• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Shots Fired! Threat Interactive calls for you to unsubscribe from Digital Foundry.

If subscribed currently, does this video convince you to unsubscribe?


  • Total voters
    201

Landr300

Member
There is a good reason why modern devs cant do this level of optimizations:

- Games have become much much larger, the amout of developers, content and assets needed for a modern game is like 100x larger than back then.
TI wants highly optimized games, and calls developers incompetent, but lacks any sort of understanding of what modern dev reality is.

- Asset quality and complexity is much higher, manual optimization has become simple impossible.
How do you optimize Quixel assets with tens of millions of polygons? Do it by hand?
How do you optimize an openworld map with hundreds of thousands of lights? Do it by hand?
TI basically suggests using a knive to open a can - as its cleaner than using a can opener machine that makes rougher edges.
Sure, this works when you have 1000 cans to open, it might work fine, but when you have 1 million cans to open, the knive solution is simply impractical!
The result would be even longer dev cycles.

- Making a modern game the "old way" with baked lighting is pretty much impossible from a productivity standpoint as it takes way too long.
Any oldschool dev can tell you that baked lighting drives you insane, as whenever you move a light or an asset, you have to redo it.
Not a big problem when your game is only made out of small rooms, but on the larger scale like openworld games, this creates so much downtime that it becomes unproductive.

- Modern tech is way better than the old tech.
TI might hate Unreal, but thats because he doesnt understand how modern games are made, nor understands the tech behind it.
Nanite and MegaLights work well when used correctly in scenarios they were made for. His examples are bad examples.
Raytraycing is also used for much more than just pretty graphics, it´s also used for Audio, AI, Culling etc.

- Developers arent getting paid to optimize, they get paid to produce content.
TI doesnt seem to understand that it isnt developers that call the shots, its the publishers.
A developer executes what the publisher demands them to do. They have the money, they pay you! As a dev you do what they say - end of story!
His angry rants against devs wont change that, unless he is the one ponying up the cash!

TLDR version:
TI offers seemingly simple solutions that fall apart pretty fast once you look at it in detail and are completely impractical on the large scale.
well, if they keep launching games unoptimized they wont sell and then makes no difference if they are making them fast or not, not all games can sell like monster hunt, just look at the ronin who was bad on the ps5 and is even worse on pc, it could have far more people than whats having right now if the games was optimzed

and the problem is not just the publishers, here we have a indie, 2.5D stilyshed, self published game, have a hard time running, and its not even unreal, nothlight or RE Engine, its Unity

its literally happening all over the industry, with or whitout publisher and doesn't even matter what engine they're using

 

Guilty_AI

Member
There is a good reason why modern devs cant do this level of optimizations:

- Games have become much much larger, the amout of developers, content and assets needed for a modern game is like 100x larger than back then.
TI wants highly optimized games, and calls developers incompetent, but lacks any sort of understanding of what modern dev reality is.
That just sounds like a workflow pipeline problem that needs immediate addressing.

- Asset quality and complexity is much higher, manual optimization has become simple impossible.
How do you optimize Quixel assets with tens of millions of polygons? Do it by hand?
How do you optimize an openworld map with hundreds of thousands of lights? Do it by hand?
TI basically suggests using a knive to open a can - as its cleaner than using a can opener machine that makes rougher edges.
But when you have 1 million cans to open, the knive solution is simply impractical! The result would be even longer dev cycles.
Problem it's not just large open world games suffering from these kinds of problems. Besides, i'd assume with so many hands working with these companies they should've been able to do that much. If they can't, it's once again a workflow pipeline that seriously needs solving.

The examples you gave also aren't very good since there are fairly straightforward ways to optimize those. With Quixel you can literally export downsized resolutions automatically, which would theoretically allow you to create LODs for example. With lights, you can just replace those very far away with bright dot textures, and those at a mid distance with ones that don't produce dynamic shadows, which doesn't necessarily translate into extensive work if planned from the start. GTA has been doing that since the ps360 days.

