VOGONS


Reply 60 of 123, by havli

User metadata
Rank Oldbie
Rank
Oldbie

Am I the only one who can't see any significant difference in those two screenshots? Yes the GF3 and R9800 are slightly different... but which is better, thats just guessing to me. I don't think is is possible to tell the difference in regular gaming.

There is a huge difference in AF quality between various GPU generations which is easy to see. For example AF on R100/200 and NV1x is so ugly it is pretty much useless. NV2x/3x/R500 on the other hand looks great, while NV4x/G7x/R300 is not so good. And GF8 up to Pascal is superior to anything in AF quality.
But again, this thread compares only bilinear/trilinear at best (whis is pointless to use on R9800 or GF3) and on top of that on super closeup screenshot of blury texture of resoution 256x256 at best. So in short this hardly can make a difference in actual gaming.

HW museum.cz - my collection of PC hardware

Reply 61 of 123, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

this thread compares only bilinear/trilinear at best

This thread does not compare mipmap filtering.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 62 of 123, by Reputator

User metadata
Rank Member
Rank
Member
havli wrote:

Am I the only one who can't see any significant difference in those two screenshots? Yes the GF3 and R9800 are slightly different... but which is better, thats just guessing to me. I don't think is is possible to tell the difference in regular gaming.

Yeah it's a fairly moot point. We're talking different patterns of banding, though all cards exhibit banding with the example textures being shown. If you have to apply a filter to the screenshot to really tell the difference, it probably doesn't matter in-game.

https://www.youtube.com/c/PixelPipes
Graphics Card Database

Reply 63 of 123, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

It's crystal clear when you look at the sky in Q3.

Also on various lightmaps in many games:

UT Software.png
Filename
UT Software.png
File size
413.51 KiB
Views
1712 views
File license
Fair use/fair dealing exception

Software render for reference

UT GeForce 3.png
Filename
UT GeForce 3.png
File size
300 KiB
Views
1712 views
File license
Fair use/fair dealing exception

GeForce 3

UT GeForce 8800.png
Filename
UT GeForce 8800.png
File size
248.62 KiB
Views
1712 views
File license
Fair use/fair dealing exception

GeForce 8800

UT Voodoo 2.png
Filename
UT Voodoo 2.png
File size
159.52 KiB
Views
1712 views
File license
Fair use/fair dealing exception

Voodoo 2 (captured via Hypersnap)

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 64 of 123, by havli

User metadata
Rank Oldbie
Rank
Oldbie

Crystal clear? I dont think so, these UT screenshots are all so ugly it hurts my eyes. It doesn't really matter which one is the best/worst. Also I don't remember UT looking this bad on any GPU (except maybe 16-bit dithering on TNT/TNT2... but thats a different issue).

HW museum.cz - my collection of PC hardware

Reply 65 of 123, by appiah4

User metadata
Rank l33t++
Rank
l33t++

If you cant see the difference above you may as well be blind.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 66 of 123, by havli

User metadata
Rank Oldbie
Rank
Oldbie

I see it, of course. But it must be some kind of weird settings used because UT usually doesn't look this terrible. The point is - I consider image quality of all four screenshots to be well bellow my level of acceptance.... and in that case I no longer care about the difference because even the best quality is useless for me.

The good way of comparing image quality for me is: decent resolution, maximum details, AA and AF that GPU in question still can run at acceptable fps... and of course screenshots taken from normal gameplay and pixep-pixel accurate position if possible. Pretty much all the Q3A screenshots comparison here looks way too "academic" with mostly zero impact on gameplay.
For example this is my way of doing image quality comparison:
https://abload.de/img/hl2_2017_01_23_21_49_hgzxq.png
https://abload.de/img/hl2_2017_01_23_21_48_chb9x.png

HW museum.cz - my collection of PC hardware

Reply 67 of 123, by kithylin

User metadata
Rank l33t
Rank
l33t
havli wrote:
I see it, of course. But it must be some kind of weird settings used because UT usually doesn't look this terrible. The point is […]
Show full quote

I see it, of course. But it must be some kind of weird settings used because UT usually doesn't look this terrible. The point is - I consider image quality of all four screenshots to be well bellow my level of acceptance.... and in that case I no longer care about the difference because even the best quality is useless for me.

The good way of comparing image quality for me is: decent resolution, maximum details, AA and AF that GPU in question still can run at acceptable fps... and of course screenshots taken from normal gameplay and pixep-pixel accurate position if possible. Pretty much all the Q3A screenshots comparison here looks way too "academic" with mostly zero impact on gameplay.
For example this is my way of doing image quality comparison:
https://abload.de/img/hl2_2017_01_23_21_49_hgzxq.png
https://abload.de/img/hl2_2017_01_23_21_48_chb9x.png

I second this motion.. I wouldn't consider any screenshots to be useful unless they're taken at maximum anti-aliasing & AF possible while still maintaining good frame rates. And from general gameplay areas folks would commonly see while running around.. preferably a large "open area" section.. not some random hallway of darkness.

Regardless of image quality.. I'm one of those folks of that I usually won't even bother playing a game at all unless I can run it at maximum settings (all settings, including AA & AF).

