Nixxes graphics programmer weighs in on how easy it is to add DLSS, FSR, and XeSS to a game. Says there is no excuse not to add them all.

giphy.gif
 
Maybe AMD is helping them out, and any optimization that they come up with would be conpatatble with the console version.
 
Maybe AMD is helping them out, and any optimization that they come up with would be conpatatble with the console version.

No it sounds more like Company A is paying money to keep Company B's reconstruction technique to not being used.

We are now living in the times of exclusive reconstruction tech. :messenger_crying::messenger_crying::messenger_crying:
 
I got my first gaming PC a few years back and i was pretty shocked by how many games lacked dlss. I just assumed every game would have it since it is such a no-brainer.
 
Programmer from dev house that only does performance optimization and never produced a single game in their lifetime claims that performance optimizations are easy.


Wow.
 
Programmer from dev house that only does performance optimization and never produced a single game in their lifetime claims that performance optimizations are easy.


Wow.
I don't think he said anything about optimization, just how easy it is to add certain features.
 
I have been desperately trying to add DLSS to RE2 to no avail. The difference between console and a pc is night and day in Village and RE4 after DLSS, RE3 also looks Fantastic in motion although we do have that specular highlight problem.

Jedi Survivor is even the better example, DLSS just shits over FSR and console FSR is just so bad it makes me question is it really improving the game? With UE and bad CPU optimised engines Consoles are really getting dogshit image quality in name of better RT features. Naughty Dogs and Insomniac have their own upscalers and to me they look better option than FSR, maybe other dev should look into tweaking Unreal's TSR for custom sause.
 
Last edited:
Strange to make such a statment like that if you never did a new game. Getting an already project done from others companies IMO is the easy part. Is like copy/paste homework from others.
Why is that strange? They have to add the support for the different solutions, wich they supposedly do by using their own wrapper between the game itself and different API's. If they developed the game or not has nothing to do with how/if you implement DLSS/FSR etc.
 
Strange to make such a statment like that if you never did a new game. Getting an already project done from others companies IMO is the easy part. Is like copy/paste homework from others.
Well, technically most PC ports are just that ports. The actual game dev is done on consoles and then a port is made for PC. At that point, the effort becomes basically what Nixxes is doing. You are taking someone else's code and implementing PC specific features which are fairly easy to implement from the sound of things. There are APIs you can use to plug into these features. You dont have to start from scratch.
 
Last edited:
No it sounds more like Company A is paying money to keep Company B's reconstruction technique to not being used.

We are now living in the times of exclusive reconstruction tech. :messenger_crying::messenger_crying::messenger_crying:
Well I was hoping Bethesda was getting more then just money with the AMD deal, so It would suck a little bit less for somebody at least.
 
This wouldn't be an issue if FRS wasn't complete trash. Amd should be dumping money to improve their tech and not blocking others.
 
"All three APIs are so similar nowadays, there's really no excuse."

So why does FSR lag behind XeSS and DLSS in image quality and performance?

FSR 2.2 is already good, much better than FSR 2.0. But that last ML pass gives XeSS and DLSS an extra polish and stability.
 
Temporal upscalers are all about performance optimization.
FSR and DLSS is not manual optimization a programmer does, he just adds an api that does the job.

"All three APIs are so similar nowadays, there's really no excuse."

So why does FSR lag behind XeSS and DLSS in image quality and performance?
They're very similar in their implementation, programming-wise.
 
Last edited:

I've just asked a friend who's a rendering/graphics engineer at a big 3rd party publisher about this topic and here's the gist of the convo:

Him: Yes, I don't know. It's mostly the same amount of work.
Actually, it might be more straightforward to integrate DLSS than FSR.
DLSS is a black box. You give some resources to an SDK and get a result.
With FSR, you have to manually integrate the shader into your pipeline. Depending on the engine, it can be more complicated. But again, if there is a difference in the effort required for integration, it is negligible.

Me: I see. There's been a huge debate lately because AMD is the exclusive partner for Starfield, and many people are afraid that the only available uspcaling technique will be FSR, which means no DLSS.

Him: Also, that person is borderline unhinged.

Him: If it comes to that, it's about deals, not about difficulty/time/budget.
Partnered doesn't necessarily mean deal. At *censored*, we've had partnered IHVs, and generally, that means they invest their time to help us with best practices.
Occasionally, Nvidia would suggest that we implement one of their features, like how we had turf grass in *censored*.

Him: Maybe they also gave money to have their logo appear or whatever, but we were interested in insider knowledge of hardware. It's super helpful to have an AMD/Nvidia engineer look at what you're doing in code and tell you that there's an alternative that would be more optimal on X hardware.
 
