[vtkusers] Strange Visualization Results With Transparency on Polydata

[Fahlgren, Eric]

That’s excellent news.  This is a very difficult problem, which I’ve watched evolve over many years and it’s great to see the payoff from all those brain cycles that have been applied to it.

Is there any means to manually force render depth for actors/parts?  We have cases where we draw a couple of overlapped X-ray images, transparent so you can see the overlap to align them.  They are overlaid with lines (for example from femur head to trochlear groove), all in the same plane with the images.  Sometimes it looks fine, sometimes the line is hidden until you rotate the image by 1e-6 degrees or zoom by a factor of almost nothing, at which point the lines jump in front; zoom a little more and they go behind.  Draw order is correct, image first, then lines “on top”, but when the image is oriented exactly in the plane of the view, then this issue pops up. (Our hack-around is to put the lines some small increment in front of the plane of the images.)  As I recall, we tried about five years ago to stick something into vtkRenderer to do this as a proof of concept, but never got anywhere near production with it.

From: Ken Martin <ken.martin at kitware.com>
Sent: Tuesday, December 4, 2018 1:24 PM
To: David Gobbi <david.gobbi at gmail.com>
Cc: Andrea Borsic <aborsic at ne-scientific.com>; Fahlgren, Eric <eric.fahlgren at smith-nephew.com>; vtkusers at public.kitware.com
Subject: Re: [vtkusers] Strange Visualization Results With Transparency on Polydata

In newer versions of VTK (8.2 maybe) this issue should not be as noticeable as we changed how we handle transparent rendering to yield better (but still not perfect) results.

On Tue, Dec 4, 2018 at 4:00 PM David Gobbi <david.gobbi at gmail.com<mailto:david.gobbi at gmail.com>> wrote:
What Eric says is true, depth peeling or Z-sorting is needed for transparent polydata.

Here's a long-ish explanation of why the red sphere is rendering wrong: since the mapper uses a lookup table that contains alpha values below 1.0, VTK decides to use the translucent rendering pass instead of the opaque rendering pass.  In the translucent rendering pass, VTK doesn't write to the z-buffer, so unless depth peeling is enabled (or polys are pre-sorted back-to-front) you'll end up with bad results.

  David

On Tue, Dec 4, 2018 at 1:47 PM Fahlgren, Eric <eric.fahlgren at smith-nephew.com<mailto:eric.fahlgren at smith-nephew.com>> wrote:
Hi Andrea,

Z-sorting of transparent polygons is always a problem.  You can cure it by turning on depth peeling, but it comes at a cost (for our models it makes renders take almost twice as long).

# create a rendering window and renderer
ren = vtk.vtkRenderer()
ren.SetUseDepthPeeling(True)

Eric

From: vtkusers <vtkusers-bounces at public.kitware.com<mailto:vtkusers-bounces at public.kitware.com>> On Behalf Of aborsic at ne-scientific.com<mailto:aborsic at ne-scientific.com>
Sent: Tuesday, December 4, 2018 11:20 AM
To: vtkusers at public.kitware.com<mailto:vtkusers at public.kitware.com>
Subject: [vtkusers] Strange Visualization Results With Transparency on Polydata

Dear All,

I am experiencing some strange visualization results in VTK 8.1.0 on polydata using a LUT with some transparency for some entries and opacity for others.

I am attaching a Python example to reproduce the problem. The code creates two polydata spheres which are spatially separated. An array with a uniform values of 1 is associated to cell data for the first sphere, and an array with values of 2 to the second sphere. The two spheres are appended together for convenience with a polydata append filter.

A LUT is created with two entries, solid red RGBA = (1,0,0,1) and transparent green RGBA = (0,1,0,0.2), the scalar range is set to (1,2) mapping the cells of the first sphere to the first LUT entry and the cells of the second sphere to the second LUT entry.

As shown by the attached rendering the red sphere also appears as semi-transparent, and in a wired way, if rotated some portions of the sphere seem more transparent, and others less. Setting the second LUT entry to solid green RGBA = (0,1,0,1) makes both spheres opaque – this is expected, but why does modifying the alpha channel of the second LUT entry affects also the appearance of the first sphere?

I am experiencing this on two different Windows 10 computers equipped with modern NVIDIA GPUs, and on a 2017 iMac 27 running Mojave, all these computers have VTK 8.1.0.

Any comment / advice is welcome,

Thanks, Best Regards,

Andrea
_______________________________________________
Powered by www.kitware.com<http://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:
https://public.kitware.com/mailman/listinfo/vtkusers


--
Ken Martin PhD
Distinguished Engineer
Kitware Inc.
101 East Weaver Street
Carrboro, North Carolina
27510 USA
This communication, including all attachments, contains confidential and legally privileged information, and it is intended only for the use of the addressee.  Access to this email by anyone else is unauthorized. If you are not the intended recipient, any disclosure, copying, distribution or any action taken in reliance on it is prohibited and may be unlawful. If you received this communication in error please notify us immediately and destroy the original message.  Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://public.kitware.com/pipermail/vtkusers/attachments/20181204/12499040/attachment.html>