- Making a modern game the "old way" with baked lighting is pretty much impossible from a productivity standpoint as it takes way too long.
Any oldschool dev can tell you that baked lighting drives you insane, as whenever you move a light or an asset, you have to redo it.
Not a big problem when your game is only made out of small rooms, but on the larger scale this creates so much downtime that it becomes unproductive.
This used to be a pain because it took time to create the lightmaps. Nowadays you can bake those on the fly since modern machines can literally calculate real lighting on run-time. It'd only take time if you decided to calculate a large world all at once, instead of only small segments of it at a time according to what you're tweaking, but that'd be a stupidity problem.

- Modern tech is way better than the old tech.
TI might hate Unreal, but thats because he doesnt understand how modern games are made, nor understands the tech behind it.
Nanite and MegaLights work well when used correctly in scenarios they were made for. His examples are bad examples.
Raytraycing is also used for much more than just pretty graphics, it´s also used for Audio, Culling etc.
Yeah, they're great tools when used correctly. The whole problem is that they aren't being used correctly and instead just became crutches for poor development practices.

- Developers arent getting paid to optimize, they get paid to produce content.
TI doesnt seem to understand that it isnt developers that call the shots, its the publishers.
A developer executes what the publisher demands them to do. They have the money, they pay you! As a dev you do what they say - end of story!
His angry rants against devs wont change that, unless he is the one ponying up the cash!
This, once again, sounds like a problem that needs to be solved, not a counter-argument. He's pointing out what's wrong, saying why that wrong thing is like that doesn't justify the problem's existence to begin with.

TLDR version:
TI offers seemingly simple solutions that fall apart pretty fast once you look at it in detail and are completely impractical on the large scale.
He's pointing out what's wrong from a technical point of view. You've mostly given counter-points that relate to workflow problems. That doesn't justify the problems, it just shows there are serious issues with the dev process that need fixing.
 
Last edited:

ScHlAuChi

Member
well, if they keep launching games unoptimized they wont sell and then makes no difference if they are making them fast or not, not all games can sell like monster hunt, just look at the ronin who was bad on the ps5 and is even worse on pc, it could have far more people than whats having right now if the games was optimzed
"They wont sell" - and yet they do! Because casuals dont care about this!

and the problem is not just the publishers, here we have a indie, 2.5D stilyshed, self published game, have a hard time running, and its not even unreal, nothlight or RE Engine, its Unity
If youre an indie and you spend alot of your money on optimization and the game doesnt sell, you go bankrupt!

its literally happening all over the industry, with or whitout publisher and doesn't even matter what engine they're using
The consumer wants more content for less money!
The money has to be saved somewhere - and that is usually QA and optimization!
 

Landr300

Member
"They wont sell" - and yet they do! Because casuals dont care about this!
they do? Outside of monsters like monster hunter we have failure over failure in pc at least, and the casuals care about ratings and what they hear from others

If youre an indie and you spend alot of your money on optimization and the game doesnt sell, you go bankrupt!
so the issue its not publishers, and again, if you dont spend on optimization you dont sell
 

ScHlAuChi

Member
That just sounds like a workflow pipeline problem that needs immediate addressing.
Problem it's not just large open world games suffering from these kinds of problems. Besides, i'd assume with so many hands working with these companies they should've been able to do that much. If they can't, it's once again a workflow pipeline that seriously needs solving.
None of his solutions will solve that, instead make it worse.

The examples you gave also aren't very good since there are fairly straightforward ways to optimize those. With Quixel you can literally export downsized resolutions automatically, which would theoretically allow you to create LODs for example. With lights, you can just replace those very far away with bright dot textures, and those at a mid distance with ones that don't produce dynamic shadows, which doesn't necessarily translate into extensive work if planned from the start. GTA has been doing that since the ps360 days.
Well yes you can, but I just used this as an illustration of the larger problem.
His optimization solutions simply dont scale!

This used to be a pain because it took time to create the lightmaps. Nowadays you can bake those on the fly since modern machines can literally calculate them on run-time. It'd only take time if you decided to calculate a large world all at once, instead of only small segments of it at a time according to what you're tweaking, but that'd be a stupidity problem.
It is more painless now, but there is just no reason to go backwards to old tech.

Yeah, they're great tools when used correctly. The whole problem is that they aren't being used correctly and instead just became crutches for poor development practices.
They have been used correctly in games, but its not tools alone that make or break a game.

