Support NeoGAF

[mckmas8808]

"THEY" are trying to silence me. Stop them by contributing just $5/month to my patreon!!

Fucking outrage culture man

No way he says this. Please tell me he doesn't say this......

 

[Sethbacca]

...no one in the world would accept sub par products and service in a retail store for example, the store would go out of buisness.

Harbor Freight, Walmart, TEMU, Ali Express, etc etc etc. There is a large market for people that want good enough knockoff junk that will last just long enough at dirt cheap prices, so it's 100% a business model that works.

That said, I'd rather not see it continued to apply to games.

 

[Thief1987]

If we moved back to a world in which every games studio needs to create their engine, often for a specific game, we'd probably have better looking and performing games overall. And a fraction of them, too!

Well, quality over quantity any day. And AAA devs with tens or even hundreds million budgets not having their own tech is an absolute embarrassment

 

[Herr Edgy]

None of which even fit the scene in question, at most walking out of the building, but with that size it should be possible to make the optimizations on the video.

Correct, which is why the whole video is unnecessary and gamer bait, as initially stated.
It's using a game dev setup for testing purposes, game devs deduce how Mega Lights can be applied, and gamers lap it up as 'performance improvement'.
The guy of the original video (not TI) posted a comparison with Mega Lights on vs Mega Lights off and saw massive perf increases with Mega Lights on.
The reason for this is because he wanted to test mega lights. The premise of mega lights is that it works with lighting setups that wouldn't have ever worked before. So he built a scene with tons of overlapping lights, turned on Mega Lights, and voila, it works!
Mega Lights was not developed to satisfy static lighting needs. Arguing about this is nonsensical.

Who said for a specific game? just make an in-house optimized engine, stick with it, and upgrade it. Deprecate it when necessary and make a new one when needed.

I'm sorry but this is the kind of ignorance I was talking about earlier. *Just do it like this* - it really is blissful isn't it.

I'm losing interest in this, I will respond if I feel there are points being made or questions to be answered

 

[rofif]

"THEY" are trying to silence me. Stop them by contributing just $5/month to my patreon!!

Fucking outrage culture man

If I had 5 minutes of fame on the internet, I would cash it in too.
5$ welcome everyone. bring it right here.

 

[Sethbacca]

No way he says this. Please tell me he doesn't say this......

They're trying to silence me and "i'm under confirmed censorship" : somewhere around the :45 second mark
"The people that support Threat Interactive will always know the truth" somewhere around the 11:00 mark.

Not an explicit call for Patreon, I was being more hyperbolic than anything, but basically "Only my followers will know the truth" type nonsense. The patreon thing I just threw in because it's so common these days .

 

[Herr Edgy]

Well, quality over quantity any day. And AAA devs with tens or even hundreds million budgets not having their own tech is an absolute embarrassment

Fair point of view, but keep in mind that this could also mean that some of the games you like or other people like would never be made. Also AAA devs with tens or hundreds of millions in budget gravitate to UE in recent years for a reason and maybe that reason eludes you.
Also these AAA devs absolutely do have their own tech, but using proven tools is a lot better than making everything from scratch.
However I wish Unreal's renderer was way more modular than it is, that would allow licensees to more easily write their own renderer implementations. Whether they make use of it or not is up to them of course, but it would make it easier. The renderer is one of the few parts that you really can't extend or change using plugins.

A good talk on rendering customization for stylized graphics which Unreal isn't good at by default

 

[Herr Edgy]

So, if the problem isn't Unreal Engine per se, but what is being done with it, then all that means is that the devs really do need to get their shit together. Which means that there needs to be more discussion and more videos on topics like this, so devs can make their games actually look and run well.

What I have seen in public discourse is that game devs will instinctively rush to defend their shit performance games, and then claim that anyone who complains or wonders why it is so horribly unoptimized (like the teeth in Cities Skyline 2) just doens't know anything. At some point you get tired of hearing it, and this circling the wagons isn't helping anyone, includding them, because when their shit running game gets released it is going to get destroyed.

Getting laid off sucks, but game devs are nto some perspecuted minority and too many of them online seem to carry this mindset and it's not helping.

I agree with this somewhat, there definitely are a bunch of toxic devs out there, I wouldn't put this on devs as a group though.
Also I wouldn't frame it as "they need to get their shit together" - game dev is more complex than it has ever been and what Epic is doing is attempting to reduce that complexity again. This benefits developers and in turn consumers.
Rendering tech is very complex and hard to understand for professional software engineers that are not rendering engineers. Add to that that rendering engineers are a rare species in general.
Rendering was not always complex, but it has become so complex that I don't think it's viable for small to mid-sized studios to have their own graphics experts. Keep in mind you also need more just one guy, the work spans many fields that you need many experts on at once.

The new tech is also bleeding edge, and projects on bleeding edge have to deal with bleeding edge problems. Lots of potential, lots of downfalls. The downfalls get alleviated over time but to get there you need to live on that edge first.
The Cities Skyline Teeth topic is interesting, good that you brought it up. I vaguely recall it and looked into it again. It seems the problem wasn't related to the teeth themselves but lack of LODs on characters in general.
This seems like a major problem that would be impossible to miss, but I wouldn't attribute it to developer's incompetence, or rather, it's hard to imagine that to be the problem. Seems more like a matter of suits & deadlines.
Also a good example for how this could have benefitted from Skeletal Mesh Nanite support that is in development. It might be better overall to have custom LODs for everything, but that is also a massive undertaking that requires resources, personnel, budget, time.
Would be great if all devs had that, but it would be even better if it wasn't required. Trust in the process.

it's interesting you mention Spiderman here, as in 2018 that game looked AMAZING on a base PS4 and ran great, on PS4 Pro it looked and ran even better, and on PS5 Pro, Spiderman 2 is an amazing looking game with great 60fps performance and ray tracing. So, it sounds like on some level Insomniac knows what they are doing. Whereas you see these UE games that run like garbage, that stutter on a 4090, and that quite frankly don't look nearly as good as Spiderman 2 on a base PS5 let alone a Pro. I have never heard anyone complain about Spiderman's graphics or performance.

