Game Development Community

Glow Buffer problem on 6200?

by Stan Miskiewicz · in Torque Game Engine Advanced · 09/26/2005 (2:07 am) · 26 replies

So I'm having some problems with the latest version of the shader engine's bloom effect. I'm basically seeing the bloom effect not clear the buffer properly, so every time the effect starts it just draws over everything without clearing. This ends up in a sort of motion trail look.
www.okita.com/files/ss03.jpg
This is the same strange effect in the demo, just double checking against the starter.fps
www.okita.com/files/ss02.jpg
This is a bit of the strange effect I'm seeing on a geforce go 6200
compiled with .net 7.1, latest d3d9 aug. update. and latest video driver. I thought that it might be my video driver at first but I tried some of the directX demos to see if they would show the same effect.
www.okita.com/files/ss04.jpgthis is a similar use of the bloom/glow effect.

www.okita.com/files/ss01.jpgthis is another d3d app running with bloom without the sort of trail effect.

Has anyone else seen anything like this? I'm hoping to get this bug into the next update.
Thanks!
Page «Previous 1 2
#1
09/26/2005 (3:59 pm)
I found this on a previous thread, could this be related?

James Thompson
Member Posted: Jul 22, 2005 11:28

Also it says glowbuffer.cs is no longer needed but in console.log it says:
Missing glow buffer datablock. Check glowBuffer.cs

Edited on Jul 22, 2005 11:34
#2
09/26/2005 (11:29 pm)
Can you test your game build on another machine? Burn it to disk and see if another
shader equipped machine works. If it does work ok then it's a card specific issue.

If it doesn't then it's your code. Just a process of elimination.
#3
09/27/2005 (2:06 am)
I'll take a look on an ATI card tomorrow at work. I'll check it out on a geforceFX as well. My desktop here at home is a geforce 5900, no pixel shaders, and to think I paid 400 canadian for that when it came out.

My first impression is that it's something specific between TSE and my video card.
The DirectX example demo's using the same glow buffer seem to work fine. Which leads me to think that TSE deviates from the microsoft stuff. Not that deviating from microsoft is a bad thing.

I tried using the HEAD release and Release_0_3_0 of the TSE, and both had similar looking results. No modifications as I'm just getting started.

I'm also thought there might be a driver issue, but once again the glow buffers from the other d3d demos worked fine...
#4
09/27/2005 (11:25 am)
www.okita.com/files/ss05.jpg
So this is the same compile on another machine, working fine. So, I'm going to guess that it's specific to the GeForce Go 6200. I'll take a look at a friends machine running a geforce go 6600, I'd be curious to see if the same trail artifacts will show up on his machine as well.

Might be an nvidia portable issue, Is there anyone else out there running a geforce go ?
www.okita.com/files/ss06.jpgthis is another screen shot of the same thing on the 6200, as you can see, its pretty broken. Anyway, I trust this will get fixed up before the final release.

good work so far!
#5
10/04/2005 (12:36 pm)
I've seen this behavior before, it's almost certainly a driver problem or a problem with your card.
#6
10/04/2005 (12:45 pm)
I see this exact same problem with my Laptop's 6200 Go (Turbocache) as well. I've tried several different drivers, and all exhibit the same problem.

The problem only manifests when I'm running the engine fullscreen. In windowed mode it works nearly perfectally, although the water flickers when walking around, and the terrian seems to "morph" a bit right after loading.

If there is a specific driver you'd like me to try, I will do so. (if it is available for the Go series)
#7
10/05/2005 (4:47 pm)
Go series drivers are slow to be updated since its up to the manufacturer to post updates. I hear thats due to how the chip interfaces with the motherboard since it is common to have a more direct connection between the chip and the motherboard, skipping a regular PCI like interface.

I'm also wondering how the glow artifact doesnt show up in the other directX demos using the same method to draw. Are the dx9 SDK examples using a different method altogeher?
#8
10/13/2005 (1:08 pm)
Yeah, they are probably doing something slightly different that doesn't affect the driver in the same way. You are pretty much f#$ked on driver updates for laptops because like Stan said, it's up to the laptop manufacturer to update.
#9
10/13/2005 (1:17 pm)
I'm currently using a hacked version of Nvidia's current (78.01) driver, and it has the same problem. The hack is fairly slight; the inf file was just tweaked to enumerate the Laptop cards. All other software I've tried is working perfectly. (Empire Earth 2, Guild Wars, World of Warcraft) Of course I have no idea if the glow effects in Guild Wars are implemented the same way as those in TSE.

