[vtkusers] Re: vtkVolumeTextureMapper2D

Lisa Avila lisa.avila at kitware.com
Mon Dec 20 11:01:07 EST 2004

Hello Julien,

The texture mapper will render your whole data set from the appropriate 
viewing direction, packing images into a larger texture for rendering. 
Since TargetTextureSize is 512x512 by default you will have a fair amount 
of waste. Also, note that the opacity/color transfer functions and shading 
are applied during every render (as the data is moved into textures) and 
the texture are created every render (not stored in display lists - at the 
time this class was written video cards did not have that much memory, and 
if you have shading turned on the textures change with every move of the 
camera or light). I would recommend trying to use 1024x1024 as a target 
texture size - that should fit either 6 or 9 slices into 1 texture (rather 
than the 4 that fit in 4 half size textures with a target size of 512x512). 
Also, you can reduce the number of planes rendered to provide interactive 
performance (at a loss of image quality) by using the MaximumNumberOfPlanes 

Over the next few months there will be a fair amount of development 
occurring in the volume rendering functionality of VTK - you may wish to 
keep an eye on this and try out some of the newer classes as they become 


At 07:45 AM 12/20/2004, Julien BARNEOUD wrote:
>Hi All,
>Is there anybody who could help me, or should I add some missing information
>to my first email?
>"Julien BARNEOUD" <j_barneoud at trophy.fr> a écrit dans le message de
>news:cps3us$omr$1 at sea.gmane.org...
> > Hi all,
> >
> > I am comparing quality and performance of the the different volume mapper
> > (ray casting and texture ones) and I have slower result with the texture
> > mapper.
> >
> > My configuration is a PIV 2.66 Mhz, with 1 GO RAM and an NVIDIA Geforce
> > GT graphic board.
> >
> > The dataset I am trying to render is made of 289 * 289 * 399 grayscale
> > voxels with 12 bits depht.
> >
> > I am surprized because the texture render is very slow (i.e. it is
> > to interact with the volume), and the graphic board has 256 Mo of Vdieo
> > so my entire dataset should be loaded and handled by the GeForce...
> >
> > I have tried to reduce the volume size using vtkExtractROI, I get
> > faster, but not really usable.
> >
> > Is there any attention to pay to the volume size: a 256*256*512 volume
> > be easier to fit  texture memory as it is aligned on a power of 2? Should
> > shift the data to be on 8 bits ? Anything else ?
> >
> > Thanks in advance for your help,
> >
> > Regards,
> >
> > Julien BARNEOUD
> >
> >
> >
> > _______________________________________________
> > This is the private VTK discussion list.
> > Please keep messages on-topic. Check the FAQ at:
> > Follow this link to subscribe/unsubscribe:
> > http://www.vtk.org/mailman/listinfo/vtkusers
> >
>This is the private VTK discussion list.
>Please keep messages on-topic. Check the FAQ at: 
>Follow this link to subscribe/unsubscribe:

More information about the vtkusers mailing list