The same convo is going on around now with Indy which leans heavily on RT and stuff and also runs very good. It, too, is not an Unreal Engine game. it seems like there is an advantage of being a dev with an in-house engine that knows how it works and how to optimize around it. As for 2016 looking games, I mean... Helldivers 2 runs on a literally discontinued engine, does actually look like a 2016 game in many ways... and is one of the most popular games of the year.

Insomniac definitely knows what they are doing! The reason I brought up Spiderman was non-specific, could have been any game with good graphics and sequel installments. The point was: would fans of a game accept a sequel with worse graphics but better image quality? Would this avoid a Threat Interactive guy? Probably not.

 

[Guilty_AI]

Correct, which is why the whole video is unnecessary and gamer bait, as initially stated.
It's using a game dev setup for testing purposes, game devs deduce how Mega Lights can be applied, and gamers lap it up as 'performance improvement'.
The guy of the original video (not TI) posted a comparison with Mega Lights on vs Mega Lights off and saw massive perf increases with Mega Lights on.
The reason for this is because he wanted to test mega lights. The premise of mega lights is that it works with lighting setups that wouldn't have ever worked before. So he built a scene with tons of overlapping lights, turned on Mega Lights, and voila, it works!
Mega Lights was not developed to satisfy static lighting needs. Arguing about this is nonsensical.

And the purpose of this video is to show there are much better ways to improve performance than turning on mega lights or nanite or whatever. And no matter how much you say they were developed for specific case scenarios, the truth is they are not only being improperly used, there's a myriad of other poor practices being employed alongside them. Thats the whole point of this and other videos of his.

 

[Herr Edgy]

And the purpose of this video is to show there are much better ways to improve performance than turning on mega lights or nanite or whatever. And no matter how much you say they were developed for specific case scenarios, the truth is they are not only being improperly used, there's a myriad of other poor practices being employed alongside them. Thats the whole point of this and other videos of his.

Yes and no, the purpose of the original video *not* to 'optimize by turning on Mega Lights' instead of using more simple alternatives. You have to understand that the lights placed were configured in a way that only makes sense with Mega Lights.
If Mega Lights wasn't used there would have never been a need to 'optimize' in the first place because nobody would place lights like this without it! It's a non issue! Newcomers to game dev would place lights like this before and wonder why perf is crippled, then google it, and find the answer "doesn't work, don't do it" and that's that.
Unreal literally displayed a red X over lights when there were more than 4 overlaps.

This is not news to anyone except newcomers.

I don't know what you are trying to argue. Can you clarify what your point is? Mega Lights bad because there are other means of doing things in certain scenarios? Welcome to game dev since forever, where your choices have consequences.

 

[Dane]

So TLDR basically is devs are overly reliant on the engine to be doing optimization that they should be doing themselves with lighting and meshes, and the engine is doing a piss poor job of optimizing on its own.

The same TLDR applies into why today's proprietary engines are more optimized than Unreal, you need to invest in infrastructure and have more passionate devs instead of relying on the pre existing factory optimization

 

[Guilty_AI]

Yes and no, the purpose of the original video *not* to 'optimize by turning on Mega Lights' instead of using more simple alternatives. You have to understand that the lights placed were configured in a way that only makes sense with Mega Lights.
If Mega Lights wasn't used there would have never been a need to 'optimize' in the first place because nobody would place lights like this without it! It's a non issue! Newcomers to game dev would place lights like this before and wonder why perf is crippled, then google it, and find the answer "doesn't work, don't do it" and that's that.
Unreal literally displayed a red X over lights when there were more than 4 overlaps.

This is not news to anyone except newcomers.

I don't know what you are trying to argue. Can you clarify what your point is? Mega Lights bad because there are other means of doing things in certain scenarios? Welcome to game dev since forever, where your choices have consequences.

What's bad is overreliance, often unnecessarily, on certain tech and excuse poor practices with it.

Before devs would follow the correct advice and "don't do it", or at least do it in a more efficient way that worked with their game. Now they "do it", turn on mega lights or whatever other thing, and get better performance than the absolute worst case scenario. Except that "better" performance is still awful, incomparably worse than if they just followed older methods.

And you can argue the new tech is technically more accurate, or works out-of-the-box in a bigger variety of different scenarios. But rare are the games where many of such scenarios even matter or happen at all, with these implementations offering marginal improvements for whats actually in the game for a huge performance cost, if any.

Like your torch example, does it even matter how accurately the light of the torch scatters and blends with the other lights in a fully illuminated hall? Vast majority of players wouldn't notice. What they will notice however is the frame rate dropping to a crawl due to the usage of some expensive and unnecessary tech.

 

[Sethbacca]

And you can argue the new tech is technically more accurate, or works out-of-the-box in a bigger variety of different scenarios. But rare are the games where many of such scenarios even matter or happen at all, with these implementations offering marginal improvements for whats actually in the game for a huge performance cost, if any.

I think that's the big takeaway here. You can turn everything to 11, but is a 60% performance decrease worth the 5% visual increase. All the devs are using the advanced features just to use it as a marketing point for their games, but the performance of the games is suffering as a result and basically tanks any benefits to marketing anyway.

