[vtkusers] Depth Peeling on Intel HD 4400

[Burlen Loring]

Hi Michale,

In the test program you pointed to, not all combinations of inputs are 
valid or will work everywhere. In your "Problem 1" you are identifying 
some scenarios that don't work on your system, but are valid use cases 
on other platforms.

> Problem 1 Summary: calling renderWindow->SetOffScreenRendering(true), 
> seems to break depth peeling
the run time selection of on or offscreen rendering doesn't work with 
modern Mesa drivers on Linux. It may work on other platforms/drivers (eg 
Windows).

> switching the parameter 'bool doitOffScreen' in the call to bool 
> IsDepthPeelingSupported(...) from true to false, charges the result of 
> IsDepthPeelingSupported(...) from false to true and results in the 
> image being displayed correctly.
Given that run time selection of offscreen rendering is not supported on 
your system, the behavior you're describing sounds correct. With your 
system/configuration you should just use on screen rendering.

In your "Problem 2" you report an OpenGL error in the 
vtkOpenGLRenderer::RenderPeel. This is perhaps a bug in VTK's depth 
peeling code, but far more likely a bug in Mesa Intel driver itself. 
Someone (I can't do it because I don't have an hd 4400 linux system 
handy) needs to step through this section of code and see which GL call 
is the source of the error and see if there's an easy work around/bug 
fix for VTK and if not disbale depth peeling for this driver/hardware 
combo and report it to the Mesa developers.

Burlen

On 1/11/2015 4:47 AM, Michael Arnold wrote:
> Hi,
>
> Am trying to get depth peeling to work to improve the display of a 
> vtkPolyData surface resulting from a delaunay triangulation. Surfaces 
> are often displayed in the wrong order.  Have been unable to get depth 
> peeling to work and applying depth sorting still results in a display 
> with surfaces in the wrong order.
>
> VTK 6.10
>
> Intel Core i5-4210U
> Intel HD Graphics 4400
> OpenGL version string: 3.0
> Mesa 10.3.5
> Gnome Version 3.14.2
> Fedora 21
>
> Based on glxinfo (attached) and 
> http://www.itk.org/Wiki/VTK/Depth_Peeling, I find all the relevant 
> extensions are available for depth peeling.
>
> When experimenting with:
> http://www.vtk.org/Wiki/VTK/Examples/Cxx/Visualization/CorrectlyRenderTranslucentGeometry
>
> Problem 1 Summary: calling renderWindow->SetOffScreenRendering(true), 
> seems to break depth peeling and results in 
> renderer->GetLastRenderingUsedDepthPeeling() and hence 'bool 
> IsDepthPeelingSupported(...)' always returning false.
>
> Problem 1 Detail:
> http://www.vtk.org/Wiki/VTK/Examples/Cxx/Visualization/CorrectlyRenderTranslucentGeometry
> As provided, always results in:
>
>     DEPTH PEELING SUPPORT: NO
>
>     CHOSEN MODE: *** DEPTH SORTING ***
>
> (A blank window is displayed under each of the 3 call scenarios on the 
> web page.  This same result is achieved under both x and wayland).
>
> But switching the parameter 'bool doitOffScreen' in the call to bool 
> IsDepthPeelingSupported(...) from true to false, charges the result of 
> IsDepthPeelingSupported(...) from false to true and results in the 
> image being displayed correctly.
>
> Problem 2: Applying the above code under QVTKWidget, we still have the 
> issue with GetLastRenderingUsedDepthPeeling(), but applying the 
> workaround above and setting 'bool doitOffScreen' to false, results in 
> new error message during rendering:
>     ERROR: In /.../VTK-6.1.0/Rendering/OpenGL/vtkOpenGLRenderer.cxx, 
> line 1068
>     vtkOpenGLRenderer (0x2c5a5e0): failed after RenderPeel 1 OpenGL 
> errors detected
>     0 : (1282) Invalid operation
>
> Problem 2 Detail:
>
> According to Sebastien on this thread:
> http://comments.gmane.org/gmane.comp.lib.vtk.user/70699
>
> We need to create a new a new render window and then attach it using 
> QVTKWidget::SetRenderWindow(...) rather than using 
> QVTKWidget::GetRenderWindow(...)
>
> Even after doing this and setting 'bool doitOffScreen' to false, we 
> still get the error message above.
>
> Similar issues discussed for ATI:
> http://public.kitware.com/pipermail/vtkusers/2013-December/082236.html
>
> Any suggestions appreciated.
>
> Thanks and regards,
> Michael
>
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the VTK FAQ at: http://www.vtk.org/Wiki/VTK_FAQ
>
> Search the list archives at: http://markmail.org/search/?q=vtkusers
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/mailman/listinfo/vtkusers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtkusers/attachments/20150111/0e63ebb1/attachment.html>