Funny & Weird Screen Shots

[Haverford]

maybe they aim at being the worst-pinged clan?

Heh, a lot of them are Turkish, actually O_o I think.

Nope k@oS is the "worst-pinged" clan! 5 of us average 130, and another guy is always 300+. And another member never has a ping above 15.

XD

[RMF]

Sauerkraut is german, and 'haufen' also sounds germanisch. But it also might be they aren't from the same country.

[GrosBedo]

/r_detailtextures 0

Indeed, r_picmip is mostly used now for visibility enhancements (although I can't see how removing textures can really help you, furthermore many servers have brightskins now...).

r_detailTextures 0 is a good one, but actually, I find that the cvars that impacts the most the FPS are :

r_swapInterval 0 // Define the vertical sync. Can sometimes do nothing, or be awful. Try to set this one to 2 to see my point.
r_showSky 0 // Remove skybox, gain quite a bit of fps but no more cute sky
r_fastSky 1 // Use in conjunction with r_showSky 0
r_subdivisions 12 // Define how the polygons of the models will be drawn (inferior values draw better, you can see the changes in realtime by looking at a model like sarge)
r_mode 5 // Change resolution of the screen, actually the best way to gain more FPS

Above all, I would say that to gain FPS, one must set r_swapInterval 0, and a lower resolution with r_mode. You can jump from 30 FPS to 100 FPS pretty easily with these.

[GrosBedo]

About the FPS decrease :

http://ioquake.org/forums/viewtopic.php?f=2&t=1593

[7]

Indeed, r_picmip is mostly used now for visibility enhancements (although I can't see how removing textures can really help you, furthermore many servers have brightskins now...).

Really? Do you notice someone more quickly when he's standing in front of a jungle background or when standing in front of a more evenly colored green backdrop without any intricate patterning? High r_picmip values do give better visibility, especially when combined with r_vertexlight. That's why it's considered cheating by some (many?) in the community and the cvars are protected in OA.

[7]

Arggh, pressed quote instead of modify again, sorry!

[Cacatoes]

Indeed, r_picmip is mostly used now for visibility enhancements (although I can't see how removing textures can really help you, furthermore many servers have brightskins now...).

Really? Do you notice someone more quickly when he's standing in front of a jungle background or when standing in front of a more evenly colored green backdrop without any intricate patterning? High r_picmip values do give better visibility, especially when combined with r_vertexlight. That's why it's considered cheating by some (many?) in the community and the cvars are protected in OA.

Lol, depends. If the player is all green then I'd see him more easily in a Jungle Some people say they avoid pul1ctf for a similar reason.

[7]

Lol, depends. If the player is all green then I'd see him more easily in a Jungle Some people say they avoid pul1ctf for a similar reason.

So you don't believe in camouflage then No offence, but in my not so humble opinion pul1ctf looks like it was designed for Teletubbies and not for a shooter. I have congenital red-green color blindness (deuteranopia) and when playing instagib I just can't see opponents on pul1ctf in time before getting killed, except if they're in the (very dark) Tony skin that is.

[Cacatoes]

On the contrary, I believe in camouflage, but camouflages can be complex or simple (flat) depending on the context. I guess a cameleon would try to get uniform colors if in a uniform environment.
If you have problems with specific colors that's a problem, I doubt we'll see ways to switch colors in games soon.

[7]

On the contrary, I believe in camouflage, but camouflages can be complex or simple (flat) depending on the context.

No argument with you there, but the point I was trying to make is that any camouflage, wether simple or complexly patterned, works better against a patterned background than against a (nearly) monochrome one.

If you have problems with specific colors that's a problem, I doubt we'll see ways to switch colors in games soon.

I have no problems seeing individual colors (though I can distinguish less tones of pinkish, purplish and greenish colors than normal people can), the problem is not seeing red-green contrasts so it's very hard for me to distinguish a bright reddish object against a bright greenish backdrop and bright greenish objects against a reddish backdrop. This form of color blindness is congenital in males and about 1% of the male population has it, some of them even without knowing it. I suspect that a lot of players who can't play pul1ctf but can't put a finger on why not, would test positive for one  form of color blindness or another (about 10% of all males has a form of color blindness).

Edit: do it yourself colorblind diagnose:

Most protanopics can't see the number 37 in this image.	Most deuteranopics can't see the number 49 in this image.	Most tritanopics can't see the number 56 in this image.

(Images taken from Wikipedia.)

[GrosBedo]

Indeed, r_picmip is mostly used now for visibility enhancements (although I can't see how removing textures can really help you, furthermore many servers have brightskins now...).

Really? Do you notice someone more quickly when he's standing in front of a jungle background or when standing in front of a more evenly colored green backdrop without any intricate patterning ?

False assumption based on a myth.

Actually, your reflexes are triggered by your side eye cells in your peripheral vision (called rod cells), the ones that are sensitive to movements, but not at all to colors.

This means that if you rely on color, you will be much less reactive than if you focus on movements. That's why I find the simpleitems and drawicons cvars very useful, to turns all static items, weapons and ammos into static icons rather than animated 3D models, which lessen your focus.