 

[Herr Edgy]

What's bad is overreliance, often unnecessarily, on certain tech and excuse poor practices with it.

Before devs would follow the correct advice and "don't do it", or at least do it in a more efficient way that worked with their game. Now they "do it", turn on mega lights or whatever other thing, and get better performance than the absolute worst case scenario. Except that "better" performance is still awful, incomparably worse than if they just followed older methods.

And you can argue the new tech is technically more accurate, or works out-of-the-box in a bigger variety of different scenarios. But rare are the games where many of such scenarios even matter or happen at all, with these implementations offering marginal improvements for whats actually in the game for a huge performance cost, if any.

Like your torch example, does it even matter how accurately the light of the torch scatters and blends with the other lights in a fully illuminated hall? Vast majority of players wouldn't notice. What they will notice however is the frame rate dropping to a crawl due to the usage of some expensive and unnecessary tech.

But that scenario is fictitious. Devs do not "do it" if it gives them bad results. You only see this with newcomer devs and gamers that are blinded by trailers.
Serious devs don't use features if they don't need them and they give them bad performance.
There are barely any big games released that run on UE5 and the instances I know of, devs evaluate if it makes sense for them to use Nanite for example, and adjust accordingly.
I won't deny that there probably are games and devs out there abusing it against their best interest, but this again is up to developers, no one is being held at gunpoint to use Nanite when you have low poly meshes or Lumen Realtime GI when all you need is static lighting.
Give them a gun and they'd probably shoot themselves, too.
I don't see people happy-posting about UE5 with Infinity Nikki despite having beautiful graphics and running really well.

As for the bolded, I am not a fan of the photorealism trend either, I prefer stylized, clean visuals, and game design and storytelling are more important to me than graphics. However this is up to individual developers and no one else.

 

[diffusionx]

But that scenario is fictitious. Devs do not "do it" if it gives them bad results. You only see this with newcomer devs and gamers that are blinded by trailers.
Serious devs don't use features if they don't need them and they give them bad performance.
There are barely any big games released that run on UE5 and the instances I know of, devs evaluate if it makes sense for them to use Nanite for example, and adjust accordingly.
I won't deny that there probably are games and devs out there abusing it against their best interest, but this again is up to developers, no one is being held at gunpoint to use Nanite when you have low poly meshes or Lumen Realtime GI when all you need is static lighting.
Give them a gun and they'd probably shoot themselves, too.
I don't see people happy-posting about UE5 with Infinity Nikki despite having beautiful graphics and running really well.

As for the bolded, I am not a fan of the photorealism trend either, I prefer stylized, clean visuals, and game design and storytelling are more important to me than graphics. However this is up to individual developers and no one else.

If I read all your posts in this thread I would think that the state of UE5 games is great, devs have it under control, they know what they are doing, they are experienced, they don’t need to discuss it with anyone least of all some whippersnapper on YouTube.

Except that’s not the case, shit UE5 implementations is basically a meme now. Everyone knows how dire it is, especially and including on the PC, even PCMR shills like Alex in DF are talking about it.

So, what next.

 

[winjer]

This guy is a bit aggressive with his rhetoric, but he is right in several points. His example of how TAA in most UE4 games sucks is very accurate.
TAA in most UE4 games is really bad. His tweaks do improve image quality. And he is not the first dev to propose tweaked settings in the Epic forums.
I have been using my own tweaked settings in UE4 games. And there are users in the Steam forums and other forums, sharing better settings, than the UE4 default.
But most devs don't care, so most games are released with the default configuration. And image quality suffers.

And then we have the UE4 and 5 stutters. So many games that have issues with performance, it's unbelievable.
Only after a decade, is Epic addressing some of these issues.
And most devs that made games in UE4 didn't even bother to gather PSOs and create a pre-compile. Or optimize the streaming system for their own game.
This idea that game devs know better is very wrong. Most have no idea what they are doing and are willing to release broken games.
There are only a few devs that actually optimize their games when using UE4 and UE5. But these are the exception, not the rule.

 

[Guilty_AI]

But that scenario is fictitious. Devs do not "do it" if it gives them bad results. You only see this with newcomer devs and gamers that are blinded by trailers.
Serious devs don't use features if they don't need them and they give them bad performance.
There are barely any big games released that run on UE5 and the instances I know of, devs evaluate if it makes sense for them to use Nanite for example, and adjust accordingly.
I won't deny that there probably are games and devs out there abusing it against their best interest, but this again is up to developers, no one is being held at gunpoint to use Nanite when you have low poly meshes or Lumen Realtime GI when all you need is static lighting.
Give them a gun and they'd probably shoot themselves, too.
I don't see people happy-posting about UE5 with Infinity Nikki despite having beautiful graphics and running really well.

As for the bolded, I am not a fan of the photorealism trend either, I prefer stylized, clean visuals, and game design and storytelling are more important to me than graphics. However this is up to individual developers and no one else.

What you also need to take into account is that many big developers are taking in tons of amauter devs while bleeding talents, devs that would do stuff like this because its easy and faster. Yes, there are devs that know better what they are doing, i've played UE4/5 games that ran perfectly fine (except for the whole poor image quality due to relying on AA techniques to look proper), but there are games that don't. Too many, enough for players to notice, and this guy is pointing out exactly where the problem lies.

 

[Clear]

Not to be a dick, but objectively speaking what are the odds on one guy -with nothing of note to show- having a level of stunning insight literally nobody else in the industry has?

Sorry its just classic internet argumentation, where the world is all wrong but they are the one with the singular vision to make it better!

Talk is cheap guys, and in cases like this the razor always favors bullshitting as the likely truth.

