VaM 1.x DLSS 5 in VAM2

[hazmhox]

Hey have you gotten motion vectors to behave properly your version? Without MVs any small movement causes lighting and details to smear around. I have them working 70% of the time for my plugin but 30% of the time some funky scene setup causes them to break in fun ways.

Yo! : )
Nope I'm not talking about smearing... I saw that effect when testing it on VAM2, and indeed that's a problem about motion vectors. In this case, I'm talking about actual lighting computation and AO interpretation. A good test case is just moving towards a character at a steady pace. You see the AO adapt and change over several frames. It's clearly a limitation of the diffusion system, as it probably can't be adequate/accurate at every distance.

I'll post a set of situations you often see if you go further than just slapping the thing on for a quite standard scene.

The conclusion I got to is pretty much "that works well under normal lighting circumstances, anything quite artistically driven immediately creates issues".
Which technically, feels logical. Most diffusion models on people was trained under very standard lighting (and I'm not even accounting for the fact that models have insane biases)... which means that anything non-standard will yield results that you can't control much no matter what Huang wants to believe.

And the settings to control intensity is just some sort of merge of both image pre and post diffusion... at least, its how it feels, which makes the result somewhat strange. (when they could probably have made several different models yielding a less heavy diffusion result).

That's probably a fun tool for a porn game like VAM (at least for people who like images that heavily lean on slop)... for production ready games, that thing is gonna create all sorts of issues sadly. And I highly doubt studios are gonna Q/A the shit out of their games to ensure every single important spot of their games are looking good with and without NR.

 

[perkel]

Question, Considering DLSS5 shennenigans. Is there any plugin that uses window camera that makes full screenshot using camera including post processing ? McGrubber Supershot doesn't take post processing in mind (or at least old version i have). I would love to do proper window camera shots with dlss5 and other post process effects.

Yo! : )
Nope I'm not talking about smearing... I saw that effect when testing it on VAM2, and indeed that's a problem about motion vectors. In this case, I'm talking about actual lighting computation and AO interpretation. A good test case is just moving towards a character at a steady pace. You see the AO adapt and change over several frames. It's clearly a limitation of the diffusion system, as it probably can't be adequate/accurate at every distance.

That's weird line of logic considering how many games work well with it with no issues. Including games that never had motion vectors (dx9 games).

 

[hazmhox]

That's weird line of logic considering how many games work well with it with no issues. Including games that never had motion vectors (dx9 games).

That's not a line of logic, that's actual tests and results. Also, that's not because you don't "see" the problem on a fast moving game without motion vectors, that it does not exists. There is not a single video I saw from every hack used to make this leak work on all sorts of games, that wasn't loosing temporal coherence/stability at some point or another.

But let's assume motion vectors would fix that 100% of time (which I think -that one is subjective- won't)... there is a stack of problems that remains. And as I said later in the post... knowing the industry quite well, deep integration of NR is not gonna be a thing overall.

If you just take Ray Reconstruction and the problems it still creates today... NR is nowhere near a stable tech.

Question, Considering DLSS5 shennenigans. Is there any plugin that uses window camera that makes full screenshot using camera including post processing ? McGrubber Supershot doesn't take post processing in mind (or at least old version i have). I would love to do proper window camera shots with dlss5 and other post process effects.

No.
You need to capture Unity's final image to have the full shot. Reshade and any plugin being an injection after Unity's rendering pipeline, you can't capture the resulting image from inside the game.

Frame your win cam manually or use something like Studio Maestro to frame your shot. Then swap to the wincam with M in desktop mode. Use DSR to push to 4K/8K to "super sample" your game, and capture with pretty much anything you want ( print screen being the simplest approach ).

 

[EvilBox]

Yo! : )
Nope I'm not talking about smearing... I saw that effect when testing it on VAM2, and indeed that's a problem about motion vectors. In this case, I'm talking about actual lighting computation and AO interpretation. A good test case is just moving towards a character at a steady pace. You see the AO adapt and change over several frames. It's clearly a limitation of the diffusion system, as it probably can't be adequate/accurate at every distance.

I'll post a set of situations you often see if you go further than just slapping the thing on for a quite standard scene.

The conclusion I got to is pretty much "that works well under normal lighting circumstances, anything quite artistically driven immediately creates issues".
Which technically, feels logical. Most diffusion models on people was trained under very standard lighting (and I'm not even accounting for the fact that models have insane biases)... which means that anything non-standard will yield results that you can't control much no matter what Huang wants to believe.

And the settings to control intensity is just some sort of merge of both image pre and post diffusion... at least, its how it feels, which makes the result somewhat strange. (when they could probably have made several different models yielding a less heavy diffusion result).