Reply 68 of 123, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

But it must be some kind of weird settings used because UT usually

UT usually looks like this +/- your gamma settings. Square lightmaps of GeForce 3 screenshot is inherent problem of all old Nvidia cards, you can't fix it by any means.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 69 of 123, by firage

User metadata
Rank Oldbie
Rank
Oldbie

So, it turns out the GeForces optimized for texturing performance over quality more than the other two outfits. An interesting find. Could be the decider in some scenario, but not likely for my own setups. 3dfx's performance is stuck in the first half of the year 2000, and ATI's stuff doesn't have quite the level of backward compatibility that GeForce2 and GF4MX, or even the next couple of generations, offer.

My big-red-switch 486

Reply 70 of 123, by appiah4

User metadata
Rank l33t++
Rank
l33t++
firage wrote:

So, it turns out the GeForces optimized for texturing performance over quality more than the other two outfits. An interesting find. Could be the decider in some scenario, but not likely for my own setups. 3dfx's performance is stuck in the first half of the year 2000, and ATI's stuff doesn't have quite the level of backward compatibility that GeForce2 and GF4MX, or even the next couple of generations, offer.

ATi BC is fine for me the 8500 is my goto card for anything dX8- and its iq is much better than gf4-

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 71 of 123, by Putas

User metadata
Rank Oldbie
Rank
Oldbie
havli wrote:

For example AF on R100/200 and NV1x is so ugly it is pretty much useless.

And I claim for the generation of R200 anything without AF is useless loss of quality.

Reply 72 of 123, by appiah4

User metadata
Rank l33t++
Rank
l33t++
Putas wrote:
havli wrote:

For example AF on R100/200 and NV1x is so ugly it is pretty much useless.

And I claim for the generation of R200 anything without AF is useless loss of quality.

R200 AF was great. No idea what he means probably he never owned an R200 card. A lot of fuss was made at the time about AF and trilinear not being possible together buy ultimately it made little difference.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 73 of 123, by havli

User metadata
Rank Oldbie
Rank
Oldbie

Are we even living on the same planet? 🤣 AF on the R2xx is extremely angle dependent and on top of that it produces rather extreme shimmering. It may look "ok" on static screenshot but not in movement... which is the case of the actual gameplay. Yes, it is fast, while NV20 AF is very slow, but the quality difference is huge. As for the never owned R200... yeah, try that on someone else. I have pretty much everything starting with Voodoo Graphics all the way up to current gen GPUs. Including R200 (more than one card), it is not that rare. http://hw-museum.cz/vga/vga-list?category=3D

I don't have anything better at hand currently, but AF tester should prove my point.

R100 http://imgur.com/dGawtOE
R200 http://imgur.com/dk4k5Y8
R300 http://imgur.com/fJU2FB5
R500 HQ http://imgur.com/UIRkKMt
R600 http://imgur.com/v38VR8D
HD 5000 http://imgur.com/4oImvyl
HD 6000 http://imgur.com/RuUYiBj
CGN http://imgur.com/c24yQ2D

NV10 http://imgur.com/9tqSCQD
NV20 http://imgur.com/t8dIPIv
NV30 HQ http://imgur.com/vInAcD6
NV40 HQ http://imgur.com/epYliBe
G80+ HQ http://imgur.com/btn2YCa

HW museum.cz - my collection of PC hardware

Reply 74 of 123, by silikone

User metadata
Rank Member
Rank
Member

Why is it that the pre-GPUs have better filtering, and why was it this bad for so many years?

Do not refrain from refusing to stop hindering yourself from the opposite of watching nothing other than that which is by no means porn.

Reply 75 of 123, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

"Optimisations" probably. Quite effective since nobody really cared or noticed all this years. Well, at least for Radeon cards. Kinda funny in retrospective, if you remember that much older cards like original Rage 128 were blamed for horrible linear filtering. I won't be surprised if Radeon filtering optimisations were actually borrowed from Rage 128 Pro.

Meanwhile there are still a lot of cards to test if anyone interested.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 76 of 123, by Putas

User metadata
Rank Oldbie
Rank
Oldbie
havli wrote:

AF on the R2xx is extremely angle dependent and on top of that it produces rather extreme shimmering. It may look "ok" on static screenshot but not in movement... which is the case of the actual gameplay.

If the shimmering was so bad that people disabled AF, than ... their perception of image is different than mine. Possible, though I doubt their numbers were significant.

Reply 77 of 123, by silikone

User metadata
Rank Member
Rank
Member
The Serpent Rider wrote:

Meanwhile there are still a lot of cards to test if anyone interested.

The compression sucks, but some interesting comparisons were done professionally back in the day.

1090787143TO62AMOEWX_4_3_l.jpg

https://www.hardocp.com/article/2004/07/29/of … ardware_guide/4

Do not refrain from refusing to stop hindering yourself from the opposite of watching nothing other than that which is by no means porn.

Reply 78 of 123, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Calling Kyle Bennett a professional would be stretching things quite a bit.

Retronautics: A digital gallery of my retro computers, hardware and projects.