The water in TSE also flickers rather badly when you walk around or rotate the view. This happens in both windowed and fullscreen modes, although the glow buffer issues happen only while in fullscreen.

These problems also exist with the latest drivers from my laptop manufacturer, and two older hacked driver sets I've tried. (both in the 77.xx series)
#10
10/25/2005 (1:05 pm)
@Stan...

Dude a 5900 has shaders.... ?? It's got up to 2.0.
". I'll check it out on a geforceFX as well. My desktop here at home is a geforce 5900, no pixel shaders, and to think I paid 400 canadian for that when it came out."
#11
10/28/2005 (5:41 pm)
We are having the same issue at ATOMIX Productions. We have the oct 2005 build of DirectX, the latest video drivers as of today and we have the same problem. We have a 5700LE on 1 machine with shaders up to 2.0 and a GeForce 6200 on another, the 5700LE works greAT But the 6200 CHOKES like the pictures above. All of the directx samples in the oct 2005 build work great.
#12
11/26/2005 (8:12 am)
I have the same problems with the glow buffer on a NV 6610. And same water flickering problems as described. This is not a laptop
system, anbd I run latest drivers from Nvidia on this particular machine.

Dan
#13
12/13/2005 (1:33 pm)
Just curious, do those of you having problems with this have Anti-Aliasing turned on in your driver settings?
#14
12/20/2005 (4:54 am)
No its off.
#15
12/21/2005 (8:42 pm)
Have you tried the 81.95 Forceware? its up at Nvidia.com
#16
12/27/2005 (1:14 pm)
Brain: I'm having some similar problems with the glow buffer with AA turned on as posted in this thread.

These artifacts are from Sky Glow(using red as the color).
AA X2:
img349.imageshack.us/img349/5615/screenshot040000019pp.png

AA X4(not much of a difference):
img349.imageshack.us/img349/4483/screenshot039000019ip.png

And here's one of some funky behavior of the lights in the Orc base(at most angles there's no glow at all).
img336.imageshack.us/img336/9276/screenshot038000021du.png

I don't have any problems with AA off(aplication controlled) in either full screen or windowed. I'm running on a GeForce 6800. I think this really needs to get adressed as I don't think anyone will want to be running without anti aliasting turned on.

Kinda related: Is it possible to control AA from TSE yet? I've searched the source a few times for some kidn of pref, but haven't turned anything up.
#17
12/28/2005 (9:15 am)
Randy,


yes thanks, I run 81.95
#18
01/03/2006 (6:30 pm)
@Josh - yes, TSE will have it's own AA controls. Some of the glow buffer code will need to be re-written to support this and I'm hoping that change will also solve 6200 and laptop problems. We'll see when we get there.
#19
01/11/2006 (7:37 am)
Just like to check in and report that I am also having problems with the glow "smearing" all over everywhere it's been. I'm using an nVidia GeForce 6600 GT 128mb PCI Express 16x card with Windows XP Home Edition SP2 and the latest stable nVidia ForceWare drivers (81.98).

Brian, since this has now been reported on a 6610 (whatever that is), a 6800, and my card at least, I think it's safe to say this problem's scope extends beyond 6200 and laptops. Perhaps it affects all 6x00 series nVidia cards with recent drivers, whcih would make it quite a serious issue indeed. I'm perfectly content to wait (this is alpha code, after all) but make sure the problem gets looked at.
#20
01/11/2006 (11:38 am)
@Sam, I can believe that it extends beyond the 6200, but it's not on all 6xxx cards as our 6600 and 6800 cards do not show this behavior. If the problem persists after the rewrite, we will be contacting nVIDIA for a fix however.

@Josh - not yet, but we'll have it (and anisotropic settings) for MS3. I believe the glow buffer is what's holding it up.
Page «Previous 1 2