 

[Herr Edgy]

If I read all your posts in this thread I would think that the state of UE5 games is great, devs have it under control, they know what they are doing, they are experienced, they don’t need to discuss it with anyone least of all some whippersnapper on YouTube.

Except that’s not the case, shit UE5 implementations is basically a meme now. Everyone knows how dire it is, especially and including on the PC, even PCMR shills like Alex in DF are talking about it.

So, what next.

I'm not saying things are perfect, I'm saying that people are dramatizing and misinterpreting things they do not understand.
"Shitty UE5 games" are a meme because people misunderstand what's happening, and using the existence of that meme as justification to fuel that misunderstanding doesn't help anyone.

If your assumption is "UE5 bad" and then I note a couple games running on UE5 that run well and the response is "well, those people know what they are doing so it doesn't count" then of course UE5 has to be bad because all bad things are attributed to UE5 and all good things to the developers.
This tribalistic behavior is found in politics and in everyday life, too. Reality is not so simple.

 

[diffusionx]

I'm not saying things are perfect, I'm saying that people are dramatizing and misinterpreting things they do not understand.
"Shitty UE5 games" are a meme because people misunderstand what's happening, and using the existence of that meme as justification to fuel that misunderstanding doesn't help anyone.

If your assumption is "UE5 bad" and then I note a couple games running on UE5 that run well and the response is "well, those people know what they are doing so it doesn't count" then of course UE5 has to be bad because all bad things are attributed to UE5 and all good things to the developers.
This tribalistic behavior is found in politics and in everyday life, too. Reality is not so simple.

If some games with UE5 are good but most of them are bad, it means most of the devs don't know what they are doing and need help that Epic isn't providing. It says that the problem isn't totally UE5 but the actual devs making games for it. It is actually a reason for videos like the one OP posted. Except instead of that these people are getting defensive and said they don't want to hear anything from this guy or anyone else. Of course a lot of times people who have been in an industry and made things don't want to be told their work is subpar, it stings the pride. I've seen it before in my line of work.

And it could just be that a studio has no business running UE5. They should be writing their own engine. Dragon Age Veilguard, say what you will, looks great, runs smoothly, is polished, runs on low-end hardware... and is a custom engine. This keeps happening.

In one sentence - gamers are tired of hearing from these devs that they know all while they put out these technical disasters that don't run well on console, or a 4090, or anything.

 

[Herr Edgy]

What you also need to take into account is that many big developers are taking in tons of amauter devs while bleeding talents, devs that would do stuff like this because its easy and faster. Yes, there are devs that know better what they are doing, i've played UE4/5 games that ran perfectly fine (except for the whole poor image quality due to relying on AA techniques to look proper), but there are games that don't. Too many, enough for players to notice,

Interesting point, I was inclined to ask whether you have sources for this because personally I can not confirm this observation.
However I did remember that Ubisoft stated that around half of their devs working on AC Shadows are juniors. Ubisoft is a bit of an outlier though, and so is Epic. Epic barely has any juniors. No clue about Blizzard and the likes.

and this guy is pointing out exactly where the problem lies.

But hard disagree with this one. Any game dev with some level of serious experience knows that game dev is a super complex process. Blaming bad perf on some graphics options specifically is the least accurate reading of the situation you could make.

 

[kevboard]

Not to be a dick, but objectively speaking what are the odds on one guy -with nothing of note to show- having a level of stunning insight literally nobody else in the industry has?

this is not about having any secret knowledge, it's about exposing how devs take shortcuts instead of actually optimising games.

Silent Hill 2 using Lumen GI being still one of the prime examples on how shitty this practice is

 

[Clear]

this is not about having any secret knowledge, it's about exposing how devs take shortcuts instead of actually optimising games.

Silent Hill 2 using Lumen GI being still one of the prime examples on how shitty this practice is

If you didn't take shortcuts you'd take an eternity to finish anything.

Like I've said in the past, and anyone who's spent time in a commercial creative environment knows damn well; its all about compromising because time is money.

Chasing perfection is not a business model!

You do the best you can with imperfect talent, tools and technique and hope you can get something great out the door before the money runs out.

 

[Herr Edgy]

If some games with UE5 are good but most of them are bad,

Can you give me a list of games running on UE5 that run bad so I have a baseline of what you think of when you say 'bad'?

it means most of the devs don't know what they are doing and need help that Epic isn't providing. It says that the problem isn't totally UE5 but the actual devs making games for it. It is actually a reason for videos like the one OP posted. Except instead of that these people are getting defensive and said they don't want to hear anything from this guy. Of course a lot of times people who have been in an industry and made things don't want to be told their work is subpar, it stings the pride. I've seen it before in my line of work.

And it could just be that a studio has no business running UE5. They should be writing their own engine. Dragon Age Veilguard, say what you will, looks great, runs smoothly, is polished, runs on low-end hardware... and is a custom engine. This keeps happening.

It's not Epic's job to help people make their games, unless they have a partnership with Epic that entails that help. Of course in terms of offering Unreal as a product, it benefits from community engagement and putting out documentation and talks etc., but that is different from having Epic jump in to put manpower onto other peoples' problems.
Anyone can make games with Unreal. You can download it in EGS right now, open up a template and get started. You don't even owe Epic any money for this until you meet some criteria. In turn they don't owe you anything either.

Also people are getting defensive because the TI guy is quite obnoxious in his behavior. If a young guy with potential was arrogant as hell while lacking the experience to back it up anyone would choose to sideeye him.
That doesn't mean all his points are invalid. But his points lack nuance stemming from his lack of experience. But there are plenty of reasons to dismiss him, even outside his arrogant tone.
Graphics pioneers at Epic that are responsible for a bunch of graphics papers *surely* know about downsides of TAA and for some elusive reason went ahead with it anyways. Don't you think? Surely a discussion can be had but if you dismiss experienced people's opinions before they even get a chance to speak (by means of his tone) you are also sure to make people dismiss you in turn.
If the guy offered his opinions but understood that real life game dev is not as simple as he makes it out to be, the response to him would be way less harsh.

