PDA

View Full Version : Colored Herbology


tialaramex
13th October 2004, 00:27
There have been several discussions in the main site & in the adjacent Windows sub-topic about the fact that many ATI owners don't get even an approximation of the colors for herbs, a critical factor in figuring out what you've just found.

Most likely this is an unavoidable technical problem, but it seems that Pluribus can at least be more specific about what it is, since he has actually seen the code.

So, Pluribus, don't afraid to use technical terms, I've written a little bit of OGL and can tell my ARB extensions from my elbow, what is going on with the drab green plants I see where others see a riot of color?

FWIW I have a Radeon 9200SE using the out-of-box drivers on Fedora Core 2, but please don't just tell me to switch drivers or use an nVidia card, I am genuinely interested in knowing what's actually wrong :)

Pluribus
15th October 2004, 04:17
Is simple... In order to support the per pixel lighting, bump mapping and specular mapping, without the use of ATI specific extensions, the ARB_fragment_program extension is required. Since this isnt supported prior to the Radeon 9500, cards of lower series doesnt have those effects. Now... That seems to preclude most of the useful things on lower end cards. As for the very dark herbs and trees on low end radeon cards, it seems to be something with the way they are treating lighting. Interestingly, turning off all effects on other cards doesnt make them go nearly as dark. Adjusting the color balances and gamma has been proven to get the plants to show correctly on some low end radeons. How to do that under linux is a mystery.

tialaramex
15th October 2004, 17:52
X controls the gamma settings for the display, try xgamma. There is a full LUT but the command line tool only provides regular gamma curves, nothing fancy. I tinkered but it didn't seem to fix the herb colors relative to everything else.

The lack of standards compliant shaders is annoying, but doesn't explain the colors being so wrong. Do the colors really look fine for anyone without working shaders? Can they post some screenshots or something?

Ah, speaking of screenshots. The Alt-C capture mechanism doesn't work any more on my Linux system. I can't be sure how long it's been broken, but it didn't work just now and my auto-screenshot reaper says it hasn't seen any for about 7-10 days. The machine pauses for a moment, as usual, but no PNG files appear anywhere.

Pluribus
15th October 2004, 18:54
Look in ~/.egenesis I have begun moving all the "non-client" stuff to user specific directories... Logs will be next to go. I want ONLY client and resources files in the eClient directory. I am working on eliminating the last few places that non-ega files are dumped into the egyptc directory...

My end goal is to allow a central client directory with per user files... I also plan on making the client not overwrite existing screenshots.

tialaramex
16th October 2004, 03:11
Ah, it's a feature. Found them and updated my script.

Thanks for all the hard work you're putting into this BTW.

tialaramex
21st December 2004, 20:28
No-one ever posted any shots without shaders, so I finally gave up and tried Software OpenGL. There's no opportunity here for ATI to have got anything wrong, since the ATI card in this machine is only acting as a 2D framebuffer for MESA's rendering.

Shrubs and herbs look the same dark green in software as in ATI hardware accelerated mode.

I think we have a bona fide bug. I'm not asking for ATITD to magically develop ATI equivalents for the nVidia shaders, but I think we can expect red things to be red, not dark green. In any case these need to be acknowledged as an ATITD bug, and not blamed on drivers as in the past.

Pluribus
5th January 2005, 04:31
All trees and plants are almost black without the shader programs... For those with nvidia cards or 9500+ cards run with the following options to prove it...

./elaunch -nonvrc -nonvrc2 -nofrg10

That will disable all the shader programs. (the -nonvrc and -nonvrc2 are only REALLY needed to show it on nvidia cards but doesnt hurt anything if there on ATI cards..)

What ever made you assume that just because you could duplicate it with software rendering which STILL doesnt dupe the nvidia register combiners or software interpret the ARB fragment programs that it must be a general client bug and not a lack of shaders?

tialaramex
25th January 2005, 03:40
All trees and plants are almost black without the shader programs... For those with nvidia cards or 9500+ cards run with the following options to prove it...

./elaunch -nonvrc -nonvrc2 -nofrg10

That will disable all the shader programs. (the -nonvrc and -nonvrc2 are only REALLY needed to show it on nvidia cards but doesnt hurt anything if there on ATI cards..)

What ever made you assume that just because you could duplicate it with software rendering which STILL doesnt dupe the nvidia register combiners or software interpret the ARB fragment programs that it must be a general client bug and not a lack of shaders?

I think it must have been this quote, which you've just flatly contradicted.


Adjusting the color balances and gamma has been proven to get the plants to show correctly on some low end radeons

All you needed to write, back in October, was that herbs and other game content would never work on these cards, instead of mumbling about color balance and gamma correction which aren't in the least bit relevant to the real problem.

TO THIS DAY the ATITD.com site doesn't warn people that some of the gameplay content simply won't be accessible unless they have much higher specifications than those listed. Is that dishonest? It's certainly unhelpful.

The generated shaders are by no means either the only way or the most sensible way to achieve the basic color shading needed to identify herbs, that there exists not even the most basic of fallbacks remains in my opinion a serious bug.

Pluribus
31st January 2005, 17:31
Sigh, There is NO contradiction there... without the shaders, the plants appear nearly black and require the gamma and color balance adjustments. Tell you what... You get Josh to write the ATI specific code paths and I will put them both the linux and OSX versions of the client.

ESRB Notice: Game Experience may change during online play... Game will look better when not using PoS hardware...

ArmEagle
31st January 2005, 21:27
ESRB Notice: Game Experience may change during online play... Game will look better when not using PoS hardware...

AMEN!