Strange to make such a statment like that if you never did a new game. Getting an already project done from others companies IMO is the easy part. Is like copy/paste homework from others.
They would be adding DLSS, FSR, and XeSS to games that didn't have any of that though so they have experience.
 
They would be adding DLSS, FSR, and XeSS to games that didn't have any of that though so they have experience.

Is not the same as making a new game from the ground dude. The time you take for making a port is very different from making a new IP/making a sequel.

Why is that strange? They have to add the support for the different solutions, wich they supposedly do by using their own wrapper between the game itself and different API's. If they developed the game or not has nothing to do with how/if you implement DLSS/FSR etc.

Is strange for me. Like, saying is 'that simple' while you never did a new IP before, just getting others people work and them port to others plataforms.
Games are getting like 5 to 6 years to make, thats why I find strange such a statment as 'There is no excuse'. In fact, there is no excuse for example to make a game running at 30 fps on next gen machines, but here we are.
 
Last edited:
Is not the same as making a new game from the ground dude. The time you take for making a port is very different from making a new IP/making a sequel.
Ok, but why is making a new game a prerequisite to knowing how DLSS, FSR or XeSS is added to a game?
 
Is it also easy to add ray tracing?

I'm of the opinion Alan Wake Remastered should have had ray-tracing. Does not make any sense whatsoever why IT of all games didn't.
 
Last edited:
Ok, but why is making a new game a prerequisite to knowing how DLSS, FSR or XeSS is added to a game?

Because you don't even know what engine you gonna use to begin with while the game is in development phase, while porting a game is already decided.
Remember Kingdom Hearts 3? This game took like 6~7 years to make and in the middle of the development they give up using Luminus Engine in favor of the Unreal Engine.
So for him to say 'is easy' and 'there is no excuse', from a new IP perspective is wrong. Maybe for remasters/ports could be because take less time. He is like walking into the 'lazy dev' argument, because if is that easy why FF devs choise to use FSR 1.0?
 
Because you don't even know what engine you gonna use to begin with while the game is in development phase, while porting a game is already decided.
I'm confused what you're trying to say. Wouldn't this be even more impressive? The fact that when porting a game it could be in any engine you're not familiar with and still add these features.
Remember Kingdom Hearts 3? This game took like 6~7 years to make and in the middle of the development they give up using Luminus Engine in favor of the Unreal Engine.
So for him to say 'is easy' and 'there is no excuse', from a new IP perspective is wrong. Maybe for remasters/ports could be because take less time. He is like walking into the 'lazy dev' argument, because if is that easy why FF devs choise to use FSR 1.0?
Well, I suppose but this isn't about performance choices of using FSR1 vs FSR2. I don't see the relevance of the timeframe to create a game from scratch and being able to comment on how easy or difficult it is to add any of the super sampling methods to a game.
 
I don't care if its easy or hard, if you don't add the best tech in your game which you ask 80€ for than you can fuck off. Stupid lazy money hungry devs.
 
I'm confused what you're trying to say. Wouldn't this be even more impressive? The fact that when porting a game it could be in any engine you're not familiar with and still add these features.

Vary from company to company. He can't say that as a well know fact.

Well, I suppose but this isn't about performance choices of using FSR1 vs FSR2. I don't see the relevance of the timeframe to create a game from scratch and being able to comment on how easy or difficult it is to add any of the super sampling methods to a game.

We can only conclur as soon as the Final Fantasy devs are open to talk about. FSR 2.0 > FSR 1.0, both in perfomance boost and image quality, but if the reason they used the FSR 1.0 is because of the time constrain between the release date?
 
Last edited:
Seems that the reason why we dont have FSR,DLSS and XeSS on all games is because companies pay to block the technology off:



It's funny how PS4/5 games on PC do a better job supporting a lot of current tech like DLSS, FSR and XeSS then a lot of recent Xbox pc games.
 
"All three APIs are so similar nowadays, there's really no excuse."

So why does FSR lag behind XeSS and DLSS in image quality and performance?
He meant that their IMPLEMENTATION are really similar, hence why they made their own wrapper around all of them. The wrapper takes the same call from the game and then "translates" it into the selected API (DSSL, FSR, etc.)
That way it's much easier for them to implement all of them in the same game.
 
Programmer from dev house that only does performance optimization and never produced a single game in their lifetime claims that performance optimizations are easy.


Wow.

Oh yea, what do they know

but ToTTenTranz ToTTenTranz knows

Spider Man Lol GIF


PureDark modder can implement DLSS in any games that has TAA




Indie guys had implemented DLSS hours after SDK was out



Very difficult, only top dev studios can comprehend an SDK after you made the motion vectors like 100% of modern games for TAA.

Fitting avatar for you

813476.jpg
 
Top Bottom