 

[kevboard]

If you didn't take shortcuts you'd take an eternity to finish anything.

Like I've said in the past, and anyone who's spent time in a commercial creative environment knows damn well; its all about compromising because time is money.

Chasing perfection is not a business model!

You do the best you can with imperfect talent, tools and technique and hope you can get something great out the door before the money runs out.

this would make sense in a world where budgets would go down while quality goes up.

but we are in the exact reverse situation. dev time goes up, budgets go up, quality goes down or is stagnant

 

[Fafalada]

But that scenario is fictitious. Devs do not "do it" if it gives them bad results.

Path of least resistance and financial realities dominate decision making more than ever (yes there's a handful of auteur studios out there that go the extra mile, but that % is diminishingly smaller portion of the industry every year). Thus bad results do ship - all the time.
Eg. - to stay with the themes of the thread - and because the example was my pet peeve of the past gen - console Doom VFR was objectively one of the worst VR experiences of that generation, period. Main culprit - use of aggressive temporal AA.

 

[Herr Edgy]

Path of least resistance and financial realities dominate decision making more than ever (yes there's a handful of auteur studios out there that go the extra mile, but that % is diminishingly smaller portion of the industry every year). Thus bad results do ship - all the time.
Eg. - to stay with the themes of the thread - and because the example was my pet peeve of the past gen - console Doom VFR was objectively one of the worst VR experiences of that generation, period. Main culprit - use of aggressive temporal AA.

Sure, but if the reality between 'We are able to make that game at 90% of our vision' and 'We are able to make that game at 100% of our vision' is 'We can actually make and ship that game', then the former choice is surely the best choice.
No clue about Doom VFR, never played it; I'll take your word for it. I get the feeling of seeing unrealized potential, and maybe that was related to budget, who knows, and I wish it was different too.
But I think games in which financial restrictions cause unrealized potential in whatever shape or form don't really matter much to this thread, no?
I guess if we are talking about TAA specifically, it does, but if the developers that are held on a pedestal for their well running and looking games still chose to use TAA then I'd trust there are reasons for this too and it's not to smite someone.

 

[DoktorCoktor]

I actually tested his TAA modification in Jedi Survivor, I got grass flickering, proceeded to ignore the channel.

Try these.
#Anti-aliasing & Sampling
r.DefaultFeature.AntiAliasing=2
r.PostProcessAAQuality=6
r.TemporalAASamples=32
r.TemporalAAPauseCorrect=1
r.TemporalAACurrentFrameWeight=0.08
r.TemporalAAFilterSize=0.6
r.Jittering=1
r.Tonemapper.Sharpen=0.7
r.Tonemapper.Quality=0

 

[Clear]

this would make sense in a world where budgets would go down while quality goes up.

but we are in the exact reverse situation. dev time goes up, budgets go up, quality goes down or is stagnant

The workload keeps going up though.

Seriously, if it was just a matter of AAA extravagance breeding inefficiency tell me why indie projects often take 3, 4. 5 years to make?

If people could just shit out good quality stuff on small budgets in the sort of timescales that used to be the norm, why aren't they doing that? It'd be a license to print money and a lot more rewarding creatively for the developers. Its not easy looking at a project after you've been working on it for years! All you see are the flaws!

Sorry but it gets really tiresome humoring this idea that the industry is somehow "doing it wrong" when there is zero evidence of an actual viable alternative!

 

[Aces High]

When I first saw the channel, I immediately thought it was run by Epic Games themselves to deflect the growing criticism of UE games onto developers.

Also, the talking head looks like CGI.

 

[kevboard]

If people could just shit out good quality stuff on small budgets in the sort of timescales that used to be the norm, why aren't they doing that?

this is a good question, but it's undeniably happening.
look at the dev time and budget of Concord, and compare it to the dev time and budget of Space Marine 2.
and now compare the scope of both titles, the amount of content they had at launch etc.

Space Marine 2, running on in-house tech by Saber wipes the floor with nearly anything on the market in this regard. and it's not like Space Marine 2 is a perfect game, it has flaws, but somewhat understandable ones.

and these might be extreme examples, but then you compare Batman Akrham Knight with either Suicide Squad or Ghotham Knights and see the same pattern, this time by the same publisher and in one case even the same studio.
lower quality, longer dev time, higher budget.

 

[Fafalada]

But I think games in which financial restrictions cause unrealized potential in whatever shape or form don't really matter much to this thread, no?

The point is more that software-dev is a constant balancing act of various constraints. The idea that 'tech is in the lead' is quaint - but increasingly rarely a reality. Basically - it's not uncommon at all to use suboptimal choices because they fit some other piece of the puzzle. Or sometimes because you simply have to make 'a' decision, even when it's not fully informed.

The irony is that this channel and many of its detractors alike are arguing from the same vantage point of 'developer is always right' just in reverse, and they're both wrong.

I guess if we are talking about TAA specifically, it does, but if the developers that are held on a pedestal for their well running and looking games still chose to use TAA then I'd trust there are reasons for this too and it's not to smite someone.

