Hi,

I've been working on an OpenGL volume rendering application for some months
now, and it's got to the stage where I really need some feedback from other
developers.

For info, the application is Windows MFC based (MS VC7) and apart from the
OFFIS DCMTK toolkit (for DICOM import), uses no external libraries. Some
features are more developed than others but a short list is as follows:

2D and 3D texture, alpha blended display
Maximum Intensity Projection (MIP)
Multi-Planar Reformatting (MPR)
Interactive fly-through
Stereoscopic window - anaglyph or dual viewport
Interactive transfer function adjustment
Iso-surface mesh generation
3D Watershed segmentation

It's optimised for Radeon 9000+ or GeForce 3/4 but I'd really like feedback
on as many different configurations as possible. I need to know:

What does it do well?
What does it do badly?
What doesn't it do at all?
When/why does it break?
What have I misunderstood?

The executable is available as a free download. Here's the link -

http://www.rmrsystems.co.uk/volume_rendering.htm

Please take a look and let me know, either via this newsgroup or by e-mail
(from the download) if you have any comments, suggestions or bug reports.

Thanks in advance,
Roger

hey,

all seems to be working just fine although on my
small (?) machine 128 MB I was getting virtual
memory problems when trying to make an isosurface.
I didnt pursue. All was a bit slow ( 5 fps or so )
Im running a Gforce4 card ~1.1GHz.

one suggestion: to be able to fly around and through
the data, clipping at the camera near-clipping plane
( at an arbitrary angle, dont know if the data allows
that ) otherwise let the intersecting point of the three
clipplanes always be at a fixed distance from the
camera and make the camera free to move around.

very impressive. just looking in someone's head, not
knowing the person is a bit eerie. Eerie but very beautiful.

J

--
-------------------------------------------------------------
  please tear the sticker off my eddress before use
-------------------------------------------------------------
"RSR" <rsr@sys.uea.ac.uk> wrote in message
news:bdbjbs$4vj$1@cpca14.uea.ac.uk...
> Hi,
>
> I've been working on an OpenGL volume rendering application for some
months
> now, and it's got to the stage where I really need some feedback from
other
> developers.
>
> For info, the application is Windows MFC based (MS VC7) and apart from the
> OFFIS DCMTK toolkit (for DICOM import), uses no external libraries. Some
> features are more developed than others but a short list is as follows:
>
> 2D and 3D texture, alpha blended display
> Maximum Intensity Projection (MIP)
> Multi-Planar Reformatting (MPR)
> Interactive fly-through
> Stereoscopic window - anaglyph or dual viewport
> Interactive transfer function adjustment
> Iso-surface mesh generation
> 3D Watershed segmentation
>
> It's optimised for Radeon 9000+ or GeForce 3/4 but I'd really like
feedback
> on as many different configurations as possible. I need to know:
>
> What does it do well?
> What does it do badly?
> What doesn't it do at all?
> When/why does it break?
> What have I misunderstood?
>
> The executable is available as a free download. Here's the link -
>
> http://www.rmrsystems.co.uk/volume_rendering.htm
>
> Please take a look and let me know, either via this newsgroup or by e-mail
> (from the download) if you have any comments, suggestions or bug reports.
>
> Thanks in advance,
> Roger
>
>
>

Thanks!

"j" <vdspek@xs4all.nl> wrote in message
news:3ef98094$0$49113$e4fe514c@news.xs4all.nl...
> hey,
>
> all seems to be working just fine although on my
> small (?) machine 128 MB I was getting virtual
> memory problems when trying to make an isosurface.
> I didnt pursue. All was a bit slow ( 5 fps or so )
> Im running a Gforce4 card ~1.1GHz.

Yes - the sample data doesn't really lend itself to the iso-surface
generation as there's no nice peaks in the histogram. I would have included
the CT data from the documentation but this is not public. Sorry.

>
> one suggestion: to be able to fly around and through
> the data, clipping at the camera near-clipping plane
> ( at an arbitrary angle, dont know if the data allows
> that ) otherwise let the intersecting point of the three
> clipplanes always be at a fixed distance from the
> camera and make the camera free to move around.

Having a free-floating clipping plane is a good idea and I'll implement this
soon. The fly through again is a little bizarre in the sample data but you
can move the camera by dragging on the preview windows or clicking
left(forward) or right(back) over the image itself. Is that what you meant
or have I missed the point?

>
> very impressive. just looking in someone's head, not
> knowing the person is a bit eerie. Eerie but very beautiful.
>
> J

Thanks again - I need this kind of feedback so I'm grateful for your time.

Best regards,
Roger

Hi,

It is a very impressive job.
Especially fly-through mode is a brake through.
It can work on $600 computer!!!

