"Ringing" on objects in CAVE system

[SteveEB]

We use Vizard 4.0 with a CAVE system to display 3D images.

We have begun to use light-baking techniques in Autodesk 3ds Max 2011 to enhance the realism of our scenes but have recently run into a problem. There are rings of different colors on flat surfaces that look similar to oil when we put on 3D glasses. The color distortion is not visible with the glasses off.

The glasses are shutter glasses, alternating on and off for each eye to produce the 3D effect.

We have had Mechdyne, the company that manufactured our CAVE out to look at the issue and they believe that it is a software issue, and not one inherent in the hardware (projectors/mirrors/screens).

Again, this issue only happens with light-baked models, viewing the scene with the 3D glasses on.

I have attached an image that demonstrates the problem. This was taken using a cell phone camera, with the 3D glasses held over the lens. Ignore the vertical shaded highlight lines that are an effect of holding the glasses over the lens. In this image, the problem is the rainbow-like effect in the upper center of the picture (the wall should be all one color). I will try to get a better image with a better camera tomorrow.

Any help would be greatly appreciated.

Thank you.

[farshizzo]

Are you using quad buffer stereo with DLP projectors?

[SteveEB]

Yes we are.

[farshizzo]

This sounds like it could be caused by the DLP "rainbow effect". Do you notice the problem when the viewpoint is still or does it only appear while the viewpoint is moving?

[SteveEB]

The issue happens when the scene is stationary, the image that I posted was just one example, this problem actually occurs on just about every flat surface in the scene. Moving doesn't seem to affect it at all.

The rainbow bands become more pronounced based on the angle at which we view them. Additionally, we can see them if we view them with shutter glasses with only one channel active. The issue also occurs in non-light baked scenes, but only in cases of extreme gradients from black to white.

We will try to use a DLSR camera with a slower shutter speed to get a better image tomorrow.

[farshizzo]

Again, this sounds like a common DLP color wheel issue. The "rainbow effect" usually appears in high contrast areas. Do you notice this problem with other programs? Are your shutter glasses properly synced with the projector?

[shivanangel]

Hi Farshizzo, this is George who runs the lab that SteveB is talking about.

We have DLP but they are 3 chip, so we would not get a rainbow effect that you are talking about.

Any other ideas?

Our shutter glasses are real 3D CE4 and as far as we know we can't adjust the glasses timing, but only the projector dark interval.

The Mechdyne crew out here servicing our CAVE says this is most likely a software issue, as the problem doesn't seem to occur in anything they've supplied.
We were wondering if it was a ghosting issue, but at this time we a being told it probably isn't.

Thanks,
George

[shivanangel]

Here is what I hope to be a better image.
A side by side of the original image and an attempt to enhance it.
It is the gradient we are dealing with, where it seems like all the midtones disappear and instead we get a nasty green or pink color in its place.
I know it is very noisy, and I'll try to get a better picture tomorrow with a camera that allows for slower shutter speeds.

~George

[tobin]

To troubleshoot this, do you have the ability to view the same or similar gradient in active stereo independently of Vizard? The fact that you don’t see it without the glasses suggests that it’s a timing issue between the glasses sync and the DLP light engine. I think it’s worth a few tests to see if you replicate the artifact in other tools. Besides using a different 3D renderer for the same scene, you could try a simple 3D stereo picture viewer such as what nVidia provides with their “3D Vision Driver” (see this page for details: http://www.nvidia.com/object/3d-vision-photos.html). For the stereo picture test, I’d suggest just loading the identical smooth gradient in both left and right eyes, engaging active stereo, and viewing with the glasses to see if that also causes the artifact. My prediction is it will.

[shivanangel]

I'll give it a shot tomorrow, that you very much Tobin.

[shivanangel]

Dear Tobin and Farshizzo,

Thank you for helping us with our problem.

We spent some time playing with our projector settings and were able to directly modify the severity of the problem. We are going to spend some time trying to identify the optimal ranges to set our projects at, but just playing around with the settings we noticed the effect could completely disappear or become far worse.

We also successfully photographed the issue and I thought I'd post two images in case anyone else has a similar issue in the future.

Thank you,
George