There are always reasons - I'm just saying sometimes they are objectively wrong, and we still ship games in that state anyway.
Doom's problem was that the type of TAA used 'might' look 'ok' on a flat-screen, but it just becomes a blurry smear on VR panels that made it painful to look at.
However, we can't just derive such examples into absolutes - ie. TAA is always bad / always good. As a counter example, I've shipped VR titles myself that used TAA that had none of the problems - there's usually nuances to all these decisions, but that gets lost in online discourse that is well... thriving on polarization rather than actual discussion.

 

[Clear]

this is a good question, but it's undeniably happening.
look at the dev time and budget of Concord, and compare it to the dev time and budget of Space Marine 2.
and now compare the scope of both titles, the amount of content they had at launch etc.

Space Marine 2, running on in-house tech by Saber wipes the floor with nearly anything on the market in this regard. and it's not like Space Marine 2 is a perfect game, it has flaws, but somewhat understandable ones.

and these might be extreme examples, but then you compare Batman Akrham Knight with either Suicide Squad or Ghotham Knights and see the same pattern, this time by the same publisher and in one case even the same studio.
lower quality, longer dev time, higher budget.

Dude, a quick check shows Space Marine 2 took probably 5 years+ to make!

Just checking the wikipedia entry shows it was first announced at the end of 2021 with a cinematic trailer. Meaning that it most likely had been in development for at least a year, (most probably longer given pandemic conditions) as it was originally pencilled for a 2023 release. It eventually comes out in September this year.

As for the Arkham games; bear in mind that the free-flow combat system was an evolution of the mechanics developed for Orchid, a failed project for PS2/Xbox that was the last thing being worked on at Argonaut before they went bust in 2004. Given Arkham Asylum didn't come out until 2009. and Orchid dates back to 2001-2002 ish... You do the math!

So no, its absolutely not happening. You'd probably be shocked by the amount of time taken to squeeze out even relatively simple stuff like mobile titles, because unless you've been in the business you wouldn't know how long the process is for crafting an app-store icon!

 

[mckmas8808]

They're trying to silence me and "i'm under confirmed censorship" : somewhere around the :45 second mark
"The people that support Threat Interactive will always know the truth" somewhere around the 11:00 mark.

Not an explicit call for Patreon, I was being more hyperbolic than anything, but basically "Only my followers will know the truth" type nonsense. The patreon thing I just threw in because it's so common these days .

He's a crazy person LOL! I hate when people do that.

 

[Herr Edgy]

The point is more that software-dev is a constant balancing act of various constraints. The idea that 'tech is in the lead' is quaint - but increasingly rarely a reality. Basically - it's not uncommon at all to use suboptimal choices because they fit some other piece of the puzzle. Or sometimes because you simply have to make 'a' decision, even when it's not fully informed.

The irony is that this channel and many of its detractors alike are arguing from the same vantage point of 'developer is always right' just in reverse, and they're both wrong.


There are always reasons - I'm just saying sometimes they are objectively wrong, and we still ship games in that state anyway.
Doom's problem was that the type of TAA used 'might' look 'ok' on a flat-screen, but it just becomes a blurry smear on VR panels that made it painful to look at.
However, we can't just derive such examples into absolutes - ie. TAA is always bad / always good. As a counter example, I've shipped VR titles myself that used TAA that had none of the problems - there's usually nuances to all these decisions, but that gets lost in online discourse that is well... thriving on polarization rather than actual discussion.

I agree, and that's what annoys me so much about this topic. Gamers often only think of games as an end result, because that's what they interface with - there is neither room nor curiosity for understanding why things ended up the way they did.
They don't need to understand either, which is why this wrongful positioning of having strong opinions on things they don't have any clue about is counterproductive.
I work in the industry, but not on graphics, so I'm familiar with the process, but I don't presume to know the ins and outs of graphic techniques. I trust that the year-long experts I'm working with have their reasons for doing things certain ways, and when I don't understand something, I'm going to assume they know better than me when I'm reaching out and ask for clarification.
That doesn't mean that I have to agree with the direction. But direction is more akin to creative decision than it is to correctness.
It is true that the choices that are made aren't always the best choices, but I would argue they are rarely the *wrong* choices.

They might be shortsighted, or suboptimal in hindsight, but provided you are working with experienced and committed professionals and not amateurs, the choices are made with the best intentions and justified with reasonable logic.
Deadlines alone will incur technical debt if you are working on software that keeps evolving. We might get told to prioritize certain work over other work due to some project's dependence on it.
Is it optimal? No. Do we have good alternatives? Also no.

When your perspective due to experience supersedes a simple "X good/bad" and you start thinking about X in terms of context, you arrive at the answer 'It depends'. But that answer is not easily understood by people who don't have insight into the topic.
Hidden within are perspectives that don't even cross people's minds and you can't easily make up for the lack of experience on the other's part by articulating yourself well.

 

[AGRacing]

He's got a touch of Boogie2988 syndrome. But maybe he's got a point on the optimizations.
Something has to explain the gulf in performance between Unreal Engine 5 games.
After Indiana Jones I'm all in on that ID Tech engine getting some more use.

 

[Buggy Loop]

Kudos on the kid, he has balls. Pompous? Yea so what. Graphic geniuses bordering on autism are pretty much like that. They just never get the mic. The Cherno appreciated his videos and that's something, the dude was a graphic engine designer at DICE, just the motherfucking PEAK of Frostbite.

I mean we have devs that can barely lift the finger to include settings like Ultrawide when script kiddies implement a fucking mod or solution to hex edit within typically 24h a game releases. For sure they're just using UE5 plugins that they don't understand or optimize for, following what Epic suggests (always nanite on, always Lumen, always... ), taking assets from the store with too high geometries and not optimize them. Studios can fire these R&D engine nerds as they don't need inhouse solutions anymore, they can lower the experience threshold of a whole studio to junior devs, even get workforce from low cost labour countries, as everyone in the market for a job in the field right now at least has to know UE. We get this slop. No it's not the engine the problem, it's the ecosystem. They're promoting a tool to remove the need to have senior devs with engine knowledge. So when it gets to optimization, it's a shitshow.

