Yeah, it’s probably that “scratches layer” that have glitched out there and used another texture entirely or something…
I wonder if this was caused by Special K’s texture cache… I can’t actually reproduce it though, as it’s just the regular scratches layer now after a game restart.
I can add some additional monitoring of Render Target Views on the swapchain backbuffer when I apply the flip model sRGB-off hack. D3D11 can create an sRGB view of a non-sRGB buffer, which is the appropriate behavior.
Most engines that use an sRGB swapchain are smart enough to be creating sRGB views, which means the change in gamma doesn’t matter. It’s only the dumb ones that are creating views without asking for any format. When no format is asked, Windows picks the closest match.
There’s an off chance this will fix-up Watch_Dogs 2,
SpecialK64.7z (7.6 MB)
It certainly didn’t fix the problem in Path of Exile, but to be honest Path of Exile’s problem is reverse. Normally disabling sRGB for flip model causes a game to go much brighter than it should be.
This version includes the new HDR presets and support for VLC, BTW.
Yeah, working for me, i had to use my controller to prompt the control panel, perhaps VLC hotkeys are interfering here?
Looks really great, i just think shadows need to be a touch higher since some stuff is hard to see. I messed around with Nvidia and VLC’s colour settings, but it’s specifically the shadows that need to be brought up just a touch, rather than the entire image’s contrast being reduced. That being said, i don’t know if dark shadows are actually a good thing in my case, since it helps make the HDR more pronounced on this 350-400 nit panel.
Edit: I have an endless amount of footage to test, lol. The more i’m testing, the more i’m split on whether the shadows are handled correctly or not. It’s still a tad dark for some footage (which are colour corrected properly), but for other footage it actually works right, and there’s also a good distinction between shadows and highlights. Hope i don’t sound confusing XD
Assuming you still have the original SK trailer file, you can check for yourself and see if the shadows are a bit too dark for you too, it’s definitely dark for me in some areas, particularly the beginning.
I just realized I’m a dumbass 
I’ve been approaching this framerate limiter thing all wrong, in the sense that I was going to release it open source and hope someone builds a tool like RTSS to make applying my limiter easier.
I still will release an open source standalone implementation, but the far simpler solution is just to hand it over to Crosire for integration into ReShade.
People with intermediate technical skills perhaps only marginally more demanding than running RTSS use ReShade all the time, distributing as part of ReShade would be an easy way to get the world’s most advanced framerate limiter into the hands of everyone 
I’ve had a change of mind after looking at the performance of my limiter in OpenGL. It still smashes the competition without all of the intricate swapchain tuning that can be done in DXGI graphics APIs, so I have an idiot proof product already and it’s time to just start getting it in the hands of users.
ReShade with a framerate limiter, well it already has a number of additional overrides so it seems like a good addition if that works out.
… wow… A reason for me to actually use ReShade for once…
To be honest, ReShade has a lot of uses
I’ve seen a lot of people build proof of concept tools using it, since it’s a pretty simple injection base that covers a bunch of graphics APIs. That’s what I see as being the primary benefit here, I can experiment with integration into other products and find any loose ends.
Because ultimately I want this thing to be available to both end-users and something that port developers might license rather than roll their own abomination of a framerate limiter that I may very well have to patch their game to remove
This article is stupid…
https://www.thegamer.com/crysis-remastered-graphics-card-4k/
My RTX 2080 Ti KINGPIN runs the game at about 50 FPS in Can it Run Crysis Mode at 4K. The game looks terrible frankly, I expected for the performance cost it’d be wowing me a little more, but they just went out of their way to make a game that’s difficult to run for the sake of continuing a meme.
The original game had great visual quality when it was released but the selective updating and what appears to be a more stylized or filmic look over the more natural though in comparison perhaps a bit duller looking visuals also clash a bit.
I think some of the comments I’ve read put it well it’s a 2007 game with updated lighting and shadows but the selective updating of assets and textures plus the performance problems are dragging it down.
The excess of some of the highest PC settings could have been better balanced too but maybe the preset a step down already does a good job with that without 8192x8192 shadow resolution and a near removal of LOD levels.
