hedwards: You have your doubts, but I don't see anything in your post that would point that way. The processing requirements of raytracing are primarily about the number of rays that have to be processed per frame. The complexity of the environment does matter somewhat, but not nearly to the degree you're suggesting.
The tech demo we're seeing is good enough for the majority of games where you'd want to be ray tracing. It may not be good enough for something like Crisis' jungle scenes, but for the types of 3d games that people typically play, this is already more than good enough.
Also, as far as framerates go, this kind of technology is much easier to target specific framerates than the current method of rendering scenes. You can scale the number of rays up and down to a significant degree without having it be as obvious that you're doing it.
As far as those render farms go, yes they exist, but they're also rendering higher resolution and more frames. They're also doing it for a medium where the only thing that matters is what you see, very different from games where you can vary the number of rays in a scene if you need to. Plus, they're often rendering a much finer level of detail, things like individual hairs that are not even remotely necessary for games when this first showing up in games. Games have been varying detail levels for years to help get through areas where the hardware can't quite keep up.
That's the thing there. Compared with 1st generation 3d graphics, 1st generation raytracing is likely to look good for decades after the point where people stop working on the games. The main difference at this point between what we're seeing and what we could see is in the models and the number of rays being used. The models are and will continue to be the thing holding realism back for the foreseeable future as raytracing is a relatively simple thing to do, just very resource intensive.
I'd be very surprised if those first games don't include some method of increasing the number of rays for when future hardware can handle more as it's unlikely to require much effort to include that. Sort of like how some older games had graphics settings that weren't practical at the time for many of the customers.
I could see that, I would not want to screw around with something like that until it had been extremely well tested, but I'd be surprised if it doesn't eventually happen.
Judicat0r: I actually wrote that the complexity of the raytracing is infinite: I don't know how to state it in an other way.
You overlook a couple of things: yes the number of rays is one the most heavy variables in the equation but you don't take into account the number of steps which, visually, and computationally is very impactful.
That, clearly, impacts fthe framerates as well, and if you need to tune the number of rays and steps until it affects badly the visual result then wat's the point of all of that? Have a shiny writing on a splash screen or on a GPU box that reads RAYTRACING?
Geometry complexity usually isn't heavy but once you factor in displacement and tesseletion I can assure you that it becomes a problem AND geometry detail is not going to decrease year on year, it's the contrary.
Also light sources, you have to keep in miind that if you want to achieve decent results you are going to need a lots of them.
In render farms it doesn't work like that: if they can output a clip in a week today, it will output the same clip in a day 5 years from now, clearly, but the demand for power will increase accordingly as the technology moves on: in five years render farms will produce professional-grade outputs in times that are comparable as today's maybe even longer.
Don't think that they render at much higher resolution than what's possible today where you can easily reach south of 50M pixels with just 3 displays.
And "much finer level of detail, things like individual hairs" factor in importantly apparently so geometry complexity is a thing then. ;)
All the above is just to address your points but you need, if you like, to understand that what we are shown is just marketing: raytracing is not viable unless it is a half assed form of it.
The tech demo features a kind of rendering algorithm based on ray casting that a involves a great deal of approximation in the form of multi samplig to achieve decent results, it's no raytracing, simple as that.
I repeat myself three years for real time raytracing?
I'm doubtful.
Edit: fixed spelling.
These are things that mostly apply to movies and TV, not video games as there are other trade offs to be had. Rendering a video game in real time via ray tracing is a much easier problem than you're letting on because you get to control the level of detail shown tot he player in a way that you can't with other media.
You're not likely to have the budget for the huge numbers of artists needed to generate the assets necessary to get that level of complexity. Even with AA games, you don't see that level of detail because of the costs and storage space associated with such levels of detail. Not to mention the increased cognitive load on the player that has to decide what to pay attention to. That's not going to change just because we start using raytracing.
Granted, you don't quite get the option of doing it via distance the way that raster engines do, but the situations you're using just won't apply to 1st gen raytracing games. And may never apply as those are mostly things that distract the player from playing the game.
As far as the sampling goes, that doesn't make it not raytracing and it's likely a lot easier to remove that than it is to get the raytracing engine going in the first place.
As far as geometric complexity goes, I never said that wasn't an issue, just that it's not something that's likely to be an issue in the first gen raytraced games. They'll just design the games around the constraint the same way they always have. They get to choose what the geometry is like and you're not likely to see them adding realistic hair like that until they have that detail solved.
In short, these seem like overly picky standards and an attempt at rationalizing why it's going to take such a long time. In practice, I doubt very much that once this demo is released that it won't be good enough for typical players, it certainly looks better than what we currently have.
DreamedArtist: I own an RTX 2080ti but tbh I do not like the look of the game and how they got rid of the griddy look of Quake. but hey I will give it a shot and see how it looks at 4k and post some screens of it here.
it is legit a PR stunt to get people to adopt RTX with a free download of there childhood classic. I seen this all before but hey there are people who will take this.
exorio: Still looks fugly and won't justify RTX purchase lol
Okay probably the ability to pull 4k+ resolution, but there're sooo many ports that able to do 4k by now.
Why not give a way a real RTX-only games like back in the days, where you get all those awesome CDs bundled with the cards.
You make it sound like it's fuglier than the game was originally. This is a tech demo as in they likely turned the effects up to make it clear to the viewer what the changes were. I'd personally wait until the actual demo is released before making a judgment call as it's pretty clear that they tuned it to make the effects as obvious as possible.
I could be wrong about that, but I'd expect that they'll tone it down a bit to make the experience a bit more consistent with what they're selling.