Blender render artifacts

wilfchun · April 20, 2023, 06:39

Hello,
I am trying to come up with a workflow for rendering reef3D outputs in blender.
An example of where I am it is here: https://youtu.be/PWSFtRq_I_g

There appears to this griding along the domain that is interfering with the visual. Is there a way to remove this either pre or post simulation?

Felix_Sp · April 21, 2023, 03:17

Hey

I played a bit with the rendering as well. What you see are the processor boundaries of the simulation. You also get these articfacts when you decrease the opacity of the .pvtu/.vtu in ParaView.

My suggestion to solve this problem is to apply a "D3" Filter on the .pvtu in ParaView (Just have a look in Filters->Alphabetical in ParaView). Depending on how you import your files into Blender you can then either export these Files directly by File->save Data ->"give name"->"put a tick into write time steps"->OK.

If you want an .stl/obj. (whatever surface data) for Blender you have to put the filter "Extract Surface" onto the "D3" filter in ParaView and then File->save Data ->"give name"->"put a tick into write time steps"->OK. I would be interested in the result of your render if you are willing to share your video

.

Hope that helps. Cheers

wilfchun · April 21, 2023, 05:42

I have tried the D3 filter and also what appears to be the newer Ghost cells generator filter, but this still leaves the vertical lines both in preview renderview (with decreased opacity) and then in the export to blender.

I am actually using Nvida Omniverse to move the scenes between paraview and blender using the .usd filetype

Felix_Sp · April 21, 2023, 07:23

This is weird. I never had a problem with the D3 Filter showing the processor lines. The ghost cells generator filter does not remove the processor boundaries as well. Do you have the MPI version or the "normal" version of ParaView?
A colleague of mine had the problem, that the filters of the MPI version and the "normal" version did work differently. I use the MPI version on an Ubuntu 22.04 Maybe try the d3 filter of the MPI version? https://www.paraview.org/download/

valgrinda · April 21, 2023, 11:04

Wow, very interesting! I always wanted to do something like this😉. Do you have plans to make your routine publicly available? That would be really great!

Regarding your problem: If nothing else fixes this, then one option could be to gather the individual .vtu files and merge them into on single global file in a post-processing step. We do something similar for our state file when using coupling. This could be integrated into DIVEMesh.

wilfchun · April 26, 2023, 08:58

Using the MPI version of paraview and the D3 filter fixed these artifacts.
The final render can be seen here: https://youtu.be/TzuywpzJqfE

Felix_Sp · April 26, 2023, 09:07

Cool that you got it running. It still buggers me that the MPI and "normal" version of ParaView work differently.

Did you use the Stop-Motion Add-On for Blender to import the time dependent surface data? I guess afterwards, you just applied some light and material properties for the water and rendered the Video/ single pictures?

Cheers

wilfchun · April 26, 2023, 09:32

I actualy used nvidia Omniverse. This installs a plugin to paraview and then saves the animation from the viewport as a .usd file, which can then be imported to blender.

And, yeah, then just a HDRI for environmental lighting and a basic water material.

valgrinda · April 27, 2023, 04:01

Wow, looks really cool! Do you think you could write up a quick step-by-step recipe/manual on how to do this? That would be really helpful!

Felix_Sp · April 27, 2023, 05:11

I do not know about Mr. Chun, but I work basically as explained in this video by the DualSPH guys. Also, the VisualSPHysics-AddOn has a pretty good water-material.

Besides the input working differently (as we somehow have to import the data to Blender), the workflow is pretty comparable. I use the Stop-Motion-addon for blender to import the surface-files of ParaView. I create the surface-files by clipping the water surface of the REEF3D-simulation, and applying an "extract surface"-filter. Furthermore, I extract this surface as .stl by file->save data->write all time steps. These surface files can then be imported via Stop-Motion in Blender. Afterwards I render as explained in the video.

Are you doing any different Mr. Chun? The workflow is actually quite easy once one understands the basics of Blender.

VisualSPHysics actually has a nice foam-simulator based on this paper. If one had a python filter for ParaView doing the same for REEF3D the simulations would actually look even more awesome.

April 20, 2023, 06:39	Blender render artifacts	#1
wilfchun New Member Wilf Chun Join Date: Apr 2023 Posts: 4 Rep Power: 3	Hello, I am trying to come up with a workflow for rendering reef3D outputs in blender. An example of where I am it is here: https://youtu.be/PWSFtRq_I_g There appears to this griding along the domain that is interfering with the visual. Is there a way to remove this either pre or post simulation?