Pluribus our :indian_ch

chisisi
3rd February 2005, 19:31
best ESRB ever

tialaramex
4th February 2005, 02:58
Sigh, There is NO contradiction there... without the shaders, the plants appear nearly black and require the gamma and color balance adjustments. Tell you what... You get Josh to write the ATI specific code paths and I will put them both the linux and OSX versions of the client.

You can't seem to make your mind up Pluribus. Let's try this slowly.

Are the plants supposed to be colored (not just all of them a constant dark green for some silly reason, but actually colored) if shaders are not available?

If they are not supposed to be colored, why not? Basic vertex coloring is a cheap and universally available feature of OpenGL. With many of Tale's plants being bright reds, yellows or blues it would be easy to distinguish them.

If they are supposed to be colored, why are the colors wrong? The rest of the game uses colors, the rest of the colors are fine. OpenGL certified software renders the exact same dark green mess, so it's ATITD at fault here.

Marrick
4th February 2005, 14:10
Ok, just so we can get the atitd blame out of the way...

Home: ATI Radeon 9800 AIW Pro

Work: ATI Radeon 9000

I use the _exact_ same catalyst drivers at home and work, and the colours are correct at home and that dark green crap at work. Tada, hardware issue.

QED

Edit: The issue is _only_ with <9500 radeons, so the ESRB is quite correct. Spend more than $20 on a video card.

Chichis
4th February 2005, 19:58
I don't think it's fair to blame the people who have the cards at least. Most of the people affected are using them as mobilities in laptops. The choices for laptop graphics are normally much more restricted, and the Radeon Mobility series 7000-9000 is STILL considered a good mobile graphics card, and is still sold in new laptops. Replacing a laptop graphics card is quite a bit more difficult than a desktop, and simply not an option for most people.

Not to mention the Mac Mini ships with a Radeon 9200 and you'll probably never get a new one in there...

Marrick
4th February 2005, 20:31
Good point, but the ire should be directed at ATI for not having the catalyst driver correctly handle the inabilities of their lower end hardware not at egenesis (or Plur) who have no control of ATI's ass driver support.

tialaramex
4th February 2005, 23:33
Marrick did you run with shaders disabled as described by Pluribus?

Otherwise you're not comparing like with like, and unfortunately that just tends to make you look like an idiot.

ATITD's code could it seems be paraphrased as

if (nvidia) {
complicated_fancy_rendering
} else if (high end card) {
moderately good rendering
} else {
needlessly broken rendering
}

The only issue from where I'm standing is, how come we get broken rendering, instead of merely lower quality. No-one's expecting a Radeon 9000 to do fancy lighting and bump-mapping. But the herbs aren't even close to the right color, because OpenGL is apparently being told to draw them all as dark green.

This is the issue on which Pluribus doesn't seem to have a coherent answer. One minute he's talking about shaders, the next he's talking about gamma adjustments. So I asked a simple step-by-step series of questions to try to find out whether he actually knows something useful, or is just plain confused.

Solalique
5th February 2005, 01:41
Good point, but the ire should be directed at ATI for not having the catalyst driver correctly handle the inabilities of their lower end hardware not at egenesis (or Plur) who have no control of ATI's ass driver support.
I don't know what you are suggesting -- that the driver should have software emulation for more advanced features that were never available in the card's hardware?

Though it is discussed less freqently, S3, SiS, Intel chipsets <900, and others have the exact same deficiency (no ARB_fragement_program support). I don't know how you would figure a "broken" ATI driver can manage that.

As far as I can tell the driver is doing exactly what it is told to do. Let's consider beetles by way of comparison.

Beetles have some difference in appearance between the "bad" and "good" cards. And yet the fundamental colors are not too extremely different between the cards. You will have no trouble telling apart blue beetles from red ones, for instance. The fragment program is used to add specular lighting, bump mapping, etc., and beetles appear "shiny" and more brightly colored on the "good" cards. The fragment program is being used to enhance the appearance.

By way of contrast, as far as I can tell, the coloration of plants (outside of the texture mapping), has been left entirely to the fragment program. Which means that if you don't have fragment program support, you lose all of that. This is a design decision by ATITD development, and has nothing to do with so-called broken drivers.

For a more admittedly simplistic analogy, suppose that a game can present words on the screen and/or play sounds of a voice speaking the words. And imagine that some users have speakers on their machines, while others do not.

One design choice is to use both written and spoken words. Those who have no speakers can rely on reading the written words, while those with speakers get an enhanced experience. This seems roughly similar to the beetles color situation.

Another possible design choice is to say: "well, we don't want the printed words; they are ugly, we'd have to maintain both written and spoken stuff, and we think everyone has speakers anyway". In this case, those users who don't have speakers are in a lot more trouble; they don't have any way of knowing what is being said. This seems to be roughly similar to where we are with plant colors. (While the gamma correction suggestion perhaps corresponds to a suggestion to "turn it up really loud").

Now your suggestion would seem to be something like: I got the latest driver on my computer (that has no speakers) and the driver is not generating printed words to fall back on. The driver is broken.

Pluribus
5th February 2005, 02:55
Look. Either way, it isnt a linux or OSX specific issue. If you have bitches about the herbs and oasis trees render, take them up with Josh. It is beyond my capabilities to write what you want.

Point blank. Without the fragment program or NV register combiner hardware support. Oasis trees and herbs look VERY dark. With gamma and color corrections it is possible to make them look acceptable. Since you dont want to hear that, Since you seem to think it somehow contradicts itself, I will say TOO BAD... Take it up with eGenesis.

tialaramex
5th February 2005, 03:41
OK, going to Josh about this makes sense. I'll mail josh@egenesis.com in a day or two with some observations and questions [unless someone suggests a better way to contact him by then].

You never know we may be a bug fix away from having blue plants look blue instead of green, even on cheap Rv100 or i8xx based laptop chipsets.