TressFX: A new frontier of Realism in PC gaming

Yep I think that's the difference. On AMD hardware the hit is similar but still results in playable frame rates because baseline performance is higher.
 
The hit is smaller on tbreak's test. Could it have something to do with the fact that they used the Catalyst 13.2 beta 7 driver instead of the beta 6? Or just a different scene?

They did mention aliasing problems even with SSAA 4X, which reminded me of Humus's Wire-AA. It sounds like something that would solve discontinuity problems quite elegantly and efficiently. That is if there are discontinuity problems, not just jagged edges.
 
Last edited by a moderator:
b74b13f3c8b2d429982652b1e9e13d02.png

http://www.pcgameshardware.de/AMD-R...mb-Raider-PC-Grafikkarten-Benchmarks-1058878/
 
My 6850 (17.3 fps) and 2500K outperforms the 7770 and 650 Ti with those exact settings on tbreak.

Tried a few more settings -

Normal, FXAA - 72fps
Normal, FXAA, TressFX - 37fps
Normal, No AA, TressFX - 42fps

Almost a 100% hit using TressFX on a 6850 at these settings. The effect is great, anybody who says otherwise is full of shit (don't judge it by the stills from the ingame benchmark). That said, it does appear to be a Tahiti card level effect for now.
 
Last edited by a moderator:
Here's quick capture of the same scenario as where the in-game benchmark takes place, in-game the hair does blow indeed to correct direction according to wind, while in bench it's being blown to opposite direction
 
My 6850 (17.3 fps) and 2500K outperforms the 7770 and 650 Ti with those exact settings on tbreak.

Tried a few more settings -

Normal, FXAA - 72fps
Normal, FXAA, TressFX - 37fps
Normal, No AA, TressFX - 42fps

FXAA incurs a 12% hit? That seems like a lot...
 
FXAA incurs a 12% hit? That seems like a lot...

I just ran it again and got 36.2 with FXAA and 37.1 without.

It's buggy and I think the game needs to be shut down and restarted before each benchmark - I've noticed that sometimes it runs through her body instead of around it and that is probably affecting scores.
 
Game runs well and looks great even on normal btw (not running TressFX as the 6850 just isn't up to it). Couple of hours in I'm really starting to enjoy it as well. My initial impressions though are that a 7870 is what you'd want minimum for High + TressFX, and almost certainly a better idea to be on an OC'd 7950. It's one of those games where 40fps feels like it's light-years behind 60fps.
 
At one point I could see a weird silhouette of her eyes, through her hair, from behind her head. And when she's upside down in the beginning, only the pony tail is correctly positioned. It behaves strangely in wind alright. Quirky. It looks neat. Definitely has the hair appearance from medium distance. But off it goes because it's stupid demanding.

Later I was near a creek and the water was all messed up, rendering above or in front of some things.

Buggy game.
 
Whats the gpu you are using ? ( i know tressfx need some fix but i have not see much complain ( maybe they was not look where they should ) .. i will not get the game before 1 week at least.. no time for play .
 
I just ran it again and got 36.2 with FXAA and 37.1 without.

It's buggy and I think the game needs to be shut down and restarted before each benchmark - I've noticed that sometimes it runs through her body instead of around it and that is probably affecting scores.

Yeah it the same with SSAA .. ( i think it is linked if you switch the setting in game and not in launcher and vice versa for the benchmark )
 
I'm playing with TressFX on, and after playing with it for over an hour, I just can't disable it, the difference and superiority is just to obvious... the hair moving really helps the game to feel more natural, even if it's far from perfect...

I'm playing with FXAA, AF 4x and tessellation OFF, the rest is at the maximum available (like on Ultra), with a relatively low resolution (1280x1024), since I'm using one of the old 5:4 screens, the VGA is some old 5850, tessellation on is just to much... if I was trying to play at 1080p I would probably need to play with TressFX off (or lower a few other things).

TressFX on = 36
off = 57

it's a huge hit, but even with it turned on it's almost always easily over 30FPS
 
Here's quick capture of the same scenario as where the in-game benchmark takes place, in-game the hair does blow indeed to correct direction according to wind, while in bench it's being blown to opposite direction

lol, i actually forgot the actual video
 
Did some testing

I have to admit that this whole TressFX thing got me interested so i decided to buy the game (especially i recently got new shiny GB 7870 GPU).
I started the game and set everything to max with AA set to FXAA. First bench run gave me 44 avg with 2fps min.! Fortunately it turned out its only kind of loading bug of the built in benchmark and second run gave me 44.8fps avg with 21.8fps min. Still i had the impression i cant spot that 22fps min. value during the bech so i decided to fire up fraps. To mu surprise that min. fps value is always recorded during the first fractions of the second after starting the test (and the image is just starting to show). During the whole test minimum recorded value was 35fps. So overall i am very pleased with performance i get with my hardware (oh and i use 1680x1050 res. because of my old LCD).
Effect itself is not perfect but i have to admit it adds to the immersion.
 
Sounds good, I have hopes for my 7870 + 3570k and I'm at that resolution as well. With single player games of this style, I'm fine with 30fps+. Thanks for the feedback!
 
Sounds good, I have hopes for my 7870 + 3570k and I'm at that resolution as well. With single player games of this style, I'm fine with 30fps+. Thanks for the feedback!

Just played a bit and i have to say that benchmark is representative of the actual gameplay expirience. Game always stays above 30 fps (i think 32 was the lowest value i noticed with hair close up) and usually its around 45-50fps. I also run the bench with hair set to normal and got almost 60fps average.
I'm just wondering if new drivers improve performance (I'm running cat 12.11 beta 4).
 
Back
Top