This, once again, sounds like a problem that needs to be solved, not a counter-argument. He's pointing out what's wrong, saying why that wrong thing is like that doesn't justify the problem's existence to begin with.
He's pointing out what's wrong from a technical point of view. You've mostly given counter-points that relate to workflow problems. That doesn't justify the problems, it just shows there are serious issues with the dev process that need fixing.
Yes because his solution would lead to more workflow problems.
Those serious issues are unfixable, unless there is more time or money!
 

bender

What time is it?
You can take any games released in the past 15 years and find things to criticize.

Moose Life

200w.gif
 

ScHlAuChi

Member
they do? Outside of monsters like monster hunter we have failure over failure in pc at least, and the casuals care about ratings and what they hear from others
If a game fails on PC but makes a shitton of money on console, do you think a publisher cares?
As for stuff being unoptimized on PC, well do what we did back then:
yvmCKJH.png

so the issue its not publishers, and again, if you dont spend on optimization you dont sell
To optimize you need to have the money or time to afford to do that.
Most indies dont - thats just the reality!
 

Landr300

Member
If a game fails on PC but makes a shitton of money on console, do you think a publisher cares?
they're making less money than what they could, and if happens to sell less on consoles because of that then yes i think they care


To optimize you need to have the money or time to afford to do that.
Most indies dont - thats just the reality!
most indies are already making games less intensive to maximize their profits on hardwares like switch or steamdeck anyway, i only used this specific one as exemple to counter this argument that the issue is publishers
 
Last edited:

ScHlAuChi

Member
they're making less money than what they could, and if happens to sell less on consoles because of that then yes i think they care
As long as the money people think it will costs them more than they will make, this wont change.
I dont like that either, but it is how it is!

most indies are already making games less intensive to maximize their profits on hardwares like switch or steamdeck anyway, i only used this specific one as exemple to counter this argument that the issue is publishers
It is basically the same issue - for AAA games its the publishers that determine devtime and money.
For indies it is their own time and money. Making smaller games is certainly a way to have more time to optimize.
 

Landr300

Member
As long as the money people think it will costs them more than they will make, this wont change.
well, with the failure rate we have i think this will change fast

Making smaller games is certainly a way to have more time to optimize.
it doesn't necessarily need to be smaller in the sense of having less content, they just dont care about new techs, the issue is devs want to be on par with new techs available
 

Guilty_AI

Member
None of his solutions will solve that, instead make it worse.
How?

Well yes you can, but I just used this as an illustration of the larger problem.
His optimization solutions simply dont scale!
Not everything needs to scale

It is more painless now, but there is just no reason to go backwards to old tech.
There are plenty of reasons to go back to old tech. In fact, a lot of "modern tech" like ray tracing is very old.

They have been used correctly in games, but its not tools alone that make or break a game.
Based on results, no they haven't.

Yes because his solution would lead to more workflow problems.
Those serious issues are unfixable, unless there is more time or money!
Again, how? And if they're so unfixable, why the hell there are games from 5-10 years ago that look and perform better than modern ones while not being particularly anymore complex or any smaller in scale?
 

ScHlAuChi

Member
Because thats just what the workflow is for modern games.
It works for them - if it doesnt work for you, well, youre not the one in charge ;)
Ultimately, thats just where the industry direction is going - its a transition, like 2D to 3D was.
Yes early 3D games were terrible, but you wouldnt have said going back to 2D was the solution!

Not everything needs to scale
I agree, but im indie, what works for me doesnt work for big AAA studios.
Im not going to suggest what they should do, as they will know better what works at their scale.

There are plenty of reasons to go back to old tech. In fact, a lot of "modern tech" like ray tracing is very old.
But you have to be realistic, it just isnt going to happen.
And yes Raytraycing is old, I did that on my Amiga back then!
But it just wasnt viable to do it realtime until now.
Some might say it still isnt, but there is no way back.

Based on results, no they haven't.
Again, how? And if they're so unfixable, why the hell there are games from 5-10 years ago that look and perform better than modern ones while not being particularly anymore complex or any smaller in scale?
Well would you say games like Lords of the Fallen were terrible? I wouldnt.
I cant think of a single game in the last 5-10 years that looks better than modern games - but performance wise I agree.
As for complexity, modern games are way more complex in the amount of details they simulate.
That said, it is a classic case of diminishing returns, as the number of people that will realize that horse testicles in RDR2 shrink when its cold is tiny!
 
