VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Thu Jun 12 2003 - 11:07:01 CDT

Hi,
 Some comments about seeing both images at the same time, and
some info on how to determine which kind of bug it is:

On Thu, Jun 12, 2003 at 05:41:47PM +0200, Bogdan Costescu wrote:
> On Thu, 12 Jun 2003, Dr. Daniel James White PhD wrote:
>
> > I mean in one there is one image, the correct one for that eye,
> > in the other there are both images, the left and right at the same time.
>
> In my case there was also the correct image for that eye plus some
> leftover. If the left and right at the same time can also come from some
> leftover, I'd guess this would point towards a VMD or a driver bug...

It should not be possible to see both images at the same time.
The only way this can happen in OpenGL is if the application draws both
the left and right images to the same buffer without clearing in-between.
If this were the case, it would be due to a bug in VMD itself. This is
easily distinguishable from driver/hardware bugs because:
  - If the application draws both the left and right eyes to the same buffer,
    one of the two images will "win", i.e. the last image drawn will be
    rendered "on top" of the other one. This situation looks _very_ different
    from what one sees if you look at a normally functioning stereo
    display without your LCD glasses. Typically, if the application is
    at fault, you will see "jagged intersections" between the geometry in
    the left eye image and the geometry in the right eye image. (assuming
    the Z-buffer was not cleared). If the Z-buffer was cleared, then
    the result is that the last image drawn (right eye) will look normal,
    but the parts of the first image drawn (left eye) will show up in places
    where they are not obscured by the last image, where one would have
    normally seen the background color.
    OpenGL does not provide a means for an app to show both left and right
    buffers at once. You can only show one or the other. The only way to
    see pieces of both left and right images is if you draw them both to the
    same buffer, but the artifacts make this easy to spot if it ever happens,
    it looks VERY different from what you'd see if you took off your stereo
    glasses and looked at a normally operating stereo display.

  - If the problem is caused by the video driver or the emitter box hardware,
    or some combination thereof, you may see the left, right, neither, or both
    images in each eye, sync may be lost and regained intermittently, or
    combinations of these problems. This type of problem is distinguishable
    from bugs in the application because it isn't possible for the application
    to request "both" the left and right color buffers to be shown
    simultaneously, so if you see both the left and right images
    simultaneously and they look just like they do if you take the
    glasses off, this would indicate a driver or hardware problem of
    some kind. Other problems like "tearing" flickering GUI windows
    or strange colors in GUI windows in one eye but not the other, are also
    all problems that occur result from the video driver and hardware.

  - If you get different behavior with different programs (as was stated below)
    then its likely that the problem is in the video driver and not in the
    hardware. Stereo emitter hardware is pretty simple, so in general the
    OpenGL and video driver software are typically the cause of most problems
    I've run into. The "blue-line sync" stereo can be flaky with certain
    video cards when they are running in certain video modes, but typically
    it would be equally flaky with all programs, not just one.

My suggestion is to contact XiG and report the problems you are
experiencing. I suspect the problem can be easily reproduced by others
if it is a driver bug. If it were a VMD bug, I would have expected to
have seen it already on one of the 5-10 stereo-capable video boards
we have in our lab, since I haven't seen or heard about this double-image
problem before, my conclusion is that you've found bugs in the XiG drivers.
Talk to XiG and see if they have tested their stereo code with the
Radeon 8500 much. It should be easy to determine which one of the
problems is occuring by looking carefully at the "double image" and
seeing which of the 3 major types it is.

Thanks,
  John Stone
  vmd_at_ks.uiuc.edu

> > this is only with VMD, not on Bodil or Pymol
>
> Well, I'd say that the way each program uses the facilities offered by the
> driver is different. So if there is a driver bug, it might still not be
> exposed by all applications.
>
> --
> Bogdan Costescu
>
> IWR - Interdisziplinaeres Zentrum fuer Wissenschaftliches Rechnen
> Universitaet Heidelberg, INF 368, D-69120 Heidelberg, GERMANY
> Telephone: +49 6221 54 8869, Telefax: +49 6221 54 8868
> E-mail: Bogdan.Costescu_at_IWR.Uni-Heidelberg.De
> 

-- 
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: johns_at_ks.uiuc.edu                 Phone: 217-244-3349              
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078