From: Sean Spicer (spicer@bme.stanford.edu)
Date: 04/19/2000 09:41:19
Mark,
The MPR distortion problem is an Octane architecture problem. My guess is
that you are rendering in perspective, which means that in order to get a
(semi-)proper image one must correct for the perspective transformation
when applying the texture. This is called perspective correction and is
usually done by the hardware...except in the case of Octane, where there
is no perspective correction hardware. The workaround that I use
(although not always pleasant) is just to render only in ortho when I'm on
Octane.
I can't answer the fog or texture subload questions, but on the
dithering...
The impact graphics set is close to 6 years old now. It was never
designed from the ground up to be a high-performance volume-rendering
board, and the issues you have with the visual being sub-standard at
resolutions above 1024x768 are just a demonstration of the old-age of the
technology. Personally, I do almost all my work on an Octane MXI (which
with larger frame-buffer memory can handle full-resolution rendering with
no dithering), however this still has the limitation of no perspective
texture correction, no post-interpolation color lookup tables, and no true
tri-linear interpolation hardware. For this reason, I also work
extensively (and will begin to do more so in June) on an Onyx Re2. This
machine, although the same vintage as Impact, corrects many of these
problems, and can be had on the open market for less than you'd think.
Generally, at this point Onyx or Onyx2 is you best bet for volume
rendering, but I'm sure sgi won't be much longer with an update to the
graphics on Octane.
Hope that helps!
sean
On Wed, 19 Apr 2000, Mark Davey wrote:
> Hello.
> I have been working with an Octane over the past couple of months and I
> have a number of problems with respect to the use of this machine with
> Volumizer, although I suspect that none of them are due explicitly to
> Volumizer itself but to the architecture of the machine.
>
> Firstly,
> When using Volumizer to produce a Multi-Planar reformatting of the
> data, where the plane is kept stationary with respect to the data,
> considerable distortion of the texture can be seen as the data is
> rotated. Has anyone else seen this? Is it an Octane problem? This also
> appears to make 'fly-throughs' behave very strangely indeed as the data
> set gently morphs as you pass through the anatomy!
>
> As I have previously mention I was having trouble using glFog at the
> same time as colour lookup tables. As pointed out, this is mentioned in
> the man pages for lookup tables. Does anyone know if the new Octane
> graphics card will have this problem?
>
> I have tried to use
>
> voAppearanceActions::textureSubload(...);
>
> on an Octane only to be told that this command is not supported on the
> Octane... Is this a fundamental problem with the Octane or is it
> Volumizer at fault here?
> This means I cannot do interactive editing of the volume.
>
> Although I have 'solved' the dithering problems encountered when
> displaying textured graphics on the Octane by reducing the screen
> resolution to 1024x768_76 this strikes me as being a poor compromise as
> PC's can do better than this at a small fraction of the cost. Surely a
> standard Octane should be able to display adequate graphics as a
> standard resolution of 1280x1024.
>
> Have I got a sub-standard machine here?
> If not then the pratical uses for Volumizer (and probably any other
> volume-slicing-based volume rendering API) seems limited on the Octane
> (especially
> considering the goemetric distortions)
>
> What would be the platform of choice for Volumizer?
> What is required to get around the above problems? Do I have to use an
> Onyx???
>
> Regards
> Mark....
>
>
> --
> From: Mark Davey
> Dept. of Medical Physics, UCL.
>
> TEL: +44 (0)171 915 1673.
> FAX: +44 (0)171 837 9279.
>
> Institute of Laryngology and Otology,
> 330 Gray's Inn Road,
> London.
> WC1X 8GE.
>
>
>
___________________________________________________________________________
Sean Spicer Stanford University Medical Center
Biomechanical Engineering Division of Vascular Surgery, Suite H3642
Cardiovascular Biomechanics Lab Stanford CA, 94305
Telephone...650.723.1695
Fax.........650.723.8762
http://solvedeath.stanford.edu/~spicer
This archive was generated by hypermail 2b29 : Wed Apr 19 2000 - 11:32:42 PDT