However there are some minor disadvantages of this cheap solution:

---------
It does not support shading for volume rendering mode. Am I wrong?
It is probably the reason why you need iso-surface generator;
otherwise it could be just an impulse opacity function case.
---------
The biggest data set I could play on my computer GForce4200-64MB(dual
Xeon2.8/533/2GBRAM) - 512x512x64/short. Probably because it may work
fast enough only if it can load all data set to video memory (probable
ratio � 512x512xshort per MB). As the result any custom 3D processing
of data set will require data set reloading to video memory.
---------

It is a very impressive job. I wish you good luck.

--George


"RSR" <rsr@sys.uea.ac.uk> wrote in message news:<bdbjbs$4vj$1@cpca14.uea.ac.uk>...
> Hi,
> 
> I've been working on an OpenGL volume rendering application for some months
> now, and it's got to the stage where I really need some feedback from other
> developers.
> 
> For info, the application is Windows MFC based (MS VC7) and apart from the
> OFFIS DCMTK toolkit (for DICOM import), uses no external libraries. Some
> features are more developed than others but a short list is as follows:
> 
> 2D and 3D texture, alpha blended display
> Maximum Intensity Projection (MIP)
> Multi-Planar Reformatting (MPR)
> Interactive fly-through
> Stereoscopic window - anaglyph or dual viewport
> Interactive transfer function adjustment
> Iso-surface mesh generation
> 3D Watershed segmentation
> 
> It's optimised for Radeon 9000+ or GeForce 3/4 but I'd really like feedback
> on as many different configurations as possible. I need to know:
> 
> What does it do well?
> What does it do badly?
> What doesn't it do at all?
> When/why does it break?
> What have I misunderstood?
> 
> The executable is available as a free download. Here's the link -
> 
> http://www.rmrsystems.co.uk/volume_rendering.htm
> 
> Please take a look and let me know, either via this newsgroup or by e-mail
> (from the download) if you have any comments, suggestions or bug reports.
> 
> Thanks in advance,
> Roger

"George Buyanovsky" <buyanovsky@attbi.com> wrote in message
news:d7221cc2.0306251120.303b9a41@posting.google.com...
> It is a very impressive job.
> Especially fly-through mode is a brake through.
> It can work on $600 computer!!!

Hi George - thanks for your comments :-)

> It does not support shading for volume rendering mode. Am I wrong?

It doesn't yet support shading but it may do soon. This was not a
requirement for our current research so I only played with it a little. It's
more important for us to generate a mesh - maybe I'll release more details
later. Anyway, I guess I could add shading with a pre-processing step to
generate normals and then use a fragment shader? At the moment I'm using
fragment program on ATI for the transfer function so maybe I'll play around
some more.

> The biggest data set I could play on my computer GForce4200-64MB(dual
> Xeon2.8/533/2GBRAM) - 512x512x64/short.

Yes - if the texture doesn't fit onto your gfx card, it won't work. I
haven't used bricks/tiles in this version.Were you using 2D textures or 3D
textures? I guess 2D but you can change the default from the Edit/Options
menu.

> It is a very impressive job. I wish you good luck.

You are very kind. Thanks for taking time to look at it ...

Best regards,
Roger

> Having a free-floating clipping plane is a good idea and I'll implement
this
> soon. The fly through again is a little bizarre in the sample data but you
> can move the camera by dragging on the preview windows or clicking
> left(forward) or right(back) over the image itself. Is that what you meant
> or have I missed the point?

what I mean is that you would basically use the
near-clipping plane of the camera as the slice.
or is this already in the fly-through and have I missed
the point because my computer was lagging too much ?

J

-------------------------------------------------------------
  please tear the sticker off my eddress before use
-------------------------------------------------------------
"RSR" <rsr@sys.uea.ac.uk> wrote in message
news:bdbvue$bnq$1@cpca14.uea.ac.uk...
> Thanks!
>
> "j" <vdspek@xs4all.nl> wrote in message
> news:3ef98094$0$49113$e4fe514c@news.xs4all.nl...
> > hey,
> >
> > all seems to be working just fine although on my
> > small (?) machine 128 MB I was getting virtual
> > memory problems when trying to make an isosurface.
> > I didnt pursue. All was a bit slow ( 5 fps or so )
> > Im running a Gforce4 card ~1.1GHz.
>
> Yes - the sample data doesn't really lend itself to the iso-surface
> generation as there's no nice peaks in the histogram. I would have
included
> the CT data from the documentation but this is not public. Sorry.
>
> >
> > one suggestion: to be able to fly around and through
> > the data, clipping at the camera near-clipping plane
> > ( at an arbitrary angle, dont know if the data allows
> > that ) otherwise let the intersecting point of the three
> > clipplanes always be at a fixed distance from the
> > camera and make the camera free to move around.
>
> >
> > very impressive. just looking in someone's head, not
> > knowing the person is a bit eerie. Eerie but very beautiful.
> >
> > J
>
> Thanks again - I need this kind of feedback so I'm grateful for your time.
>
> Best regards,
> Roger
>
>
>
>