Hi Ken - Wow - that's awesome! Thanks very much! If I build ParaView from
the master branch of GitLab, will I get these changes? If so, I can try it
and let you know what I see. Thanks again! -- Adam
___
Powered by www.kitware.com
Visit other Kitware
Adam,
You should be able to use the latest 5.2-RC1 binaries to test Ken's
performance improvements out.
Utkarsh
On Tue, Oct 11, 2016 at 10:58 AM, Adam Lyon wrote:
> Hi Ken - Wow - that's awesome! Thanks very much! If I build ParaView from
> the master branch of GitLab, will I
I finally did get around to this issue and have completely reworked the
CompositePolyDataMapper2 for the OpenGL2 backend. Now there is only the
fast path and your original dataset ***should*** render quickly with
OpenGL2. These changes were merged into VTK last week and hopefully most of
the kinks
Hi - Just thought I'd let you all know that I've finally made progress on
this problem. To recap, I have a MultiBlockDataSet (MBDS) containing
PolyData objects. The PolyData objects don't all have the same attributes
(some have Normals, some don't; some have particular FieldData, others
don't).
Nice job Adam!
I've though about your issue some more and I think the best solution is to
revamp the vtkCompositePolyDataMapper2 to work a little differently.
Basically always have it group together blocks that have the same
properties and then render each of them with a helper mapper. That would
This code is all new hence the change. On the plus side if you moved from
4000 blocks of 10 triangles each to 4000 blocks of 1 triangles each the
rendering performance would probably not slow down. The overhead is all in
starting and stopping each of those 4000 mappers.
The old backend wrote
Hi Ken and Utkarsh, Checking back in with this issue... I still find that
interaction with the g-2 ring in ParaView v5rc2 is very slow compared to
v4.4. It looks like you all have a solution in mind. I'm happy to build
from master to try it if the solution is there. Thanks again and Happy
Hi Ken - thanks for your reply. It's looking like the last approach - I
figure out why I'm getting different blocks - is the way to go. It's
already clear that not all of the blocks have colors. Some of the blocks
are created by vtk sources (e.g. VtkConeSource) and others are created by
hand (me
I don't really have a solution planned for that case. The data size is
trivially small but it falls into the slow path because each block of data
may or may not have normals or colors or tcoords. The nature of the blocks
is heterogeneous. So you end up with 4000 mappers each rendering 10
On Wed, Dec 16, 2015 at 13:22:09 -0600, Adam Lyon wrote:
> I'm curious as to why I didn't have any of these problems in v4.4 -- or is
> this two path multiblock rendering code all new for v5? Thanks again!! --
4.4 defaults to the OpenGL1 backend (glBegin/glEnd); 5.0 is using the
OpenGL2 backend
You could maybe have an actor per LOD and then swap actors in and out.
Actor owns the property, backface property, and texture. Changes to those
objects can cause VBO/IBO rebuilds (switching from surface to wireframe,
adding a texture map, flat versus phong shading, edge visibility, line
width,
> I did notice PV is forcing the actor to be modified every time the LOD
> changes (which is a lot). Modifying the actor causes the VBO/IBOs to rebuild
> so it would be better not to do that unless you really need to and in this
> case I'm not sure why it is being done. Specifically I think it is
I took a quick look at it. I see everything from 16 to 111 fps on my laptop
which is a pretty wild range of numbers. I'll do a release build with
OpenGL1 to see what that shows but it will take a while. This dataset is
using the slow path because it has some blocks that have normals and some
that
Yes, I am noticing that slowness too. Ken, mind taking a look at that?
Utkarsh
> On Nov 19, 2015, at 12:38 AM, Adam Lyon wrote:
>
> Indeed you are right! I see the rendering you show for 5.0. Interesting
> change with regard to setting the solid color first.
>
> One thing
Indeed you are right! I see the rendering you show for 5.0. Interesting
change with regard to setting the solid color first.
One thing I should note is that image manipulation (pan, zoom, rotate) is
quite slow in 5.0 compared to 4.4. For example, rotating the ring with a
solid color, I get
Yes, thank goodness, I thought this dataset looked familiar. Now I remember
looking into this a while back and yes, the decision we landed on was as
you specified. The new code is different from the old version, but to me a
much safer and better approach. If the data is missing for a block we use
Ken, what Adam has is a multiblock dataset with field data array named
"Color" (with certain blocks missing this array). Now when you color by
this array and then uncheck "Map Scalars", I get the attached images. I'd
contend that 5.0 with OpenGL2 rendering is correct. In 4.4 it would just
bleed
Hi Ken - Thanks for your e-mail. If you look at the first message in this
thread with the pictures, you'll see that the OpenGL2 image doesn't look
like the OpenGL1 image. The OpenGL1 image is correct. The OpenGL2 image
should be identical if things are working. That message has a link to my
I'd be happy to address this if someone can explain to me what the problem
is. A failing VTK test would be really nice as I'm a bit lost from the
email. I just don't understand the specifics of what is going on here. It
sounds like some sort of CompositePolyData2 with a field data where the
field
Hi - I know you all are busy preparing for SC15, but this bug persists in
the ParaView v5 release candidate, which makes it difficult for me to use.
Is there a chance this problem can be looked at and resolved before the
final release? Thanks! Happy SC15!! -- A
*--*
*Adam L. Lyon*
Awesome! Thanks -- Adam
On Friday, August 7, 2015, Utkarsh Ayachit utkarsh.ayac...@kitware.com
wrote:
Adam,
The problem is that the OpenGL2 implementation doesn't work correctly for
field-data arrays yet.
I've reported a bug: http://www.paraview.org/Bug/view.php?id=15627
We'll have in
Hi Joachim, Do you want me to send in a bug report for these problems? I
assume there's no work-around that I can do to make things work. Using
OpenGL2 is quite a bit faster, so I'm hoping a fix can come out soon.
Thanks! -- Adam
*--*
*Adam L. Lyon*
*Scientist; Associate Division Head for
Hi Joachim - Yes, I noticed that the default color is now vtkBlockColors.
However I changed that to the color vector in my data and that change was
not honored by the program. Thanks, Adam
*--*
*Adam L. Lyon*
*Scientist; Associate Division Head for Systems for Scientific Applications*
Adam,
The color difference can be explained by the default coloring mode
introduced recently (before the git tag version you tested with).
Multiblocks are now automatically colored by blocks. Toggle the Color
Legend and see the coloring array, vtkBlockColors is selected wereas in
previous version
Hi, I'm trying to use ParaView built with OpenGL2, hoping to see some
speed improvement when I interact with my visualization. Indeed for a
complicated geometry I see significant speed up (e.g. ~25 fps with OpenGL2
vs. 5 fps with regular ParaView). But I also see serious errors in the
display.
25 matches
Mail list logo