Plus the 8192x8192 texture resolution for the nanosuit which barely shows up at all after the lads all disperse past the first map leaving the player to look at his 8k arms for the remainder of the game.
EDIT: Well at least the MP outfits kinda resemble uniforms I suppose.
EDIT: The engine itself is still really impressive however, shame Kingdom Come couldn’t get the newer Cry Engine V code base but maybe the sequel can since it’d be a huge undertaking remaking much of the data.
Beta version still holds up with it’s usage of light and shadows and the woods in particular though said beta is very crash prone but the full game also holds up well even if scaled back somewhat.
Crytek and their own work with Hunt Showdown probably as a even bigger tech showcase for the engine itself.
Wonder if Crysis 1 Switch was the primary target here and then updates and releasing that version (Based on the already somewhat updated version on X360 earlier.) became a bit of a extra though delayed slightly due to criticism.
But as probably could be expected a month or two isn’t enough time for a more thorough overhaul.
Hopefully some code fixing and optimization can be done at least.
In all seriousness though, the game really needed a few more months in development, but ig next-gen consoles releasing would’ve stolen the spotlight. I think this looks great for a remaster, outside of some non-updated textures, but it’s definitely held back by not being a remake. Worth noting they somehow pulled off RTX alternatives in real-time and API-agonist([ish) - you have SVOGI which is similar to Metro Exodus’s raytraced GI, raytraced reflections, and then there’s the shadow coverage which is similar to Modern Warfare’s raytraced shadows. Of course, none of these looks as clean as the RTX versions, but it’s impressive they can all run in real-time on consoles.
I think that was part of the goal though, to just get Nvidia’s RTX features running in real-time all at the same time somehow, that’s the impressive feat of this remaster. But outside of that, some of the 8k textures are good, but there’s still a ton of super low res textures. I feel hypocritical pointing this out, considering my current texture packs still have some textures left to cover, lol.
Plus why did they go with basic lightrays that we got from PS2 games? I’m guessing full volumetrics was too difficult to pull off appropriately on a per area basis. I think if they sorted that out, updated some more textures, redid the meshes of weapons, increased the polycount of your fps hands you see all the time, and updated their few character models (even if just their body and textures), then it would’ve looked fine.
That being said, i’m convinced that if they had the funds to make a Crysis 4, it would’ve looked seriously mindblowing, because the assets and fundamental art design wouldn’t hold it back.
I’m probably the only person that’s disappointed by the remaster because of the AI, not the graphics. The original game was always held back by the enemy AI for me, they just have eyes on the back of their heads as soon as cloak mode runs out, and can track you crawling through grass from half a mile away. And then sometimes they don’t even shoot at you when you’re right in front of them, it’s really awkward.
Well, technically a lot more are saddened by the AI — if only because it seems they didn’t properly fix the multithreaded issue with the original design of the game, so AI and a few other CPU heavy workloads are still seemingly only limited to three threads or so…? 
That’s what Fortnite just did I think, takes a big bit of performance to run the shadows (+AO), reflections and GI but it looks pretty nice though until more users have 3070’s and above the performance hit is pretty noticeable even with also adding DLSS support.
(UE4 has it’s own TAA upscaling algorithm in 4.19+ but it’s pretty rough in motion albeit better than regular methods but still not something I’d recommend.)
Probably going to be a tech demonstration for the Ampere and on NVIDIA cards or maybe that’s Unreal Engine itself going forward particularly the new 5.0 version.
EDIT: Plus that’s just the start similar to Rainbow Six Vegas or Roboblitz as the earlier UE3 titles.
Daylight as the earliest proto-UE4 game.
EDIT: For Fortnite the reflective eyes is a nifty little touch too.
EDIT: Though maybe not worth halving the framerate over. 
I struggled to see any improvements besides reflections in the trailer, i don’t know if it’s just the art design of the game that stops it from really benefiting from raytraced effects, at least in my eyes.
Going by the UE5 demo on PS5, i’m staying well away from that. 1440p upscaled to 4k according to DF, and the artifacts are horrible in motion.
Reflections are what I expect to see as the focus in particular, GI’s secondary lighting pass is nice but very subtle unless you compare.
