yet to see any developer really tap into that.

Thus, it'd be interesting, IMO, to see people talking about these things. I understand this goes a bit beyond just the tech, hence, not being sure if this was the correct sub-forum. As well, it may be that we just don't have people remaining that can really go in depth about this (lighting artists or engine architects, for example). For me, I'm extremely curious about why despite taking a massive leap forward in terms of lighting tech, we've also taken, IMO, 2 steps backwards WRT how lighting is used in games. Budget ... time, money or hardware cost? A mistaken belief by game producers/developers that once you have RT lighting you don't need proper lighting artists? Something else?

And I know that I'm not alone in seeing this. As we get numerous people commenting how X game with RT doesn't look as good as older game without RT and then mistakenly, IMO, attributing this to a bad RT implementation. The reality is likely closer to being that the RT in the new game is much more correct, but as less effort was put in to using the cinematographic lighting techniques that it also ends up looking worse despite having better lighting ... as realistic natural lighting in media isn't terribly interesting. As an example, the new Forza has obviously better and more correct lighting, but to many people it looks worse than the previous Forza despite having better and more accurate lighting.

Regards,
SB

 

[Shifty Geezer]

That is good to hear (read? ). So, then it's possibly likely the case that it's either that, for budget reasons, games implementing RT just implement RT and call it a day without going through the extra work of having a lighting artist actually go through and manipulate the lighting to make the lighting look good, impactful, emotional, clear (for dialog), etc. Instead relying on the "wow" factor of RT and light behaving more realistically and calling it a day to save money. Or alternatively, maybe the cost in terms of rendering time to actually do all that extra lighting (or light removal ) work in hardware is too cost prohibitive (for a realtime game) for the lighting artists to do much beyond just implementing RT lighting.

I think the second. Firstly, I'm not sure the current acceleration structures to hit realtime RT can accommodate novel techniques. For example, to skip an object from casting shadows you'd need to trace against objects and then reject those marked as 'no shadows'. That means finding a hit and then rejecting it and having to pick up the ray again. Or casting a new ray from the object and pass the results back into the previous ray. I don't know how programmable the pipelines are, but to get maximum speed I expect a lot of assumptions will be made. You'd also be adding a lot of additional work, in the same way glass is costly. Secondary rays increase work exponentially over just sampling opaque surfaces. I dare say technologically it can be achieved but not in realtime, and devs are more focussed on getting decent lighting working at a decent speed than pushing the engine to create nicer cinematography at single digit framerates. Then they'd have to simulate existing real-world techniques and as you've rightly pointed out, that's in an environment where they can't control the camera, so can't really have a large reflected off-camera bouncing fill light onto a face when the players can move the POV and see the guy standing there holding a large piece of cardboard...

Anyway, the main purpose for this thread, or at least my hopes for it was to move the discussion on lighting beyond just hardware RT = good/non-hardware RT = bad as I've yet to see many, if any instances where the RT lighting is used to really good effect other than to present a more coherent lighting model. The potential it represents is obviously leaps and bounds beyond traditional lighting "hacks", but I've yet to see any developer really tap into that.

I'd argue that an issue here is looking at a subset of games. There are lots of games out there with lots of render styles, many which don't care for GI. The moment you go for realistic like a racer, you are limited to certain lighting schemes for the already mentioned reasons.

 

[digitalwanderer]

I don't think the method really matters to the end user, just the results. I recently got an nVidia card and finally got to check out RT first hand and it is gorgeous and much more realistic, but if it was the only way to do lighting in a game I wouldn't be able to game. The performance impact is just too great and I ain't got the budget to get hardware that could do it.

We've had beautifully lit games before RT and we will after. It isn't perfect for every application right now or every person, but it is pretty and will continue to grow as hardware and games adapt more. Game designers should use what they can to achieve what they want, the end product's quality is all that really matters.

 

[Shifty Geezer]

We've had beautifully lit games before RT and we will after. It isn't perfect for every application right now or every person, but it is pretty and will continue to grow as hardware and games adapt more. Game designers should use what they can to achieve what they want, the end product's quality is all that really matters.

There's a cost and convenience benefit to using a GI checkbox feature. Your game automatically looks good, if not artistic. Whereas using other methods to make something that's good and artistic is a lot more work.

 

[DavidGraham]

Yea, pathtraced animated movies have been around for over a decade (Monsters University) , and still cheat to hell and back for both lighting and camera placement.

Which is a distinction that seems to have slipped @Silent_Buddha's mind, all the cinematic CGI shots are made through path tracing.

So, some things become easier with RT versus lighting "hacks". But you also lose some of the benefits of those more limited lighting models.

You can mix and match the types of light in the same scene. Use RT lights to obtain dynamism, and use Raster lights for cinematic shots.

Many games already do this, Alan Wake 2 uses hardware Path Tracing but uses many raster lights for cinematic shots, it used baked light maps combined with a limited form of RT global illumination to achieve certain looks.

Avatar does that too, it uses hardware ray tracing for global illumination and reflections with a software fallback. Unreal 5 titles uses Lumen which is a software or hardware ray tracing solution for global illumination and reflections, with as much options to use raster lights as you want. You mix and match to your heart content.

Dying Light 2 uses heavy post processing and certain raster lights to achieve it's cinematic look while using hardware ray tracing for global illumination and reflections. The baked version of Dying Light 2 looks decidedly less cinematic or next gen compared to the RT version.

The Witcher 3 is another good example of this, it uses hardware ray tracing for shadows, lights, global illumination and reflections, yet you can easily mod the game to tamper with everything, remove RT from certain lights, expand RT to more lights, particles, .. etc. You can increase or decrease the range of RT global illumination and reflections, add more RT shadows for lights and objects, or remove shadows, you can increase the range of RT shadows to cover more draw distance or reduce the range.

You can do the same for Cyberpunk Path Tracing/Ray Tracing, you can disable RT shadows/reflections/global illumination at will.

Also, excluding RT lights from casting shadows is already done on current games, you can exclude entire objects from the acceleration structure to prevent them from appearing in reflections, cast shadows or be affected by light.

You can ignore shadows all together for all objects, disable shadows/reflections/indirect illumination for entire areas, and enable them for other areas .. etc.

None of this is new or hard to get it working, we already have it.