Add FXAA support to OA and Arma3- Many AA bugs fixed

[maddogx 13]

I agree with PuFu, AToC should be set to 0 by default.

Performance comes first. People wanting to tweak their visuals can be expected to look around for config tweaks like AToC, but the average user shouldn't have to fiddle around with hidden config settings in order to get the best performance.

[sickboy 13]

Disable ATOC and FXAA (once implemented) out of the box

[PuFu 4599]

cheers SIXboy :P

[sickboy 13]

cheers SIXboy :P

Who's boy? Sick's ? :D

[-=seany=- 5]

ATOC should be disabled by default, I agree. But FXAA was already going to be disabled by default so it's no big deal really. Also FXAA does not have nearly the same sort of visual side affects and perf issues that ATOC has.

I'm fine with anything being controlled by command line only (at least for a while). The main thing is not to have these things enabled by default. Again, we already know FXAA will be disabled by default, so no big deal.

ARMA 2: Operation Arrowhead

setting not available in UI. Default is Disabled ( FXAA=0). enables Anti-Aliasing technique via post processing

0 - Disabled

1 - FXAA_QUALITY_PRESET_10

2 - FXAA_QUALITY_PRESET_11

3 - FXAA_QUALITY_PR......

[BasileyOne 10]

yup and MLAA support too, probably ;)

[bangtail 0]

I posted this in the Troubleshooting thread but it seems relevant here as well :)

http://forums.laptopvideo2go.com/topic/29042-v29518-windows-7vista-64bit-asus-mobile/

295.18 drivers that allow you to force FXAA through the Nvidia CP.

[langgis08 10]

thx for this nice config tool :)

A2FXAA V0.22b works @my side after having updatet directX (in the first run my system missed D3DX9_43.dll, like it was told in some posts above ... so I knew what I had to do ;) thx)

It runs well here @vista 32, screenshots are being produced well, also.

One little thing, maybe of some interest: I have to close ATI Tray Tools (Radeon 4870, single gpu) in order to get the tool work @XP, otherwise I get CTD (w/o on screen error message) during game loading screen the first 5-10 seconds.

Without ATT (then having CCC loaded instead) it runs as well as @vista. So if someone suffers from instant CTD while loading -> troublemaker could be ATT (XP + ATT + A2FXAA => CTD, at least for me ;) ).

[BasileyOne 10]

maybe. ATT had bogus kernel-level driver, which causes hangs, alters haddrives MBR and tamper OS security policies processing/function as well as with backup stuff.

[MavericK96 0]

Looks like there's an SMAA injector now...

http://mrhaandi.blogspot.com/p/injectsmaa.html

Gonna give this a try when I get a chance. Supposedly has very little performance hit like FXAA but without the blurring.

[maddogx 13]

Looks like there's an SMAA injector now...

http://mrhaandi.blogspot.com/p/injectsmaa.html

Gonna give this a try when I get a chance. Supposedly has very little performance hit like FXAA but without the blurring.

Unfortunately it suffers from the same lack of subpixel information as FXAA, which you can see on some thin pieces of geometry. The result looks good on screenshots, but in game you will see moving artifacts on those edges, which to me is almost even more distracting than the aliasing.

Just like the FXAA in Arma2, I would only use it in combination with standard FSAA/MSAA.

[BasileyOne 10]

actually more enjoy sharpening filter, FXAA injector introduced, than FXAA blurring itself, especially forks where mediocre Unsharp Mask used instead of uniform sharpening.

[Dwarden 1125]

maybe. ATT had bogus kernel-level driver, which causes hangs, alters haddrives MBR and tamper OS security policies processing/function as well as with backup stuff.

the crash happen because ATT has injector for overlay and other features so double injector = fail

now tell me how does ATT's driver alter MBR and tamper OS security policies ?

or just random rant w/o clue? :D

[BasileyOne 10]

the crash happen because ATT has injector for overlay and other features so double injector = fail

now tell me how does ATT's driver alter MBR and tamper OS security policies ?

or just random rant w/o clue? :D

so, ATT need different method of displaying it ?[like Unwinder-introduced in MSI Afterburner]

what you want to hear ? see a dump ? or get dump of altered MBR or see system integrity suite report about it ?

"or just random rant w/o clue ?" © ?

basically you can get WINDDK and kernel debugger and check for yourself if you not believe me.

