postprocessing effect chain graphical anomaly

No, with other postprocessing projects, such as the shader postprocessing workshops, it does not happen here. But it's hard to tell because the effect is rare. It looks as if the data buffer on the 3D card jumps out of sync for one frame dependent on the content of the image. I'll check the problem in more detail in the next days.

If you have a _simple_ project where this problem happens too, please send it to the support. Maybe the reason is something in the shader code.

I just sent a small level to support. Hopefully this helps.

When will this be fixed?

After purchasing and putting Gstudio7 on my new computer I have this problem too(as in Slin's Photos). But it is random to an extent. It only happens on certain levels at a certain area all the time. As described here, it has to do with the new postproccessing commands, it affects the bloom, DOF, and heathaze shaders. The DOF and heathaze shaders are the recent ones just released as "Shade-C" in user contributions.

When I turn off all post processing, the problem goes away. It also seems to affect areas with more particles effects, I can even make it happen in areas it otherwise wouldn't using certain weapons that have more particle effects and sprites. And this seems to have a greater effect when the camera is moving fast. But it does happen when the camera is still. It also happens in fullscreen, just not as often.

Comp Specs:

Nvidia 8880 Ultra
Intel Core2 QX6850 Extreme @ 3.0Ghz
Windows Xp Media Center, latest updates
Latest Nvidia Drivers(Forceware 169.21)

After more testing, I noticed it's very widespread and appears in all my levels quite frequently. This is a pretty bad problem...

Yeah, it's still a large problem for me as well. I've been working on a postprocessing related project that I will need to submit in less than a month, and it appears this glitch will exist in my published project. Perhaps only for users with certain video cards, but still...

The 8800 series are very common though, and that is a problem.

I have already forwarded this problem to nVidia, but have no response yet. So I can't say when it will be fixed.

And I am assuming they will respond and fix it soon. Please update as soon as you fix it, it is a terrible thing to not be able to use any of the post processing updates for lite-c, this is why I converted to lite-c in the first place. Thanks.

According to my experiences with nVidia in the past, when they can reproduce a problem it is usually fixed in the next or the next-but-one driver update. I could not easily reproduce the problem here. I tried it on several machines but it happened not very frequently. I hope that nVidia has some worst case configs where they can better reproduce such problems.

I am mostly worried that it will take a couple more months to get this fix. Nvidia's last driver release was back in December. It looks like they are readying another one for release soon(I downloaded the beta but it did not fix it). And considering this thread was started 2 months ago, and they never responded yet, it would seem a long shot that this will be fixed. If this is a problem with the 8800 series, it would most likely affect the new 9800 series as well, which is alot of graphics cards without post-processing. I am looking for a publisher very soon for my project and this is a real bad bug, especially considering I want to showcase the latest effects(bloom, DOF, haze) in my project.

I noticed that it is extremely frequent on this machine, and it definitely must be connected to the 3dgs particle system somehow. As the anomalies on the screen are usually directed at particles, and my game uses alot of them for both gameplay and level effects. The anomalies appear very heavy with the more particle effects going on. I tried setting the NOPARTICLE flag for the post processing views but it did not change anything. What was the test system you had it on? Did it at least contain an 8800 card with the latest updates, and many different types of particle effects? Perhaps it's not an Nvidia problem but something related to the engine and particles(other games must have effects chains.. and they run fine)?

Yes, I'm afraid we can't do much to speed this up. You can argue that this problem only happens with A7 and you're possibly right, but this does not help - we can't fix it here because we are not connected to nVidia and don't have the source of the code where this happens.

The problem is probably unrelated to the 8800, it happens with the driver version 169 and is caused by rendering 3D and 2D together - I can tell this much. That's why you have it with postprocessing, which is 2d rendering, and why it happens more frequently when you have particles active, which are also rendered in 2D. You can try to improve the behavior by switching to 3D particles - for this set the experimental particle_mode variable at 1.

ooh, "experimental"! sounds cool. what does it do? is there any noticeable difference, or is it just how the particles are handled? and how is it different?

julz

I tried setting the flag but it made no difference on the bug. However, I must say, the 3d particles really look great! Although many of my effects don't work, and many are overly bright, I like the extra definition it seems to give. Please continue with this as an option.

As for the bug, would you be able to at least give me an assurance that this will be fixed(Nvidia will get back to you)? In the meantime, I think I will try out some older drivers and see if it's a driver problem, at least if it's a driver issue I will be able to work with an older driver for the time being. Thanks JCL.

I tried installing the old "163.75_forceware" drivers from November of last year and this is still a problem. It's not exclusive to the 169 drivers, or the latest ones. Did Nvidia reply back to this yet?

No results yet, but they asked for an app to better reproduce the problem, and I'm trying to put something together for them. It's sort of difficult as at the moment I can not see the effect here at all - there's something random to it. It only happened on two machines that both had the 169 drivers, so I assumed the driver was the reason.

I will continue until the problem is resolved or at least the reason is definitely determined. When there's something new I'll post it here.

The problem occours more frequently when there are many particles at many different positions in the level. And it also has to do with the number of stages on top of the main view.
The problem occours very badly in this project: http://files.filefront.com/DemoLevel+V05rar/;10064920;/fileinfo.html especaially if I also activate DoF and blurred Stencilshadows. This version is without source but I could give it to you if needed. Or just the postprocessing part (each effect can be applied through one function call) or something like that.
I already got used to the error, but within thisproject it is just too bad if even the movement seems to lag because of this (the fps stays nearly the same...). And it is hard to create screenshots with each effect visible and without any strange obstacles...

