Dolphin, the GameCube and Wii emulator - Forums

Full Version: An alternative to super sampling, MLAA(Morphological AA)
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2 3 4 5 6
Since when is SSAO a filter?

Also I'de like to point out that how good MLAA is will depend on the game, in particular how much contrast their is between objects. In SC2 it looks almost as good as SSAA and a hell of a lot better than MSAA, but with almost no performance hit. It does make text more difficult to read but only at lower resolutions. It seems that the higher your resolution is the less it affects the readability of text. As for the blurry textures, that can sometimes happen but that is an inherent flaw of MLAA that you have to live with no matter how it is implemented. And considering SSAA will cause the same problem with most game engines....

I also read somethign else about MLAA that intrigued me. Many people using ati's implementation of it are reporting that it usually only works with dx9 games. This is odd for several reasons. AMD states that it will work with any dx9/10/11 game, and in theory since it's a post processing filter it should not care which api is being used by the graphics engine. And of course it uses directompute11 which should not be affected by which api the graphics engine is using. Yet many state that forcing crysis or bc2 to run in dx9 mode causes it to work while it does not work if dx10 is being used. I wonder why that is, that doesn't make any sense.

Quote:If you are using rivatuner try overclocking just the shaders clock of your card. It is the only thing that matters with efb scale, a 10% increase in shader's clock can give you a 10% increase in framerate if possible. It will probably make up for those 5 fps that are dropped. Btw with my 9600gt i am getting similar slowdowns with both opengl and directx9 plugin when upping the efb scale, there is no "bug" between plugins.

Then how come going from an 8800GT to a GTX 260 C216 made no difference for me with efb scale at 3x? Considering the shader throughput is double you would expect it to make some improvement if insufficient shader throughput was the problem. I am quite positive at this point that shader throughput has absolutely nothing to do with how much of a performance hit you receieve from efb scaling.

(10-25-2010, 04:23 AM)NaturalViolence Wrote: [ -> ]Since when is SSAO a filter?

Since Nvidia implemented it in the CP as one?
(10-25-2010, 04:23 AM)NaturalViolence Wrote: [ -> ]
Quote:If you are using rivatuner try overclocking just the shaders clock of your card. It is the only thing that matters with efb scale, a 10% increase in shader's clock can give you a 10% increase in framerate if possible. It will probably make up for those 5 fps that are dropped. Btw with my 9600gt i am getting similar slowdowns with both opengl and directx9 plugin when upping the efb scale, there is no "bug" between plugins.

Then how come going from an 8800GT to a GTX 260 C216 made no difference for me with efb scale at 3x? Considering the shader throughput is double you would expect it to make some improvement if insufficient shader throughput was the problem. I am quite positive at this point that shader throughput has absolutely nothing to do with how much of a performance hit you receieve from efb scaling.

Well you can try it yourself. Personally i used 2 games for my testing, mario galaxy 2 (the "mario planet" that you move around with) and
mario kart wii (the "rainbow road" track) since they seem to be GPU limited in my machine. I have a core i7@ 3.2ghz and a 9600GT. An increase to the shader clock in both situations using rivatuner gave me an almost linear increase to the framerate when using a high efb scale.
I will try to post shots later...

Edit:Ok here they are. The first 4 are with default shader clocks while the other 4 with the shader clocks overclocked. Efb scale 2x and 3x respectively by the order they are posted. You get about half of the overclocking percentage of the shaders as an fps increase.

[Image: dwm2010102501243745.th.jpg][Image: dwm2010102501464665.th.jpg][Image: dwm2010102501473337.th.jpg][Image: dwm2010102501541803.th.jpg][Image: dwm2010102501530851.th.jpg]

[Image: dwm2010102501354563.th.jpg][Image: dwm2010102501442874.th.jpg][Image: dwm2010102501433489.th.jpg][Image: dwm2010102501595332.th.jpg][Image: dwm2010102502011373.th.jpg]
Quote:Since Nvidia implemented it in the CP as one?

It's not a filter. It's a rendering technique.

@Link_to_the_past

Well I'm not seeing any difference on my GTX 260 between 1246MHz shader clock and 1400MHz. Yet I do see a drop in performance when raising the efb scale.
(10-25-2010, 10:21 AM)NaturalViolence Wrote: [ -> ]
Quote:Since Nvidia implemented it in the CP as one?

It's not a filter. It's a rendering technique.

@Link_to_the_past

Well I'm not seeing any difference on my GTX 260 between 1246MHz shader clock and 1400MHz. Yet I do see a drop in performance when raising the efb scale.

The increase should be minimal, a couple of fps depending on the game. Above i got a 3fps increase for Mario Galaxy 2 using efb3x (7% increase in performance for a 12% increase in shader clocks). If things were exactly the same as you imply, no matter what card you have you would have the same results. And that is not what i have experienced. A cousin of mine with a worse processor got better speeds than me when efb is set to 3x, because of his graphics card(previous gen 4-core intel at 2,8 ghz vs core i7 at 3,2ghz).

oofloom

I hear this creates problems in motion. "Crawling" on edges and such. Supposedly the God of War devs worked it out though.
(10-20-2010, 08:55 PM)StripTheSoul Wrote: [ -> ]@ Squall: Well, I guess it was another emulation-related forum where I saw you then Wink

He is from NGEmu forums. Also known as the "Enable Triple Buffer Guy".

Sorry, Squall. I'm just kidding. Couldn't resist Tongue Don't take it serious.^^
(10-25-2010, 01:44 PM)rodiabloalmeida Wrote: [ -> ]
(10-20-2010, 08:55 PM)StripTheSoul Wrote: [ -> ]@ Squall: Well, I guess it was another emulation-related forum where I saw you then Wink

He is from NGEmu forums. Also known as the "Enable Triple Buffer Guy".

Sorry, Squall. I'm just kidding. Couldn't resist Tongue Don't take it serious.^^

eh?, im against triplebuffering in D3D as it introduces tearing and anomalies (D3D only supports chained queue and displays every frame regardless of whether its needed or not).
I was refering to the many times I did read your posts at NGEmu suggesting people to enable triplebuffer in many situations regardless the emulator.
Anyway, I was just teasing you. Let it go, dude. o/
And thanks for always trying to help people there. Much appreciated^^
Hahaha, Mr Triple Buffer, huh? Good name Smile
I always enable triple buffer when enabling VSync. Ain't that right?
Nah, anyway, I guess we get a little off-topic here Wink
Pages: 1 2 3 4 5 6