[Paraview] [Non-DoD Source] Eye separation has been wrong since Immersive ParaView 5.2 - here's how to fix it

Mon Jun 26 11:42:05 EDT 2017

Hello,

I am the one who requested the changes,
they were made to correct the stereo separation with working with zSpace
hardware
for data that was small in size (floating point values of coordinates, >> 1,
for example data that ranged from 0.0000000 to  0.0000001
I checked its use with data across 13 orders of magnitude from data of size
1,000,000 to data of scale 1/1,000,000.
I admit my fix is an approximation, but I am attempting to get a reasonable
agreement of size with  the non stereo operation

  double eyeSeparationCorrectionFactor = 10.0;
  double shiftDistance = this->EyeSeparation / (2.0 *
eyeSeparationCorrectionFactor);
  if(this->Distance < 1.0)
  {
    shiftDistance *= this->Distance;
  }

  // Front (aka near)
  double F = E[2] - (this->Distance + this->Thickness);//E[2] -
10000.0;//this->ClippingRange[1];
  // Back (aka far)
  double nearDistanceCorrectionFactor  = 1000.0;
  double B = E[2] - (this->Distance / nearDistanceCorrectionFactor);//E[2]
- .1;//this->ClippingRange[0];

Without these changes with small data the stereo separation is incorrect,
(too large)
also small data will not even show up in the render window as the clipping
planes will be wrong.

This change

"  if (cameraParallel == 0 && dot(normalVCVSOutput,vertexVC.xyz) > 0.0) {
normalVCVSOutput = -1.0*normalVCVSOutput; }"

I did not make and do not know its purpose.

I think we  need a solution that works for both caves and planar display
hardware,
and I think the underlying issue is that there is different render code
being used,
that is a subset of the full normal rendering solution, which is creating
these issues.
I think the correct solution is to remove  the separate render code and use
the primary
code for both client and server rendering of mono and stereo. I do not have
a cave to
test with, but I do not see why we can not find a solution that works for
everyone.

Thanks,

Joe

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Joseph G. Hennessey Ph.D., SAIC
Team SAIC
Army Research Lab
DOD Supercomputing Resource Center
Aberdeen Proving Ground, MD 21005

-----Original Message-----
From: ParaView [mailto:paraview-bounces at paraview.org] On Behalf Of Stock,
Mark
Sent: Friday, June 23, 2017 5:23 PM
To: Gena Bug via ParaView <paraview at paraview.org>
Subject: [Non-DoD Source] [Paraview] Eye separation has been wrong since
Immersive ParaView 5.2 - here's how to fix it

All active links contained in this email were disabled.  Please verify the
identity of the sender, and confirm the authenticity of all links contained
within the message prior to copying and pasting the address to a Web
browser.  

----

Hello,

If you're an Immersive ParaView user, you may have noticed that 5.x versions
don't work as well in your CAVE. Specifically, I've noticed the following
differences when compared with ParaView 4.4.0:

1) I can place an object in space properly but it seems to move 10x as fast
when I move my head.
2) If I move the object just a little farther away from its initial
position, it clips out (vanishes).
3) When I am viewing a triangle mesh, some triangles seem to flip normals
(and get dark) when I move the object around.

The first two problems are caused by a change in the VTK source code from
March 2016 which arbitrarily multiples the eye separation distance by 10,
and also tightens the near and far clipping planes. The third can be
corrected by adding a single comment (which will help CAVE and Immersive
users, but probably nobody else). Both fixes will require you to build
ParaView from source - which you're probably doing anyway. I can't comment
on how these changes will affect HMD VR users.

The first and second problems are caused by a single commit to the VTK
repository that began to appear in ParaView with version 5.2.0. The problem
code is in VTK/Rendering/Core/vtkCamera.cxx starting at line 452 in versions
5.3 and 5.4. This problem can be fixed by replacing

  double eyeSeparationCorrectionFactor = 10.0;
  double shiftDistance = this->EyeSeparation / (2.0 *
eyeSeparationCorrectionFactor);
  if(this->Distance < 1.0)
  {
    shiftDistance *= this->Distance;
  }

with simply

  double shiftDistance = this->EyeSeparation;

The second problem (clipping planes) is just a few lines down. I fixed it by
replacing

  // Front (aka near)
  double F = E[2] - (this->Distance + this->Thickness);//E[2] -
10000.0;//this->ClippingRange[1];
  // Back (aka far)
  double nearDistanceCorrectionFactor  = 1000.0;
  double B = E[2] - (this->Distance / nearDistanceCorrectionFactor);//E[2]
- .1;//this->ClippingRange[0];

with

  // Front (aka near)
  double F = E[2] - 10000.0;//this->ClippingRange[1];
  // Back (aka far)
  double B = E[2] - .1;//this->ClippingRange[0];

The third problem can be fixed by commenting out this line in
VTK/Rendering/OpenGL2/vtkOpenGLPolyDataMapper.cxx (around line 1198 in 5.1,
1416 in 5.2 to 5.4):

"  if (cameraParallel == 0 && dot(normalVCVSOutput,vertexVC.xyz) > 0.0) {
normalVCVSOutput = -1.0*normalVCVSOutput; }"

I hope these fixes help some of you out there get your CAVE updated to the
latest ParaView.

Mark

Mark.Stock at nrel.gov

_______________________________________________
Powered by Caution-www.kitware.com

Visit other Kitware open-source projects at
Caution-http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at:
Caution-http://paraview.org/Wiki/ParaView

Search the list archives at: Caution-http://markmail.org/search/?q=ParaView

Follow this link to subscribe/unsubscribe:
Caution-http://public.kitware.com/mailman/listinfo/paraview
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5615 bytes
Desc: not available
URL: <http://public.kitware.com/pipermail/paraview/attachments/20170626/9aa57927/attachment.bin>