Marty McFly's Ambient Obscurance (MXAO) with IL

Anything that uses last frame data is hard to do with AFR SLI, since the last frame hasn't finished rendering when the current frame starts on the other GPU. ReShade itself will likely not work correctly with SLI, since there is no explicit resource management for two GPUs, so it will likely work on only one GPU and miss all the frames the second GPU renders or fails to do anything with them since they are located in different memory.

crosire wrote: Anything that uses last frame data is hard to do with AFR SLI, since the last frame hasn't finished rendering when the current frame starts on the other GPU. ReShade itself will likely not work correctly with SLI, since there is no explicit resource management for two GPUs, so it will likely work on only one GPU and miss all the frames the second GPU renders or fails to do anything with them since they are located in different memory.

I've used Reshade with SLI for years now and haven't had any issues related to it.

I'm totally out of the loop here, I've never used SLI and can't say anything for certain. Easiest way to find out is using the motionblur shader from the old ReShade Framework and test it. If it works, TSS works. Either way, if you have a SLI setup, you hardly need such performance tricks.

Marty McFly wrote: I'm totally out of the loop here, I've never used SLI and can't say anything for certain. Easiest way to find out is using the motionblur shader from the old ReShade Framework and test it. If it works, TSS works. Either way, if you have a SLI setup, you hardly need such performance tricks.

I only use MXAO in older games, which I also downsample at really high resolutions, 5K to 8K, reducing the resolution scale for the AO helps with performance but it starts getting really noisy and ugly, so this would help a lot.

Also in general Reshade shaders, even those that don't use depth buffer, seem to have a larger performance % impact at these resolutions, I don't know why, I am assuming it's because they are also being rendered at the game's res.

Zireael wrote: I only use MXAO in older games, which I also downsample at really high resolutions, 5K to 8K, reducing the resolution scale for the AO helps with performance but it starts getting really noisy and ugly, so this would help a lot.

Also in general Reshade shaders, even those that don't use depth buffer, seem to have a larger performance % impact at these resolutions, I don't know why, I am assuming it's because they are also being rendered at the game's res.

Obviously all effects are rendered (if rendered to backbuffer) in the original game's resolution. Those resolutions have huge pixel counts..

Marty McFly wrote:

Zireael wrote: I only use MXAO in older games, which I also downsample at really high resolutions, 5K to 8K, reducing the resolution scale for the AO helps with performance but it starts getting really noisy and ugly, so this would help a lot.

Also in general Reshade shaders, even those that don't use depth buffer, seem to have a larger performance % impact at these resolutions, I don't know why, I am assuming it's because they are also being rendered at the game's res.

Obviously all effects are rendered (if rendered to backbuffer) in the original game's resolution. Those resolutions have huge pixel counts..

Is it possible to have a resolution scale option that works across all shaders?

If so, this would require modifying ReShade itself. Crosire?

Just render to a texture with a lower resolution and as a final pass blit that texture to the backbuffer. One can already do that.
Making a global option which does that internally is certainly possible, but not something I'd invest time into right now.

Well, Zireael wanted this globally for all shaders - that'd require to edit all of them.

crosire wrote: Just render to a texture with a lower resolution and as a final pass blit that texture to the backbuffer. One can already do that.
Making a global option which does that internally is certainly possible, but not something I'd invest time into right now.

I understand that, if you could just keep this in mind and do it at some later time, I would greatly appreciate it, as would many other users I'm sure.

Improved TSS technique, here's a pic with TSS - 8 samples:

And no TSS, 48 samples:

As you can see, with TSS you can get the same quality at a fraction of the taken samples, increasing perf immensely.
Whenwegetthis: very soon, as soon as I updated the doc thorougly.

Looks great, can't wait. Could this be sufficient to create a Temporal SMAA shader?

do you can add transparency and color saturation for MXAO from adapt from all arts of games ??

Not sure I got what you mean?

@ivanosky: not really. I mean, I guess you could jitter samples per frame and interpolate but then you'd get some motion blur-esque effect and the AA benefit would only be visible on static scenery. I never really worked with 2D stuff so I can't say for sure.

add color saturation and transparency from this shader because of many different art of games

Awesome. Thanks working on the docs. It really helped me out before. Looking forward to try the new functions

Sooo, the time has come to make it public. I'll push it to the github repo within the next days.
Behold, MXAO 3.0:

www.mediafire.com/file/zaoo6yy5b7y2hq5/MXAO30.zip

Ow sucks. Going on a airplane and have to install a new GPU. But most definitely will take it for a spin when I get all stuff sorted. Very interesting with temporal sampling )

Hmm. There is a significant performance increase, but only depending on how you set it up.
I wanted to squeeze the most performance out of it, and it's pretty affordable now.
[strike]However, seems like smoothed normals break on 4K once again (as in, barely smoothed).[/strike]Ok, apparently it just wasn't as noticeable before, for whatever reason, but it also existed.
Also, unless I go Ultra or Maximum quality, everything seems very blurry. Up until Medium, there's barely any difference in performance, and it fixes some flickering.

EDIT: TSS really does not do well with my monitor and its "response time image optimizer thingy". Ghosting becomes very visible in dark scenes.

ShoterXX: Ah yeah, I made normals searching depending on pixel size. Will fix it, no worries. About blurryness, if you don't have debug mode enabled, you'll hardly see a difference but areas where the blur algorithm fails, you see it immediately. Hence I opted for worse visuals in debug mode and better ones in regular gameplay. I made a lot of quality options on purpose, it depends on video card how much difference there is. Also, I lowered the blur sample count for higher AO sample counts, so you have the least amount of blur for every sample option. TSS is more of an option for lower end users, if you can go 4k I doubt you'll ever need it.