That's probably a fun tool for a porn game like VAM (at least for people who like images that heavily lean on slop)... for production ready games, that thing is gonna create all sorts of issues sadly. And I highly doubt studios are gonna Q/A the shit out of their games to ensure every single important spot of their games are looking good with and without NR.

I agree that there are problems. But saying that it doesn’t work at all, or that the results are terrible, would be lying to ourselves. I’ve noticed a really interesting pattern on Twitch. I remember when DLSS first appeared people used to complain about it constantly. Now if a game launches without DLSS support, people immediately start asking - “Where is it? Why isn’t it there? Where MY DLSS!!!!” well the technology is still rough and clearly needs more work, but it’s definitely not hopeless.

 

[hazmhox]

I agree that there are problems. But saying that it doesn’t work at all, or that the results are terrible, would be lying to ourselves. I’ve noticed a really interesting pattern on Twitch. I remember when DLSS first appeared people used to complain about it constantly. Now if a game launches without DLSS support, people immediately start asking - “Where is it? Why isn’t it there? Where MY DLSS!!!!” well the technology is still rough and clearly needs more work, but it’s definitely not hopeless.

I think you're reversing the feeling. People tend to want this to be nice and throw that reversed way of thinking "you're forgetting that it looks nice".
No not really, I tested over 30 scenes and more than 50 static scenes (for beauty shots and whatnot) I made over the years to test drive the system. I also tested a dozen standalone games. I'm looking at this from an artistic approach and a tool that I could/might use in my line or work. The results I'm seeing are just sending me a huge warning on the additional work it would require for this to work as I want.

If you want another way of explaining that: look at some games that feature raytracing or lumen in UE5 games. The moment you don't have RT enabled, the games looks like ass no matter what you do... because studios don't / can't produce a solid homogenous result for both "with and without" the tech. NR would yield similar flaws... and even induce lazyness (which is already massive due to execs strategies nowadays) from the dev team... to let that stuff lift the end result.

Which transition quite nicely in what you're saying about DLSS: if we remove from the equation at the time... people bitching for the pleasure of bitching, and take actual solid arguments about DLSS detractors when it got announced, the main and most valid one was "games are already not really optimized, this is gonna make things worst". I was one of these people. I was happy for what the tech could bring for bottlenecked situations like raytracing or huge resolutions, but quite scared of the industry impact it would have.

And here we are, in 2026: people don't whine when DLSS is not here "because it's cool and we changed our minds". People whine when DLSS is not here because most games runs like ass. DLSS is lifting the inability of most games to run at a solid base framerate without it.

I don't know if you play a lot... but I do. If you objectively look at games today, they look overall worse if you don't have proprietary techs trying to fix the output. You always have perfs issues at release. And the overall latency on display as increased a bunch, you know the whole adaptation effects, noise, smearing, ghosting and so on... thanks to all the temporal methods used because the hardware can't keep up with what we throw at it.

Yes, some results with NR are nice looking, I never said otherwise : )
But it would be nice to not invert the problem. People have the right to enjoy this to fap twice a week. But these people also can be objective and accept facts and technical limitations of said tech.

Said with an analogy: anybody has the right to enjoy McDonalds and say they do. But no matter how much they like it and will try to defend it... it will never make McDonalds an healthy choice of food.

 

[perkel]

That's not a line of logic, that's actual tests and results. Also, that's not because you don't "see" the problem on a fast moving game without motion vectors, that it does not exists. There is not a single video I saw from every hack used to make this leak work on all sorts of games, that wasn't loosing temporal coherence/stability at some point or another.

But let's assume motion vectors would fix that 100% of time (which I think -that one is subjective- won't)... there is a stack of problems that remains. And as I said later in the post... knowing the industry quite well, deep integration of NR is not gonna be a thing overall.

If you just take Ray Reconstruction and the problems it still creates today... NR is nowhere near a stable tech.

I think you are throwing out baby with bathwater here. RR is CRITICAL for pathtraced games. The alternative is smeary pathtracing game aka looking bad.

Same story here. From my testing DLSS5 scene beats any day of the week non DLSS5 scene simply because VAM1 lighting engine is super crude and DLSS5 fixes a lot of those issues. The alternative is having crude graphics.

Or VAM2 with pathtracing.

 

[EvilBox]

People have the right to enjoy this to fap twice a week

I understand your concern! I look at this myself and think. Okay, what the hell would I actually need to do to make this work really well in one of my own projects or games? quite a lot of work.
And yes, I don’t play that many games. The last one I played was Resident Evil Requiem I really liked the way DLSS and rt were used there. There are games where the rasterized version already looks excellent and RT only improves certain aspects. And there are also games where the lighting really does fall apart without RT. It all depends on how the game was designed and built in the first place. I’ll be honest about what I think: soon we may not even have a choice between healthy food and McDonald’s there may only be fast food left. we’ll be lucky if we can even afford that. PC hardware prices are already astronomical, and it’s entirely possible that owning a decent personal computer will become a luxury in the near future.