He's bringing the receipts. He shows the slop that finds its way into 9th gen games. Image quality especially has hit an historic low. A real regression there. He's not alone either mentioning these problems, go visit UE forums.

But they are the case most of the time because most games are interactive and dynamic by nature.

No, most aren't. Not in the sense of an engine built around a cash cow like Fortnite where buildings are built/destroyed and needs really dynamic lights.

Raster had all sorts of solutions for dynamic lights, cmon, we aren't in the pre-baked lightmaps over texture days anymore ala Quake/Half-Life.

2005 F.E.A.R.

2007 Stalker with its xray engine


With addons



I mean, kind of shame isn't it? UE5 and Lumen for Stalker 2, that's not even shadow casting like the original.

2007 also had Crysis which had a ton of dynamic scenes, you could destroy houses etc. Can it run!

2009 Killzone 2

2011 Crysis 2


2013 Battlefield 4 (with destruction)

2014 Driveclub

2016 Quantum Break

2017 Horizon Zero dawn

2018 Forza Horizon 4

And I pass a shitload more examples because I would be spending my whole day here.

The argument that games were lacking in dynamic lighting when using raster doesn't hold.

On top of that, peoples are using the additional computational power and always refining the precision or finding alternative ways to use raster/software GI that closely match ray tracing.

Star Citizen's upcoming software solution is based on AMD's GI for their RT, but they used the same probe approach for software raster, which uses radiance caching with a ton of probes (again, not computationally heavy nowadays).



Radiance cascade in 3D is also the same idea with alternative tweaks

There's still a lot of brilliant peoples in that field of research and they are still finding solutions.

I already answered the case of *what if your game still doesn't need all this*, well the answer is, you don't use a dynamic setup that costs perf? Is that not obvious?

Then you miss a marketing opportunity, the marketing department will say, since UE5 buzzwords and early gen promises was a lot of hype and inflated expectations. So why bother? They'll run Lumen at crazy refresh times in a static scene like we see in Silent hill 2, as if the environment will explode like Fortnite or something.

Developers on UE forums have mentioned they want Lumen for sake of brevity in creating scenes and lighting them. This is fine. But the engine has no smart features when Lumen should refresh, it's always refreshing like it's fortnite. These should be triggered by game events that require a Lumen refresh of the scene. Smart use of GPU usage. If the game is 99% static, why refresh this motherfucker like a stroboscope.

 

[Inviusx]

Guy is literally just campaigning for better optimisation in games, I don't see what's not to like? He's obviously pro consumer and he's talking about stuff that everyone has been thinking about. Goodluck to him I say.

 

[Spukc]

Is this sheldon?

 

[bender]

This guy is a bit aggressive with his rhetoric

I chalk up his tone and demeaner to his youth. I know someone who was incredibly bright in his youth and his way of proving himself on a team full of much older coworkers was to be an aggressive asshole. He got away with it because he was most often correct. I'm not saying this is the correct approach nor that this person is correct, but I do appreciate his passion even if he can be insufferable.

 

[Myths]

That's basically all the engines out there.

WTF is this guy smoking and why is trying to become an authority on game dev?

He’s working the algorithm. Mostly the average gamer will be impressed, maybe a few devs but I’ve seen a number of devs on sub reddits and comments call this guy out. Pretty well known TAA hater and know-it-all without any intimate knowledge on pipeline production and workflow.

 

[Neilg]

Not to be a dick, but objectively speaking what are the odds on one guy -with nothing of note to show- having a level of stunning insight literally nobody else in the industry has?

Most people who have got to the point to learn about this are represented by an employer that would not appreciate them making content like this. Or indie devs would not want to jeopardize publishing help. A lot of people who use ue daily agree with him.
He's a little green, far too enthusiastic with his rhetoric, and bullish to the point where I think he's going to trip over himself - but currently, not totally wrong.

More experienced people know the compromises / complexity behind decisions which he's completely ignoring.

 

[M1chl]

This thread reeks of ableism towards an autistic kid, who is really passioned about better graphics in games. (hyperbole)


And its a weird grift, when he actually shows the result, its not like he is just talking about some shit, did people actually watch his videos.

Granted I stopped watching in half, because I have no desire to see the part when he is addressing others, I simply don't care. But that first half of the video is solid

 

[ScHlAuChi]

So what? He seems to be doing a better job of properly optimizing video games than all the folks working in the big budget studios, and he's the only developer out there willing to call out the fact that modern games have 2018 visuals while requiring 2027 hardware. 99% of reviewers out there have never made a damn game.

He optimized a SINGLE scene that had ONE optimization solution that ONLY works in this specific case!
Do you think his "solution" would scale and work on any modern AAA game?
How long would he need to optimize something with the size of Assassins Creed or GTA5?

This guy is akin to a snake oil salesman saying his solutions are superior than everyone elses while not having any actual game dev experience!

 

[Herr Edgy]

No, most aren't. Not in the sense of an engine built around a cash cow like Fortnite where buildings are built/destroyed and needs really dynamic lights.

Raster had all sorts of solutions for dynamic lights, cmon, we aren't in the pre-baked lightmaps over texture days anymore ala Quake/Half-Life.