Last edited:

YCoCg

Member
GAF hate on DF but they still make threads about it and it’s most popular threads around here.
Negativity THRIVES on GAF, that's why positive and threads praising games barely get two pages but outrage bait will get you 10+

It's also why the politics section used to be super busy before it was closed. People in GAF just prefer to be angry and post about what they hate.
 

Bieren

Member
DF or not, I just want everyone to stop over analyzing console graphics. It's a console, tell me which one the game runs better on and send it. Leave the extremely detailed graphic chasing to PCs.
 

Danjin44

The nicest person on this forum
Negativity THRIVES on GAF, that's why positive and threads praising games barely get two pages but outrage bait will get you 10+

It's also why the politics section used to be super busy before it was closed. People in GAF just prefer to be angry and post about what they hate.
GAF it’s too fucking jaded for their own good to the point if someone spends too much time here they would think gaming gone completely to shit and there is nothing to play but reality entirely different.
 

Guilty_AI

Member
Well would you say games like Lords of the Fallen were terrible? I wouldnt.
I cant think of a single game in the last 5-10 years that looks better than modern games - but performance wise I agree.
Cyberpunk 2077, RDR2, Days Gone, TLoU2... heck, i could point out select few modern games that run way better than others. Why the fuck Kingdom Come Deliverance 2 can make 1080p high 60 fps on a rtx 3060 and Monster Hunter Wilds can barely hold 30 on similar settings? The latter certainly isn't more complex, nor does it look better.
 
Last edited:

bender

What time is it?
As the most non-jaded robot on GAF, I can honestly say that gaming has gone completely to shit and there is nothing to play and reality is no different that what all my circuits perceive.
 

ZehDon

Member
Is that your capture? Seem overly bright to begin with

My turn

m4Fmhi9.jpeg


XmhwaPN.jpeg


DN1S9Qe.jpeg


7mQJaeo.jpeg


LhTYioc.jpeg


Path tracing indirect lighting was something else

iDhpkzU.jpeg


XPHRC7y.jpeg


ibpkgkd.jpeg


ID tech and Machine games here totally deserves praise.
Considering how well it performs, The Great Circle actually really surprised me. The trailers didn't do it justice - either in terms of gameplay, or in terms of visuals. Game looks absolutely stunning at times, pretty well optimised, and is pretty close to beating Fate of Atlantis for my favourite Indiana Jones games. I'm not sure if its graphics of the year, but it's absolutely in contention if for no other reason than it looks better than most games running at half its frame rate.
 

Salz01

Member
They need to get rid of Rich, his yellow, teeth and his stupid merch. Everytime he is on I scrub on the time line. Alex has calmed down so he isn’t as annoying any more. John and Tom are good.
 
Beat you to it years ago! I hate the whole concept of DF. It just fuels fanboys and great distracts from the actual game experience.
The actual game experience is the thing that those guys miss entirely. Wouldn't surprise me if none of them actually play games.
 
Negativity THRIVES on GAF, that's why positive and threads praising games barely get two pages but outrage bait will get you 10+
It's also why the politics section used to be super busy before it was closed. People in GAF just prefer to be angry and post about what they hate.
There's no progress without outrage.
 

adamsapple

Or is it just one of Phil's balls in my throat?
Is that your capture? Seem overly bright to begin with

My turn

m4Fmhi9.jpeg


XmhwaPN.jpeg


DN1S9Qe.jpeg


7mQJaeo.jpeg


LhTYioc.jpeg


Path tracing indirect lighting was something else

iDhpkzU.jpeg


XPHRC7y.jpeg


ibpkgkd.jpeg


ID tech and Machine games here totally deserves praise.



PC version ran great at launch with little to no typical issues and a suite of upgrades over the console, which itself ran with RTGI at pretty high resolutions and a damn near locked 60, even on Series S.

They definitely knocked it out of the park.

On a sheer moment to moment, Hellblade 2 looks better in terms of sheer fidelity, but as a combination of fidelity, ray tracing and performance.. I can understand why both DF and NXG would give this their best graphics award for last year.

And it was a damn fun game to play, despite what any shiny metal assed robots might say.
 

Codeblew

Member
This guy comes off as a bit of an asshole, but that doesn't mean he is wrong. My biggest mentor in my career was deemed an asshole to almost everybody else, but I learned a lot from him. RIP David.
 