 

[perkel]

this is the difference i am talking about:

 

[hazmhox]

I think you are throwing out baby with bathwater here. RR is CRITICAL for pathtraced games. The alternative is smeary pathtracing game aka looking bad.

It is not. And half the time it's breaking some critical shaders. I had no massive visible difference in Alan Wake 2 with RR, but I had critical shaders not working, particles disappearing, and volumetrics behaving inconsistently.

When one thing is fixed on one aspect (let's say some smearing on surface normal animations) another breaks somewhere.

And for what it's worth, it's validating what I was implying above: the more we go forward, the more the GPU spends time fixing shits it broke in the first place. Smearing and ghosting is induced by technics used to allow the hardware to run the sort of techs it can't (ie raytracing), then you slap more techs to fixe these.

Anybody not having these techs, or these GPU gens, or whatever driver version locked behind a specific generation... will have games that look like ass... because, reasons?

Same story here. From my testing DLSS5 scene beats any day of the week non DLSS5 scene simply because VAM1 lighting engine is super crude and DLSS5 fixes a lot of those issues. The alternative is having crude graphics.

I'm talking in a general manner since the beginning, I'm talking about the tech itself. VAM is an old engine and an old lighting method... this is just a fancy "reshade" type of approach improving the result. Again, not against the idea... i'm pointing out the problems of NR.

You're again proving the point I'm making at the end of my last answer: you're ignoring all the things I'm saying just to fall back on the "yes but my VAM 1 looks good"... yes yes, we understood that, NR lifts the oldish render of VAM 1, ok. This does not make the problems disappear : )

There are games where the rasterized version already looks excellent and RT only improves certain aspects. And there are also games where the lighting really does fall apart without RT. It all depends on how the game was designed and built in the first place.

True dat! The industry has widly different result depending on the games.
Capcom being an expert at this, with their last Monster Hunter or Dragon's Dogma running like ass and looking bad... when the RE are absolutely beautiful ^^

I’ll be honest about what I think: soon we may not even have a choice between healthy food and McDonald’s there may only be fast food left. we’ll be lucky if we can even afford that. PC hardware prices are already astronomical, and it’s entirely possible that owning a decent personal computer will become a luxury in the near future.

True again. Hoping datacenter dies ; )
Owning computers is already a luxury... the ram I got 6 months ago got from 750 to 2200 where I am. Nobody can afford that

 

[EvilBox]

I don’t know where this will all end up. I hope the technology keeps improving. And if you manage to make something like this work with VaM2 I think it could definitely help the game attract a bigger audience. It would be another big point in its favor. At the same time I understand that this is a hell of a lot of work with a lot of unknown variables. Ideally it would be great if we could somehow mark assets in a scene and decide what should be processed by dlss and what should be excluded from it. I also want to point out that facial expressions on characters can already look really impressive in VaM1 with this. I don’t really play VaM my entire gameplay basically consists of building a scene and setting up a couple of poses. But with this thing pine cone started smoking...
Anyway, even if you decide not to use dlss in VaM2 someone from the community will find a way to bolt it onto the game. Then why should devs worry? xD
I like the way emotions look:

Spoiler: I like the way emotions look:

You can throw x100 tomatoes at me if you want. right now I see more positives than negatives with dlss. Feel free to call me a neural slop eater I’m fine with that. Maybe it’s just the wow factor talking. I don’t know

 

[MegaDez]

Is it possible to make unity work with dlss 4.5? Or it works only with 5 version?

 

[MrGabbe]

As soon as I start the game after setting it all up I get an error saying: Runtime error, this application has requested the runtime to terminate it in an unusual way. Its only happening when the Feeder and Neural Rendering are on together. Its with the LumeniteFX-mainline methood not with the asian made Bridge methood..

 

[EvilBox]

As soon as I start the game after setting it all up I get an error saying: Runtime error, this application has requested the runtime to terminate it in an unusual way. Its only happening when the Feeder and Neural Rendering are on together. Its with the LumeniteFX-mainline methood not with the asian made Bridge methood..

You don’t need feeder or ReShade at all. It’s a standalone hooking tool completely separate from ReShade. DLSS works on RTX 50-series and RTX 40-series GPUs. It may also work on older cards but I can’t confirm that.
1. VaM DLSS NR and DLAA Plugin by UncleBurrito - VaM DLSS NR and DLAA Plugin
2. BepInEx 5 x64 (BepInEx_win_x64) and extract it directly into your vam root folder, where VaM.exe is located. (You’ll find the github link on the plugin’s page)
3. Extract the DLSS plugin archive into the VaM root folder as well.
4. Launch VaM and press F10 to open the DLSS control panel (both DLSS/DLAA are disabled by default so enable them manually)