So, if you want to enhance your reflexes, better use your peripheral vision. After all, human survived to predators thank's to these cells.

More infos about rod cells :
http://en.wikipedia.org/wiki/Rod_cell

[Cacatoes]

I was trying to make is that any camouflage, wether simple or complexly patterned, works better against a patterned background than against a (nearly) monochrome one.

Right, I was a bit tired when I answered (I'm still, heh), guessed it was something like that.
Seems I'm a bit triceratopnic, that 56 is damn well hidden.

So... when will we see shaking skins ?

[fromhell]

angelyss shakes a lot why doesn't she get used

[MyLittlePwny]

Not sure exactly how it applies relative to movement, concealment, etc., but I bet some "Dazzle camouflage" skins might have an interesting effect...

[7]

So, if you want to enhance your reflexes, better use your peripheral vision. After all, human survived to predators thank's to these cells.

I play on a 19" with a viewing distance of about 40 cm so the field of view my monitor occupies is about 60 degrees. I don't think I'm using al lot of peripheral vision when I'm playing, but I could be wrong.

[Falkland]

About the FPS decrease :

http://ioquake.org/forums/viewtopic.php?f=2&t=1593

This is not true at least for me ... I've obtained quite the same amount of FPS while playing OA080 between the ( recompiled ) default OA engine and the ( recompiled ) jackthompson quake3 1.32b based engine.

Back to recent past , I obtained quite always fixed 125 fps with ioquake3 ( patched engine , not a vanilla ) , in both OA081/085 and CPMA+ioq3 while playing on Ubuntu 8.04 Hardy Heron , with a custom RT kernel.

After upgrading to Lucid Lynx 10.04 I am able to play both games at a maximum of 65 FPS on the same system with a Low Latency but NOT RT kernel.

And I've used for first the same binaries compiled on Hardy , then recompiled on Lucid with the same optimizations.

I've noticed two things :
1. the CPU is never running at 100% while playing , always at max 65/75 % ; it was always fixed at 100 % while playing on Hardy. As a consequence the CPU fan is never running at its maximum , always at mid speed ( as now I DO never even run the risk the get it overhated while playing )

2. when running glxgears on hardy it started by 7500/7600 fps every 5 seconds till 10200/10400 fps. Now it starts by ~5000 till 10100/10200.

Probably there's some software bottleneck somewhere ( in the kernel or  in the graphic subsystem or in the Mesa Driver ... )

[7]

Probably there's some software bottleneck somewhere ( in the kernel or  in the graphic subsystem or in the Mesa Driver ... )

If the bottleneck is in the kernel it should be easy to detect, just reboot the kernel with profiling options, run OA and run readprofile.

I run Fedora 13 64bit with a standard kernel (2.6.33.6-147.2.4.fc13.x86_64) and I get constant 125 fps easily, but only if I restrict OA execution to a single core with taskset. I get best performance and lowest latencies when I force OA to run on the core with the highest number, dunno if that's Fedora specific behavior.

[Falkland]

Probably there's some software bottleneck somewhere ( in the kernel or  in the graphic subsystem or in the Mesa Driver ... )

If the bottleneck is in the kernel it should be easy to detect, just reboot the kernel with profiling options, run OA and run readprofile.

I'll do that with the standard kernel because I've disabled profiling options at compile time in the custom kernel ; anyway my system is an old and single-core low end machine and the standard Ubuntu kernel is compiled with some options that make it working bad on that machine ( NO PREEMPTION , 300 HZ , Swapaccount resource counter)

I run Fedora 13 64bit with a standard kernel (2.6.33.6-147.2.4.fc13.x86_64) and I get constant 125 fps easily, but only if I restrict OA execution to a single core with taskset. I get best performance and lowest latencies when I force OA to run on the core with the highest number, dunno if that's Fedora specific behavior.

This is probably a problem related to the amount of SMP interrupts that occur while playing since the engine has not native and working SMP support ( r_smp is reported to not work anymore since a long time ) and maybe also to the load_thick_balance_task BUG that affects all recent versions of the linux kernel ( from 2.6.31 till the latest 2.6.35 : the BUG will be fixed in the 2.6.36 and it is currently fixed only in the -tip tree ).

Someone at ioquake3 forum has published a patch to add transparent SMP support to the engine ; I've also reported it in the binaries development thread .

- The discussion on ioquake3 forum : http://www.ioquake.org/forums/viewtopic.php?f=2&t=125#p861
- The patch : http://odin.mercenariesguild.net/quake3/smp_r1515.patch

[PopeJo]

* PopeJo gives Falkland a cake

I reached the same conclusions by own experiences, blurred imagination of processes I never understood and a lot of reading  
thx for making it precise and giving links

[faoa]

during a bored session on ROFL CTF 4 Fishes:  

[chaoticsoldier]

I took these today. For some reason the /screenshot command stopped working properly, but /screenshotjpeg still works fine.