Well, it was almost too fast on my Duron 800 with GF2...

Anyway, stencil tests are "free" if you do depth testing, and if you don''t use depth testing, well... It "costs" as much as a normal depth test. I.e. stencil buffering is _not_ slow. The only thing slow about it is that you often have to draw more geometry (since you often first draw some geometry setting the stencil buffer, and then some other geometry to mask against the stencil buffer). Anyway, for what you''ve done here, it''s perfect. The only backside of it is that it''s not supported on all hardware. There are ways of doing pretty much the same thing with destination alpha though, which is supported on almost all hardware. Don''t know exactly how though.

Dentoid, thanks for your reply. it was very useful. i''ll have to sort things out with the guy who''s making glfw, which is the window creation library i''m using (to achieve muti-platform support). gotta make sure that his library does a good job, and if i don''t get better consistency, i guess i''ll have an option to toggle it. but users with stencil buffer enabled will have an advantage over those whose is not, since thoses without stencil support will only see other players whose ceter is not blocked by a wall, whereas someone with stencil buffer enabled will see just a tiny part of a player in the same situation. and without it, you can''t tell (well, it''s harder to) where you can see, and where you can''t, to make sure no one sneaks up behind you.

---
shurcool
my project

Hi!

About 140 FPS, the enemy finds and kills me in one sec.
AMD 800 MHz, GeForce 2, Windows XP

Max FPS = 176.4


[edited by - axodoss on June 26, 2002 11:52:06 AM]

Looks pretty good, can''t wait to see the end result.....

~20 fps windowed, but around 1-2 fps in fullscreen mode.
Don''t know what could cause the big difference between windowed and fullscreen mode..... maybe the TNT2 ??

AMD 750mhz
64mb ram
NVidia TNT2 32mb

I''ve got a TNT2. Around 1.25 fps for me as well.

"Don''t be afraid to dream, for out of such fragile things come miracles."

yep, i've looked around and found out that stencil buffer is free if you use depth testing, so it's cost is no more than depth testing. which isn't bad actually. the problems arise with older video cards, so like i said, i will probably have an option to toggle it. but weird things like difference in fps in windowed and full-screen mode (most of the time full-screen gets (and should get) better fps!), i'm worrying that it might be glfw's (window creation library) fault. i will look into that. thanks to all for testing this demo. (and more fps scores are welcome)

ps. i made the bot walk, instead of run, giving you an advantage, and making things easier. i also made it display the fps in-game, instead of in window title, letting you know what your fps is in full-screen mode. i might update it with more minor fixes, so try it again sometime.

---
shurcool
my project

[edited by - shurcool on June 26, 2002 12:43:55 PM]

Thats really neat!... 130fps or so Died before i could even move

p4 1.7ghz
640mb ram
geforce2 gts 32mb
winxp

yeah. I realise what you were saying. what I meant was that if you drew the map again (not enimies) in a different colour (eg, red) to fill in the black, it''d look a lot better. so you''d be able to see the map, but not the enimies.

>>~20 fps windowed, but around 1-2 fps in fullscreen mode.
Don''t know what could cause the big difference between windowed and fullscreen mode..... maybe the TNT2 ??<<

as davepermen saiz u need to be in 32bit colour for hardware stencil to work (obviously when its fullscreen u have 16bit colour)
this is for practically all cards (including tnt2)

http://uk.geocities.com/sloppyturds/gotterdammerung.html

quote:

---

Original post by RipTorn
yeah. I realise what you were saying. what I meant was that if you drew the map again (not enimies) in a different colour (eg, red) to fill in the black, it''d look a lot better. so you''d be able to see the map, but not the enimies.

---

riptorn!
that''s exactly what i did in eX0_4; u haven''t tried it, have you? well, here''s a shot:


zedzeek,
i''m working on that problem, it seems to me that (from the specification for nvidia riva tnt2 vid card) tnt2 cars (and many more) do support stencilling, so it must be something wrong w/ what i''m doing. anyway, i read that most card (including tnt2) have a 32 bit z-buffer (depth buffer) w/ 8 bit stencil buffer. in eX0_4, i allocated 0 bits to z-buffer (since i don''t need depth - it''s 2d), and 1 bit to the stencil buffer. in this executable, i changed that to 32 bits z-buffer w/ 8 bits stencil buffer, just like the specification says. please try it, and tell me if it works for you (never mind the name, cuz tnt2 and matrox g400 max are very similar):
download the executable here. (if you have matrox g400 max or tnt2, please give it a try)

thanks to all.

---
shurcool
my project