Thanks for the new file. Anything that reproducibly causes the effect can help.

Even with the new file I can't reproduce the effect here anymore. I've tried all machines that we have. First I thought that's because the engine I use is slightly different, but I also can't reproduce it with the 7.07 release version. nVidia couldn't reproduce it either.

I know that I've seen the effect, but it seems to be mysteriously disappeared.

Can everyone whose PC is suffering from this effect post his 3D hardware type and driver version here?

I´ve got an Xpert Vision 8800GTS with 640mb. I just reinstalled the ForceWare driver version 169.21.
Edit: I am using Windows XP Home and the latest DirectX 9.0c release. I know of two computers not having the problem with Windows Vista 64bit and an 8800GT and 8600GT.

I just send you an eMail with a downloadlink for my projectfiles as you asked me to in this thread. Press F1, F3 and F4 and go to the hangman on top of the hill and look to the house. That looks like this for me all the time:

Windows XP Media Edition
4 GB Ram
Intel Quad Core 6850
8800 Ultra
Any of the last 2 Official Nvidia Driver releases(but it happens with there new beta one as well)

A7 Beta that was released in March

I am thinking that you guys must not have a machine with an 8800 in it. It is so very visible here one every level, almost all the time. If I add even more post processing(DOF, Haze), and particle effects, it is all you see at all times. Which is frustrating since I want to showcase my latest effects in a video to a publisher soon but cannot.

I think later tonight I can try make a demo for you to try of this effect.

I sent a demo to your conitec.net address. The problem occurs about 50% of the time in this level for me. Here is a screenshot:

The same what slin has. With level with river and with slin's demo.
I have 8600GT, 32bit XP.

thats a warning from GOD.
"You can't copy the perfect mother nature"

if i may drop something too..

I had this problem before too but with A6, and now i also have it with Torque Game Engine but not Torque Game Engine Advanced.

I'm sure this problem is not related to 3DGS. I somehow think it has to do with the DirectX SDK, but not sure yet.

BTW i have an ATI Radeon X1600XT, so no nVidia this time.

What's the status on this?

Brief update. According to the reports I received so far, the problem seems to occur with several 3D cards and several drivers, but only under Windows XP. So it's apparently not directly related to the driver or 3D hardware.

I had it on one 8600 GT system here also, but that turned out to be temperature dependent, and thus was obviously a hardware problem. I can not observe it here on any other 8600 or 8800 system.

The pattern of the problem suggests that nVidia can probably do nothing on their end. My guess is now also that it's a DirectX bug. The next step for me would be to test if a change of the DirectX version has an effect on this problem.

Has anyone ever observed this on a non-XP OS?

What I don´t understand is, why this problem does not occour anywhere else for me, only in A7.
Even my probably very dirty postprocessing class for irrlicht (rendering with D3D9) works fine. All it does is to create a screenaligned quad with a shader on it. This is rendered into a texture, then there is another screenaligned quad with a shader on it which gets passed the texture and so on, the last "stage" renders to the screen.
I thought that is how it is basicly done? So what is different in GameStudio?

I agree, it's something A7 only. I've played a many different games and have never had this problem yet with the exception of A7. So I highly doubt it could be a DirectX bug, if it was, wouldn't every other game out there have this as well?

And I haven't observed it on a non-XP OS, but then again, the large majority of people use XP to play and develop, so that's probably why there is no one reporting it if it does exist on Vista as well.

yeah, i guess some newer version should be use instead of d3dx9_30

The reason for d3dx9_30 was that this DLL is available on a majority of systems.

Of course I don't know if the bug or whatever it is was fixed in any DirectX version. As it seems, it only appears on very few systems that only have XP in common. The only thing I know is that it's not a Gamestudio bug and that I can not reproduce or test it. Therefore, using a different DX version is the only thing I can do at the moment. If someone who has this bug volunteers, I can send him a special engine with the latest DirectX version for testing.

I could give it a try. The problem is very heavy on my system and I have got a cuple of easy and fast to use postprocessing shaders. My eMail adress if needed can be found in my profile.

Thanks. I'll send you the new version by Monday.

Some good news: After spending about $1500 for new test systems we finally found a system where the problem could be reproduced.

We believe that it's a bug in the nVidia Win2K drivers. It does not happen with any other nVidia drivers, such as the Vista drivers, even on the same system. The Win2K drivers are used for XP.

Hopefully with this new information we can find either a workaround or a way to get it fixed by nVidia.

Good news really.

Great to hear this! I hope things go smooth in finding a solution. And thank you for purchasing new equipment to reproduce this; I really appreciate your efforts.

Update: The problem looks like a command buffer overflow in the driver, caused by certain shader commands. That's why it only happened with certain postprocessing shaders.

I think we can implement a workaround so that it does not happen anymore even with the current nVidia Win32K drivers. Unfortunately this is too late for the upcoming update that I don't want to delay further. But it will be addressed after that.

Thanks for the update JCL, it's good to hear there might be a solution. Looking forward to test as soon as possible.