Myths

Member
He might be right, but I couldn’t get past the two minute mark. I don’t care for his delivery at all. Bullet points anyone?
- I’ve never shipped one game
- I don’t know a single line of code for any graphics API
- I think I know how to better optimize various rendering techniques than those who have either experience or formal education on the matter.
- I’m young and feel as though I’m leading a noble cause like every other dim-witted, naive young adult thinking they’re a revolutionary martyr.
- I’m grifting for attention, maxing out the algorithm, while posting on duplicate accounts to counter pushback.
- I also need money to fund other people to do the actual work because I need to spend more time yapping about how it could be better rather than spend that same time doing it

That about sums it up.

This guy comes off as a bit of an asshole, but that doesn't mean he is wrong. My biggest mentor in my career was deemed an asshole to almost everybody else, but I learned a lot from him. RIP David.
Except the part where the person who taught you actually knew stuff, of course. This guy on the other hand is sophomoric.
 
Last edited:

spons

Gold Member
DF is graphics whoring at a professional level. I get technical complaints about framerate and annoying shit like pop-in, but most of it is talk we should've moved on from decades ago.
 

Black_Stride

do not tempt fate do not contrain Wonder Woman's thighs do not do not
They need to get rid of Rich, his yellow, teeth and his stupid merch. Everytime he is on I scrub on the time line. Alex has calmed down so he isn’t as annoying any more. John and Tom are good.

Rich is Digital Foundry, the other guys work for him.
What you are proposing is the other DF members go start another channel doing what Digital Foundry does.
 

Knightime_X

Member
As with anyone, even DF isn't the voice of god.
I'll look at what they have to say and take things into consideration. I'll either agree, disagree, or just flat out don't care.
This TI guy is off his rocker. He does bring up some solid points, but the more I watch him, the harder I have to roll my eyes.
 

panda-zebra

Member
Best part was hearing about Alex's stinking, worthless fish. I'd go for a more in-depth video that focusses on this angle as long as it's in a highly charged tone or my ears remain deaf to it.

As others pointed out, pretty shocked to see yt wasn't set to 150% replay no matter how many times I checked... this guy has overclocked his mouth or sth, some real life hax shit.
 

PaintTinJr

Member
Sure, if someone wants to find out, they can with enough effort, but people are lazy.
DF is an easy target as they are public.


Well that is a complex topic, but for example the designer had to make sure that the player cant see more than X number of BSP and meshes at any time.
So if your limit was say 30 visible BSP, you had to do your leveldesign around that limit which of course often leads to strange architecture or layouts.
One of the most famous things in those older games is the "S" corridor which was basically used to make sure the player couldnt see the previous section so it could be unloaded.
Or for a game like Far Cry instincts you basically had a map made out of 64x64 meter patches and the designer had to manually define which patches would be loaded/visible from the patch the player was in.
Very interesting. What was the cause of the BSP(Binary Space Partitioning) planes limit - in a tree? Or are you referring to tree node count limits inside the main root node tree?

I'm just trying to visualize a scene and the technical issue. Was it something like a technical issue where some programmer had used a 5 bit Identifier for each tree node - which then meant for the active part of the loaded scene, partitioning to make rendering occlusion more efficient to cull geometry from rendering was at odds with improving visuals and to stay within the unique identifier 5bit key space? Or was it something completely different?
 
The guy has actually valid points but his delivery is what people have a problem with and I can see why. I don't mind he taking shots at DF for their coverages at all.

I'm not subscribed to them and never will.
 

Landr300

Member
Some might say it still isnt, but there is no way back
How can you be so sure of that? The devs surely may want that but the ones who ultimate get to decide are the consumers who's really paying the paycheck of the devs

There's no such a thing of "there is no way back" if the alternative is destroying an entire industry and get fired
 
Last edited:

Astray

Member
DF have gone incredibly downhill in recent years, and I've been critical of them on here on a lot of occasions.

But this Threat Interactive guy is an absolute joker and kind of an example of how the YouTube algorithm poisons online discourse.

Did I miss something? Where does all the DF negativity come from?
Constantly creating reasons for negativity in the gaming community eventually catches up with you.
 
Top Bottom