[Dwarden 1125]

in 99% cases all OSD hooks are 'hacks' using less or more abrupt methods and thus in the end not compatible one with another

i have ATT and i don't see it altering MBR or else ...

cause if it was true it would be in security reports all around world already :)

what security suite and what exactly it reports?

[BasileyOne 10]

in 99% cases all OSD hooks are 'hacks' using less or more abrupt methods and thus in the end not compatible one with another

i have ATT and i don't see it altering MBR or else ...

cause if it was true it would be in security reports all around world already :)

what security suite and what exactly it reports?

lucky you, then[or less careful]

good for you.

thats wasn't likely ;-) and common mistake - many exploits was well-used for years ;-)

i prefer not reveal system intergrity software im use. but basically you can notice it, comparing you mbr with default one, analysing you harddrive outside OS on it[ie loading from other. or CD or USB-flash1]

after that time i was discover RivaTuner and not regret about until Unwinder stop developing/improving/updating it :[

some was suggest RadeonPro or PowerStrip, but thats wasn't same league.

[Dwarden 1125]

like i said ATT does not change MBR ...

it's driver load normally like powerstrip and rivatuner driver

and while it's not as advanced as Unwinder code i find You claims bogus

if you think You right, why You not bother post on the authors forum

(which is btw. the exact same forum as rivatuner/msi afterburner author uses)

anyway we crossed offtopic and i simply don't like things being thrown into air w/o single proof

[sickboy 13]

LOL. Yea ATT author hasn't got any better things to do than corrupting MBR's (on purpose) and ... well, what possible benefit could mr ATT author have of corrupting/adjusting MBR's :O :D

+1 to Dwarden's post; Go report issues on their support platform / issue tracker instead of spreading base-less accusations :D

Edited December 5, 2011 by Sickboy

[Dwarden 1125]

Unfortunately it suffers from the same lack of subpixel information as FXAA, which you can see on some thin pieces of geometry. The result looks good on screenshots, but in game you will see moving artifacts on those edges, which to me is almost even more distracting than the aliasing.

Just like the FXAA in Arma2, I would only use it in combination with standard FSAA/MSAA.

enhanced SMAA in detail:

SMAA 1x is enhanced MLAA implementation.

SMAA T2x is temporal SSAA 2x + MLAA.

SMAA S2x is MSAA 2x + MLAA.

SMAA 4x is MLAA + TSSAA 2x + MSAA 2x.

it's nothing like just FXAA or just MLAA ... especially on higher quality levels

Edited December 6, 2011 by Dwarden

[MavericK96 0]

Yeah, so far the SMAA injector implementation is only 1x...hopefully T2x will be implemented soon.

[Dwarden 1125]

hence why i would prefer to achieve native eSMAA support in engine :)

[metalcraze 290]

I've tried SMAA injector and hey it looks better than FXAA.

Yeah with the current version of it (1.1) stuff like wires still looks barely different from no-AA but all other "ladders" are smoothed without turning the picture into mud like FXAA does.

Ditch FXAA, do a better SMAA implementation instead.

[Dwarden 1125]

edit smaa.fx and add these 2 lines

#define SMAA_PRESET_ULTRA 1

drop jaws :)

Edited December 6, 2011 by Dwarden

[metalcraze 290]

Almost like 8x FSAA (save for thin objects but that's understandable)

Except at 4-5 FPS price. No AToC obviously though. And no muddy textures like with FXAA.

Again ditch FXAA, give us this. Since loading screen after alt-tabbing and some corruption of 2D graphics (like in the editor) are somewhat annoying.

[Dwarden 1125]

* SMAA_PRESET_LOW (%60 of the quality)

* SMAA_PRESET_MEDIUM (%80 of the quality)

* SMAA_PRESET_HIGH (%95 of the quality)

* SMAA_PRESET_ULTRA (%99 of the quality)

also i suggest try use

#define SMAA_PIXEL_SIZE float2(1.0 / 1920.0, 1.0 / 1200.0)

where 1920 and 1200 are your w / h on resultion

also if you use

#define SMAA_PRESET_HIGH 1

according to the paper there should be no ghosting for quality 1 and 3

which equals LOW and HIGH

2 MEDIUM and 4 ULTRA are using double resolve

Edited December 6, 2011 by Dwarden