Hi Klaus,

Apologies for the delayed reply. We are aware that there are still some
issues with rendering large datasets (e.g. program aborts under
Windows), and your report (black window/memory leakage) adds to this.
We'll have a look at this for the next MITK release (due in a couple of
weeks).

> 3. When I click the left mouse button in a 2D window, then it is like
I
> clicked it in the 3D window. The quality gets worse and as soon as I
> release the mouse button it is rendered in high quality again. Seems
to
> be unnecessary (it just slows down everything).

This re-rendering occurs since when clicking into a 2D window, the 2D
planes are moved accordingly. Since these planes are (by default) also
displayed in the 3D window, it becomes invalidated and has to be
rendered again (the planes should move also in 3D). The behavior of
sequentially switching from low resolution to high resolution is due to
the Level-of-Detail (LOD) mechanism for volume rendering: the low
quality "preview" image is rendered fairly quickly, and the high
resolution volume is rendered later on, when the user stops interacting.


> 4. I am using a debug toolkit chain (vtk, itk, qt3, mitk, everything
is
> compiled as shared) and it took 2:45 minutes to render the volume.
Even
> in the release version it takes very much time.
> Some time ago, I implemented a small example just using vtk, itk and
> wxwidgets to load a volume and to render it and it was never that
slow.
> So why is this function in mitk/mainapp that slow? What happens under
> the hood? (I noticed for example, that a copy of the volume is made)

This is probably due to the configuration of the volume rendering used
by MITK. The high-res mapper of the LOD approach (you can have a look at
its implementation in mitkVolumeDataVtkMapper3D) uses rather small
sample distances, making the overall quality better, but the rendering
slower. In the current version VTK also supports a 3D texture based
volume renderer, but it its implementation is still rudimentary (as
compared to state-of-the-art approaches) and is therefore not yet
utilized by MITK.


Regards,
Mathias

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
mitk-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mitk-users

Reply via email to