Parhelia not happy with fsaaview?

[Tatchan]

I started to work with Parhelia today, and wanna have some easy way to compare the AA quality, but I found that fsaaview is not showing the effect of Parhelia's AA, can someone help me there?
BTW, any good tests should I perform? I've tried to run 3dmark, quake iii, villagemark, templemark with different settings, any method to show the Aniso Quality easily? Many thanks.

 

[Doomtrooper]

I think the most popular Anisotropic filtering shots are Serious Sam: Second Encounter Citadel demo ...

http://www.beyond3d.com/articles/8500gf4iq/index.php?page=aniso.inc

FSAA tests...Dungeon Siege would be a good one, Return to castle wolfenstein forest level or MOHAA or even old Quake 3 ...

 

[Tatchan]

I think the most popular Anisotropic filtering shots are Serious Sam: Second Encounter Citadel demo ...

http://www.beyond3d.com/articles/8500gf4iq/index.php?page=aniso.inc

Yes! That's good, but how can I show those red/green/blue color range?

FSAA tests...Dungeon Siege would be a good one, Return to castle wolfenstein forest level or MOHAA or even old Quake 3 ...

I'm using Q3 now, it's really good actually, just wonder why fsaaview is not working....

 

[Basic]

If you're refering to my program, then the black/white edges on the sides should work. But the dots in the middle can only show effects of supersampling, since it's just one big quad with a very special texture. And as we know, Parhelia FAA doesn't touch the internals of polys.

OGL FSAAview 0.3 has a geometry mode which could work. But as with SuperScene there can be difficulties. Parhelia has some logic that determine if a pixel should be AAed or not, and it could get confused by the massive amount of edges my prog generates in geometry mode.

Mail me if you want that version.

There is no D3D FSAAview 0.3.
D3D and I have a mutual dislike for each other.

I assume you're doing a (p)rewiev, for what site?


Btw, I'm just working on "OGL AnisoView 0.1". A program just in my übersyntetic test taste. I don't have to much time right now though, so it might not be worth waiting for it. And either way, it would probably not reveal anything earthshattering about Parhelia. It's more interesting for another popular dispute right now.

[Edit]
Download page.

 

[Althornin]

It's more interesting for another popular dispute right now.

You are gonna finally, absolutely, once and for all solve the Radeon Aniso question?
yay!

 

[Basic]

Yeah, I think I could show the sampling pattern quite clearly. Which would change the discusion from:
A > It's done like this.
B > Are you stupid?

to:
A > It still look better/worse.
B > Are you stupid?

One problem though, is that I only have a GF2, and it doesn't do much difference with aniso on or off (still a very slight difference). So it might be somewhat difficult for me to debug.

 

[Althornin]

got plenty of testers right here.

 

[Tatchan]

If you're refering to my program, then the black/white edges on the sides should work. But the dots in the middle can only show effects of supersampling, since it's just one big quad with a very special texture. And as we know, Parhelia FAA doesn't touch the internals of polys.
I assume you're doing a (p)rewiev, for what site?

Yes, I'm trying to use your program, even I've d/l the latest version, you can see the result below. The result seems to be the same under D3D and OGL.

I'm doing a (p)review, really, as Matrox only give me the card for 2 days only...I'm just trying to do my best for my local magazine....well, in Trad. Chinese 8)

Normal:

FAA-16X:

4X:

 

[Kristof]

Odd doesn't seem to work at all... I would have at least expected 4x traditional AA to work. Maybe it does not work in window mode due to some driver issue ? Try fullscreen if possible.

K-

 

[Tatchan]

Odd doesn't seem to work at all... I would have at least expected 4x traditional AA to work. Maybe it does not work in window mode due to some driver issue ? Try fullscreen if possible.

K-

Right, that's what I concern most...Can I put this program in fullscreen mode?
Anyway, FAA-16X works very well on Quake III

 

[Basic]

Kristof probably has the reason. I believe it was the same for V5. Let's realy hope that it's just a driver issue, since it could be very bad for their chances in the professional market.

Sorry that there's no fullscreen mode in the app. I might add it some time, but I don't have any programming time within your 2 days limit.

Doesn't Q3 have a windowed mode? Have you tried it, and in that case did F(S)AA work?

Althorin:
OK, I might contact you later when it's time for testing. Right now it's on ice, I won't have time until earliest next week. Do you have access to both R8500 and GF3/4?

 

[Althornin]

nope.
Sorry, i only have access to GF3 and R100 (radeon)
But i'd be gald to help with those.

 

[Tatchan]

Kristof probably has the reason. I believe it was the same for V5. Let's realy hope that it's just a driver issue, since it could be very bad for their chances in the professional market.

Sorry that there's no fullscreen mode in the app. I might add it some time, but I don't have any programming time within your 2 days limit.

Doesn't Q3 have a windowed mode? Have you tried it, and in that case did F(S)AA work??

Thanks Basic, I'll try to run Q3 in windowed mode tmr, wish that helps me to sort out the problem..

 

[Tatchan]

Confirmed, it's not running FAA-16X nor FSAA 4X in windowed mode(at least for fsaaview and quake iii)