Shadows actually get softer though the grounded ambient occlusion is a step up even over voxel methods like NVIDIA’s attempts with VXGI or Crytek with SVOTI/SVOGI
EDIT: Just need to get the developers to get materials a bit more tuned in not make everything 100% mirror or chrome shine.
Plus some of the Fortnite comparisons seem to mix ray-tracing with cube mapping as well.
Wonder if the extra drawcalls make the remaster even more demanding on the CPU than the original game, in that case.
One reason a native Vulkan or D3D12 implementation could have scaled a lot better.
If it had been implemented well and utilized fully that is.
Think reflections are also only used on plain surfaces and not water because the displacement is very costly to use this implementation of ray tracing against in addition to other trade-offs also used here like the coarseness of the effect a bit akin to path traced Quake 2
Though not that extreme plus path tracing is even costlier so that working at all is really impressive despite the more simple implementation.
EDIT: Q2VKPT not Q2 RTX that is.
Think that’s part of the RTX and hardware edge allowing computation of this along with de-noising.
With this new generation of RTX cores being more effective.
But I don’t know how well you’d be able to compare that, maybe in a pure ray-tracing / RTX tech demo the 2000 and 3000 series and potential “up to” hardware gains could be measured easier.
(Games using the effect as a extra would have the other rasterization performance and possible bottlenecks to the overall results and not all games or game engines scale well.)
EDIT: Suppose if Bethesda and id Software remembers that bit about DOOM Eternal ray tracing and implements that there would be one title with good hardware utilization and performance scaling to use for testing though.
I’m just someone who’s obsessive over shadows, lol. I just want shadows to be done right, i feel like it has a long way to go and would help pave the way to photorealism. I’ve even tested Shadow of the tomb raider with raytracing, and it just feels like PCSS with a bit more shadow coverage.
I want to see shadows where even the smallest blade of grass and a bump in the ground casts a shadow, and different lights conflict with each other, causing meshes to cast multiple shadows (even on top of each other), and have shadows not always being one set amount of darkness all the time.
I honestly think i’d sacrifice decent AO for truer to life shadows, that’s how much i care for realistic shadows. Of course, real shadows would require high polycounts to work with, Modern Warfare 2019 suffers from artifacts because of the polycount holding the shadows back. Rant over, lmao.
Doom Eternal could benefit from even basic raytraced shadows i’d say, there’s barely any shadows in the first place such as on your fps hands and weapons, and the enemies are missing a lot of smaller self-shadowing. Modern Warfare even without raytracing actually does a good job with the weapons shadows etc, it’s kind of what Doom Eternal is missing.
You can see the shadow difference on the weapons here, Doom Eternal would benefit from this sort of upgrade.
Doom
Oblivion E3 compared to launch and the lack of most real time dynamic shadow casters.
It’s probably always going to be about tricks and how to make mostly accurate shadows with good performance although since the early attempts with stencil shadows and initial pixel shader shadow effects it’s gotten fairly good but yeah there’s usually some trade-off.
Witcher 3 and many other open world games sacrifice a lot of dynamic shadow casting sources or fake it with blob shadows or spotlight shadows during cutscenes and what not plus usage of gradually lesser detailed cascading shadows to attempt for more distant shadow coverage even if it’s at a very low resolution.
Video game and graphical effects and how that compares to the real world I could see ray tracing work well here but since it’s going to be similar to HFTS the softer look might make people think it’s lower quality because the common is these hard lines with gradual softer edges.
Vegetation shadows as another common issue particularly smaller flora though vegetation itself is a whole other area that video games have had to balance between realistic and performant.
Crysis despite it’s age still has that clunky but visually effective physics system for vegetation and a effective destruction system as well even if it’s based around segments or parts but many other games omit that entirely or use static vegetation even some decade on and a few more years still.
DOOM Eternal I believe is also a case where it’s clear the framerate and 60 FPS was a important target so the engine is really good and it performs incredibly well but it’s clearly focusing on speed and optimized for maintaining that 60 FPS over additional visual effects.
Guessing COD is the same but the PC ports have been all over the place though I don’t know much about the last four (five?) games since Activision split from Steam with these titles.
(I would guess the MP levels in particular are also more sparse to keep framerate high.)