2005 F.E.A.R.

Mate I am spent so I will keep this brief but if you are talking about solutions for dynamic lights and then post a gif that shows BOTH static & dynamic lighting and a light function, in just one light, it just makes it so easy to dismiss your point.
This gif has been posted a couple times as a means to show awesome dynamic lighting that was done many years ago but a single look at that pillar shows how it isn't all that dynamic and the new tech is being developed because of things like these specifically. Don't have to like it but don't kid yourself that it serves no purpose other than making your perf bad.
It is being developed because there are unsolved problems out there worth addressing.

EDIT:
On second view I'm wrong on the pillar, that's dynamic also. Regardless the point about it being just one light stands, Mega Lights or Many Lights as it's also called is about supporting effectively an unheard of amount of shadowing dynamic lights, and all the examples there were regular instances of a single or just a few lights at once casting shadows. Fundamentally misunderstanding what Mega Lights is about.

 

[Bernardougf]

Anyone thats okay with the current state of games, and on top of that is DEFENDING modern developers who produce shit that cant even run at NATIVE 1080p60fps anymore should be checked into a insane asylum.

I think gaming is the only hobby where the majority of customers are such meek little sheeps that wont demand quality, no one in the world would accept sub par products and service in a retail store for example, the store would go out of buisness.

Most people get up in arms over the 4k releases on movies that have used AI for upscalling and looks like shit, but apperantly to be critical of game developers is to much, i can play games that came out 10 years ago that looks better than modern games and runs on a fucking potato. It's insane how much this hobby has degraded

Agreed... but is peek modern participation trophy mentality .... everything must be good and beautiful trough a certain point of view... so nothing can really be bad , or ugly , or stupid/dumb, everything is subjective, there can be no objectivity whatsoever....

Any criticism of games are taken personally as a personal attack ... people project their own selfs on this products and companys and just have to defend every shit they can.


Games are not anymore seen as a product.. but a extesion of the devs who made it, the reality around it and who buys it .. so criticism is not allowed... you talk bad about any aspect of the game and people will talk bad about you and be anger at you like you attacked their own mother. Is really bizarre.

Seen the last thread about interstellar you would think that half the people there has bald mom's dying of cancer or that their main fapping search on porn hub is "masculinized bald chick" so talking about an ugly bald virtual character is some fucking personal insult that generates anger..

 

[ScHlAuChi]

What's bad is overreliance, often unnecessarily, on certain tech and excuse poor practices with it.

That overreliance is a result of modern development realities.
Gamers want their super highend games with super highend graphics in as little time as possible.
Most publishers simply dont want to spend that extra time/money.

Before devs would follow the correct advice and "don't do it", or at least do it in a more efficient way that worked with their game. Now they "do it", turn on mega lights or whatever other thing, and get better performance than the absolute worst case scenario. Except that "better" performance is still awful, incomparably worse than if they just followed older methods.

Before, games werent as complex and large as they are now. Just look how much GTA evolved with each iteration. The simulation complexity of a GTA6 is probably 1000x higher than a GTA3!
Is that really needed? Probably not, but as with horse testicles in RDR2, people will admire so much crazy detail.

And you can argue the new tech is technically more accurate, or works out-of-the-box in a bigger variety of different scenarios. But rare are the games where many of such scenarios even matter or happen at all, with these implementations offering marginal improvements for whats actually in the game for a huge performance cost, if any.

The biggest difference with the new tech is that it actually allows devs to still make games without going insane!
Baked lighting on big complex scenes would simply be unfeaseable!

Like your torch example, does it even matter how accurately the light of the torch scatters and blends with the other lights in a fully illuminated hall? Vast majority of players wouldn't notice. What they will notice however is the frame rate dropping to a crawl due to the usage of some expensive and unnecessary tech.

Not every game needs dynamic lights, but it all comes down to what game you are making.
Take something like Thief for example, would you say accurate dynamic light isnt important for such a game?

 

[Wildebeest]

This guy makes people angry. The fact is that there is no magic bullet for optimisation of 3d. It was a problem from the start of the 3d era that framerates could vary so wildly from one scene to the next, and tracking down the reasons why was a massive pain. Epic and others offer magic technology to get past the stressful optimisation process, which often means changing level geometry or scrapping art assets, but their magic depends on future hardware that doesn't exist yet and can compromise image quality.

 

[Herr Edgy]

As a final thought on the matter, the guy's aim is collecting 900k USD by means of YouTube donations with 0 visibility or accountability while he can not be arsed to pay 15 bucks per year to buy a domain for his website. He also has not demonstrated any capability in the rendering programming realm (I do not doubt he did some hobby projects, but has not demonstrated anything). He is not stupid, but you have to be stupid to believe what he says at face value.

I think he started out with youthful and good intentions but has devolved into a grifter seeing how he got attention.

 

[Buggy Loop]

Mate I am spent so I will keep this brief but if you are talking about solutions for dynamic lights and then post a gif that shows BOTH static & dynamic lighting and a light function, in just one light, it just makes it so easy to dismiss your point.
This gif has been posted a couple times as a means to show awesome dynamic lighting that was done many years ago but a single look at that pillar shows how it isn't all that dynamic and the new tech is being developed because of things like these specifically. Don't have to like it but don't kid yourself that it serves no purpose other than making your perf bad.
It is being developed because there are unsolved problems out there worth addressing.

2005 dude, 2005, about to be 2 decades old. On DX8 API and the very first shader pipeline cards.

Do you understand that I tried to make an evolution of dynamic raster lighting? Your point is bullshit, raster could/can work around dynamic worlds, there was a massive shift circa 2011-2013 where engines started to implement light probes.

I can also spot many places where the latest tech with huge hardware requirements and runs like pixel soup on consoles fails miserably for no benefits.

In fact, BRAVO Lumen for not shadow casting enemies nor have gun nozzle flashes on Stalker 2 in 2024