Weird problem with GeForce 4 Fullscreen
Blitz3D Forums/Blitz3D Programming/Weird problem with GeForce 4 Fullscreen
| ||
| Ho-hum... looks to me like it's Z-fighting of some sort... have you tried to optimize your CameraRange values for maximizing the Z-buffer accuracy? Also although I'm not 100% sure but windowed and fullscreen apps might use different Z-buffer depth depending how's your desktop bit depth is set... I do know that apps can LOOK different when you compare fullscreen and windowed screens... my work R97p for example has different gamma settings at least in windowed and fullscreen mode... IMO running apps windowed is like asking your butt kicked because Windozer gets in the way then and can ruin your gfx. |
| ||
| Well for various reasons I did have the camera range set to 0.1,30000. I changed it back to 1,1000 to see if that eliminated the problem, it did not. I use 0.1 for the near range as I have particle effects that get chopped if the camera is very near them. I use 30000 for the far range as I have a distant comet, sun and back ground. But it appears not to be the CameraRange causing the problem. And I would have thought that in windowed odde things might misbehave not in full screen mode... |
| ||
| Are you running in 32-bit depth? Maybe your desktop is but the fullscreen game isn't? I have found that colour depth can have funky effects with the Z-buffering. |
| ||
| Few articles about Z-buffers: http://www.extremetech.com/article2/0,1558,1171859,00.asp http://encyclopedia.thefreedictionary.com/Z%20buffer |
| ||
| I have a GF4 and have had the same issues, 16 bit res is the easiest way to cure it. |
| ||
| GW, really? What causes it? Are you saying run your desktop AND app in 16 bit? Or just run the app in 16 bit? It seems strange that it only does it on certain entities and only in certain situations. |
| ||
| "GeForce 4 (latest drivers installed)" I suspect thats the problem, every time I have tried a driver set about the 44.03 on a geforce4 I have problems like that, it seems almost every other driver they get it wrong then fix it the get it wrong ;) |
| ||
| Well the installed drivers are 61.76, however 61.77 are available so they are not the VERY latest. I will try that too. Keep up the suggestions... |
| ||
| OK. An update. Desktop 16 bit, app 16 bit = PROBLEM! Desktop 16 bit, app 32 bit = ALL OK! Desktop 32 bit, app 16 bit = PROBLEM! Desktop 32 bit, app 32 bit = ALL OK! From this I deduce that running the app in 16 bit is the problem. This is good news as it means I have to run in 32 bit. The app happily runs the update world and render world at 85 FPS in 800*600*32 AA on. I will ensure that the default bit rate is 32 bit and document this 'feature' in the help docs. Thanks for all your help. |
| ||
| The 16 bit mode uses w buffering. So if you want to use w buffering anyway: In fullscreen after you set the graphics mode, enter the line Wbuffer True |
| ||
| BloodLocust for President! You are the man. Now we allow the user to disable w-buffering. Cool. |
| ||
| Your camerarange is still very wide in range. 0.1 to 1,000 might sound like having 1 to 10,000, but its much worse. You should scale everything up by 10 for a massive improvement. Your near number should be as large as you can manage, the far number can be much bigger with little effect. You can of course force 24(32) bit or the Wbuffering. Take a look here for more info: http://www.sjbaker.org/steve/omniv/love_your_z_buffer.html |
| ||
| It's blitz man. Especially on my g5, it's *very* buggy in windowed mode.(As above, polys disappearing. (Close ones..and not due to z-fighting)...a real bloody mess.(In 16bit mode..it's better in 32bit, but not perfect) In full screen it's fine though..so there's a chance you have a whole new problem..or it is just something as simple as z-fighting. |
| ||
| I also recommend doing two renderings, one for the space background and another for the nearby objects. Use a different camerarange for each rendering. |
| ||
| Beaker, that is a VERY interesting link - thanks! But I have a question. How do I stop very near particles from being clipped if the near distance is set to say 10? |
   |