Sulphur on 15/5/2020 at 17:09
The Team Bondi example may not be the best one. Their face scanning tech was really odd because while that bank of cameras capturing facial motion and detail was really cool, it also needed people to be stuck in position for the capture to work, which meant they couldn't move while talking, so they had to stick that performance on top of separate full-body mocap. And the end result wasn't even that good! It looked like they'd rotoscoped faces onto blotchy pudding on top of a stiff polygonal model. Compare that with what Ninja Theory did for Hellblade with its single camera motion capture (tech they made specifically to work with the low budget they had), and it's no contest which one looks better. All that expensive tech for LA Noire, when other methods would have worked out just fine or better.
Star Citizen is also not a great example because it's the gold standard for feature creep and monetisation practices instead of just making the game and releasing it. Anything in the top-line games bracket that takes more than 3-4 years to make is likely suffering some kind of development hell in the current AAA environment. 9+ years is not the standard (SC is 8 as of now I reckon).
That said, I don't disagree with the overall point. There's better things to be done that recruiting and tasking a 500-strong or more art department with filling out a game's graphics - I'd trade every single Assassin's Creed ever made for a simple-looking Thief/Hitman clone with advanced AI and cutting edge sound design.
The other questions you have are iterations of the same ones we've had at least as far as 15 years ago. It's always going to be a reality of the commerce of gaming, and the balance depends on what people are going to do to make these high fidelity recreations of reality easier and cheaper to create. That's part of why UE5 is doing what it does; and dema's vision for procedural art isn't that far into the future. Banks of hand-crafted+procedural assets massaged into whatever game being made is the immediate future.
Starker on 15/5/2020 at 17:27
Quote Posted by Sulphur
Epic mentioned that they didn't really optimise it hard or target a high resolution, so that demo's at 30 FPS and (a possibly dynamic resolution) 1440p. And sure, not many AAA games are going to shove in that kind of asset detail in there, but if there's one thing we do know, it's that they'll find a way to use available space anyway.
Was lack of optimisation really the reason that forced them to have the demo at that resolution and FPS, though? Or was it that they simply hit the ceiling of what the PS5 is capable of under this kind of graphical load?
Quote Posted by Sulphur
The Team Bondi example may not be the best one.
Star Citizen is also not a great example
Sure, I didn't really mention these examples because they are typical of game development, but rather as the extreme where the pursuit of technology can push game development into.
Quote Posted by Sulphur
The other questions you have are iterations of the same ones we've had at least as far as 15 years ago. It's always going to be a reality of the commerce of gaming, and the balance depends on what people are going to do to make these high fidelity recreations of reality easier and cheaper to create. That's part of why UE5 is doing what it does; and dema's vision for procedural art isn't that far into the future. Banks of hand-crafted+procedural assets massaged into whatever game being made
is the immediate future.
And yes, my concerns don't really have much to do with UE5 in particular as much as with the general push for realism and quality having a negative impact on the medium.
Sulphur on 15/5/2020 at 17:28
Quote Posted by Starker
Was lack of optimisation really the reason that forced them to have the demo at that resolution and FPS, though? Or was it that they simply hit the ceiling of what the PS5 is capable of under this kind of graphical load?
Honestly, we don't know what kind of graphical load there was, as barring a second or two where you could see frame rate dips, it was a seamless experience. In comparison, the last real-time demo from Epic for PS4/UE4 (Agni) had a lot of places where the frame rate jumped up and down and just kind of struggled, but it's been bested both in performance and fidelity by the PS4's most technologically capable games like God of War. This means that there's probably headway for things to look
better on the PS5 relative to the demo's graphics.
Nanite scaling both up and down the hardware power ladder means we have no idea how expensive it is coupled with Lumen - though the fact that there wasn't any RTX in play during the demo is telling.
Starker on 15/5/2020 at 17:35
Yeah, the lack of RTX is part of why I'm suspecting they hit a ceiling. And at the end of the day, it's a tech demo. You don't really know how feasible this is in a full-scale game under the hardware limitations until you have tried it.
Sulphur on 15/5/2020 at 17:53
I'm not seeing anything that disproves they had a limited budget relative to AAA, or that says they abused their resources? They obviously have years of expertise and talent, but that's never going to be a guarantee that your game will sell well.
Starker on 15/5/2020 at 18:05
Huh? But that's not what anyone has claimed or what the article is trying to say, though? It is saying that what was accomplished with Hellblade was in large part due to their unique circumstances and therefore serves more as an example of an exception to the rule.
Sulphur on 15/5/2020 at 18:24
My citing of Hellblade was directly in relation to the Team Bondi facial tech example. There were different, smarter ways of going about it that didn't need to blow a hole in the budget. Larger systemic issues with the industry aren't going to be proved or disproved by taking only a few examples either way.
Starker on 15/5/2020 at 19:28
Yes, I get it. They are different teams doing different things at a different time and Team Bondi doing what they did in the 2000s is not comparable to what Ninja Theory was doing in the 2010s. I understand all that. My point, though, is that Team Bondi's LA Noire is an example that proves the trend of budgets and teams getting more and more bloated and development becoming more and more expensive whereas Ninja Theory's Hellblade is an anomaly in the large scale of things.
But sure, Hellblade was an amazing game done by exceptional people with a razor-sharp focus and very likely at a much, much lower price than a big AAA game (though likely also not making as much profit). You do have to wonder, though, if they were making a game of LA Noire's scale and ambition, how much less would it have really cost?
Gryzemuis on 15/5/2020 at 20:27
Right now we see a jump in engine-technology. Your remark (as I understand it) is: what use is new engine-technology when it is way too expensive to develop content for that engine ? Well, the answer is (I think): better tools.
In my own area of expertise, 20-30 years ago, the big challenge was to create new technology, and make it scale. Nowadays, for the last 10 years, nobody really cares about newer or better technology. Yeah, of course they want faster and they want cheaper. But the issue that is hot right now is: how do I manage this technology ? So all the focus over the last 10 years has been on tools: how do I configure my stuff, how do I troubleshoot my stuff, how do I manage software versions, etc.
I think gaming will go through similar cycles. We get new engine-technology. But it's impossible to use on a large scale. Too expensive. Then someone (or some companies) come up with new technology to create content for those engines. Easier and cheaper ways to do motion capture. Easier and cheaper ways to create a 3D environment from simple pictures or movies of existing things. Right now nobody is encouraged to create those (relatively expensive) tools. Because creating content by hand is cheaper. But once creating content by hand gets too expensive, people will be encouraged to develop those tools.
Same could be true of the content too. Not just the tools. If you need a picture of a cow, do you drive to the country and take a picture yourself ? Nope, you buy a picture from a specialized website. Same could be done for game content. What Sulphur calls "a bank of hand-crafted+procedural assets". I think this is not common yet, because it's still cheaper to produce your own models/textures/meshes. But once such banks become cost effective, they will lower the cost of producing games.
I do hope that by that time, studios will have rediscovered that "making games that are fun" can make them good profits too. And they will focus less on milking their games for every penny they can, and taking minimal risks. But that's a whole different issue.