April 21, 2023, 11:04		#5
valgrinda Super Moderator Hans Bihs Join Date: Jun 2009 Location: Trondheim, Norway Posts: 403 Rep Power: 19	Wow, very interesting! I always wanted to do something like this😉. Do you have plans to make your routine publicly available? That would be really great! Regarding your problem: If nothing else fixes this, then one option could be to gather the individual .vtu files and merge them into on single global file in a post-processing step. We do something similar for our state file when using coupling. This could be integrated into DIVEMesh. __________________ Hans Bihs Team REEF3D www.reef3d.com

April 26, 2023, 08:58	Fixed	#6
wilfchun New Member Wilf Chun Join Date: Apr 2023 Posts: 4 Rep Power: 3	Using the MPI version of paraview and the D3 filter fixed these artifacts. The final render can be seen here: https://youtu.be/TzuywpzJqfE

April 27, 2023, 04:01		#9
valgrinda Super Moderator Hans Bihs Join Date: Jun 2009 Location: Trondheim, Norway Posts: 403 Rep Power: 19	Wow, looks really cool! Do you think you could write up a quick step-by-step recipe/manual on how to do this? That would be really helpful! __________________ Hans Bihs Team REEF3D www.reef3d.com

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
SnappyHexMesh GUI add-on for Blender	tkeskita	OpenFOAM Pre-Processing	22	April 22, 2024 12:45
[Other] BVTKNodes - photorealistic rendering of OpenFOAM results in Blender	tkeskita	OpenFOAM Community Contributions	5	October 15, 2021 08:55
[Other] How to get 3D model from Blender with different Image views?	sunagra27	OpenFOAM Meshing & Mesh Conversion	0	June 1, 2020 03:53
FYI: SwiftSnap now can also export on Blender to enGrid	wyldckat	enGrid	0	October 6, 2013 11:35
[blockMesh] Tool Block generation with blender	wikstrom	OpenFOAM Meshing & Mesh Conversion	8	August 12, 2009 18:09

April 21, 2023, 03:17		#2
Felix_Sp Member Felix S. Join Date: Feb 2021 Location: Germany, Braunschweig Posts: 88 Rep Power: 6	Hey I played a bit with the rendering as well. What you see are the processor boundaries of the simulation. You also get these articfacts when you decrease the opacity of the .pvtu/.vtu in ParaView. My suggestion to solve this problem is to apply a "D3" Filter on the .pvtu in ParaView (Just have a look in Filters->Alphabetical in ParaView). Depending on how you import your files into Blender you can then either export these Files directly by File->save Data ->"give name"->"put a tick into write time steps"->OK. If you want an .stl/obj. (whatever surface data) for Blender you have to put the filter "Extract Surface" onto the "D3" filter in ParaView and then File->save Data ->"give name"->"put a tick into write time steps"->OK. I would be interested in the result of your render if you are willing to share your video . Hope that helps. Cheers

April 21, 2023, 05:42		#3
wilfchun New Member Wilf Chun Join Date: Apr 2023 Posts: 4 Rep Power: 3	I have tried the D3 filter and also what appears to be the newer Ghost cells generator filter, but this still leaves the vertical lines both in preview renderview (with decreased opacity) and then in the export to blender. I am actually using Nvida Omniverse to move the scenes between paraview and blender using the .usd filetype

April 21, 2023, 07:23		#4
Felix_Sp Member Felix S. Join Date: Feb 2021 Location: Germany, Braunschweig Posts: 88 Rep Power: 6	This is weird. I never had a problem with the D3 Filter showing the processor lines. The ghost cells generator filter does not remove the processor boundaries as well. Do you have the MPI version or the "normal" version of ParaView? A colleague of mine had the problem, that the filters of the MPI version and the "normal" version did work differently. I use the MPI version on an Ubuntu 22.04 Maybe try the d3 filter of the MPI version? https://www.paraview.org/download/

April 26, 2023, 09:07		#7
Felix_Sp Member Felix S. Join Date: Feb 2021 Location: Germany, Braunschweig Posts: 88 Rep Power: 6	Cool that you got it running. It still buggers me that the MPI and "normal" version of ParaView work differently. Did you use the Stop-Motion Add-On for Blender to import the time dependent surface data? I guess afterwards, you just applied some light and material properties for the water and rendered the Video/ single pictures? Cheers

April 26, 2023, 09:32		#8
wilfchun New Member Wilf Chun Join Date: Apr 2023 Posts: 4 Rep Power: 3	I actualy used nvidia Omniverse. This installs a plugin to paraview and then saves the animation from the viewport as a .usd file, which can then be imported to blender. And, yeah, then just a HDRI for environmental lighting and a basic water material.

April 27, 2023, 05:11		#10
Felix_Sp Member Felix S. Join Date: Feb 2021 Location: Germany, Braunschweig Posts: 88 Rep Power: 6	I do not know about Mr. Chun, but I work basically as explained in this video by the DualSPH guys. Also, the VisualSPHysics-AddOn has a pretty good water-material. Besides the input working differently (as we somehow have to import the data to Blender), the workflow is pretty comparable. I use the Stop-Motion-addon for blender to import the surface-files of ParaView. I create the surface-files by clipping the water surface of the REEF3D-simulation, and applying an "extract surface"-filter. Furthermore, I extract this surface as .stl by file->save data->write all time steps. These surface files can then be imported via Stop-Motion in Blender. Afterwards I render as explained in the video. Are you doing any different Mr. Chun? The workflow is actually quite easy once one understands the basics of Blender. VisualSPHysics actually has a nice foam-simulator based on this paper. If one had a python filter for ParaView doing the same for REEF3D the simulations would actually look even more awesome.