Ah! That is expected 
If you zoom in until you can see the thickness of the elements (silk, copper, etc) you will notice that the thickness is not rendered while you are moving. Less triangles to render == It will make the movement faster
OK, makes sense that for pan/roll you want to get it as fluid as possible, but … the 3d view should be able to render full quality if possible when no speed is needed.
I mean it’s already absolutely stunning, I just wonder if something of this leaks through to the Raytracer engine and causes that one to make jagged edges and holes as well as that’s what takes the quality of those renders down a couple of notches?
OK, #6107 is here, installing and checking if that ones does something different - then off to bed. 
The AA issue is not related with 3D-viewer. I mean, It may be, but not related with the render. I am investigating.
You should try to build it from source. It will be easier after everything is setup.
In some different life maybe… I got like 20 bigger projects in my head and environment currently that need attention and I don’t want to add a 21st
Thinking about this the ‘holes’ must be the bright FR4 material underneath the soldermask shining through in a low-res setting… so one pixel it makes it and the next pixel it doesn’t. If I look at the board in the 3D view from a shallow angle (if I would zoom in I would be able to see the gap between soldermask and FR4) the issue is mostly gone in that area (other bright speckles turn up then between pads and paste and silkscreen).
It’s definitely missing the AA supersampling there.
Probably similar problems for all other layers you got there when you build the board.
And this somehow makes it through to the Raytracing engine.
Btw, #6107 doesn’t show VRML models for me either.
[EDIT] OK, I’m out. Need sleep. 
please have a look if you have enabled SMT and TH 3D models?
These are new options and I don’t know how they are set on first run…
In the new 3D viewer have been added 3 new options for 3D models
As in the pcbnew menu:
Normal is referring to TH
Normal + Insert is referring to SMT
Virtual is referring to models related to i.e. enclosures or mechanical objects useful to show/hide designing
3d viewer Virtual option
Thanks to Mario and Wayne to have added the Virtual option/code that I had in my personal branch in the new branch 
You can also reset all the options to default with a simple click.
An other option could be a problem in environment vars path but it seems to be fine in 6111 that I’m running
Maurice
1 Like
Since this thread has already been more or less hijacked by feedback on Mario’s branch about the new3d-viewer:
Just a heads up, nightlies for Mario’s branch is availabe for OS X now:
http://downloads.kicad.org/osx/testing/kicad_new3d-viewer/
Windows builds on:
http://downloads.kicad.org/windows/testing/kicad_new3d-viewer/
Thank you Nick!
Let me know if anyone tested it on OS X.
This thread was not hijacked, the new 3D viewer is a must for test, as the initial Cirilo’s post, the OCE plugin:
FYI the new 3D viewer has been merged to the product branch:
3 Likes
Hey!
I am just checking in, building the kicad from git now, (using the arch aur package). I was wondering how i use the kicad_oce_3D package? building it seems like no trouble. but where do i put the files?
You should be able to just do: make install DESTDIR=(whatever your kicad install dir)
If you want to copy the file manually, move libs3d_plugin_oce.so to KiCad’s 3D plugin directory; on Linux that’s ${CMAKE_INSTALL_PREFIX}/lib/kicad/plugins/3d. I think the default CMAKE_INSTALL_PREFIX is /usr/local.
I noticed that new 3D viewer create a lot of files *.3dc in %APP_DATA%/Local/kicad/3d under Windows. I used the 3D viewer maybe five times, which created about 30MB of these files. Are these cache files (I think) should not be deleted when you close the window/program?
Cache files will persist to let you a fast loading of the 3d render…
May be a menu option to clean these files on request would be useful to just throw away old cached files and refresh the cache
1 Like
Some method of clearing the cache to avoid stale entry errors is always a good idea
It’s a tricky issue. When you browse for parts now a cache file is created for every model you look at. Deleting after closing the program will cause more computations later to restore the cache files which are needed. For VRML models that’s generally not an issue, but for STEP models that can take a long time. I think maui’s suggestion is best - create some menu item to delete all the cache files.
3 Likes
Why can’t kicad just autoclean it after some time, like when the cahce files are one month old for example?
Then the disk usage would not be monotonically increasing and you not need to take care of it manually.
1 Like
That would be a nice option, but still a manual clean all would be a useful feature
1 Like
Okay so I compiled kicad and I managed to compile kicad_oce_3d, then I moved the libs3d_plugin_oce.so to the /usr/lib/kicad/plugins/3d, next to the ones already there.
I then went to freecad, created a cube and exported to .step. Then I went into a random testproject, added a component and added the exported step file to the 3d settings of the component. I cannot see the 3d model in 3d-viewer.
I also get the error when entering the properties of the schematic component:
/opt/vtk6/lib/libvtkRenderingOpenGL.so.1: undefined symbol: _ZTI23vtkPainterDeviceAdapter
When starting kicad from the command line I get:
13:32:54: Warning: Mismatch between the program and library build versions detected.
The library used 3.0 (wchar_t,compiler with C++ ABI 1009,wx containers,compatible with 2.8),
and your program used 3.0 (wchar_t,compiler with C++ ABI 1010,wx containers,compatible with 2.8).
I am using manjaro-linux and built kicad from the aur package kicad-git, and built the OCE plugin from github by creating my own FindOCE.cmake file
Does kicad itself work? The ABI warning suggests that something was built with different versions of compiler, but it is not clear what (maybe kicad vs wxWidgets).
How was FreeCAD installed and are you using the same OpenCascade or a different one?
The ABI is not the issue. I have that on archlinux, which manjaro are using packages from and I have not seen any issues. It is simply a warning and is not related to the plugin at all. but I don’t understand why @TotalKrill has made his own FindOCE.cmake file. It is a long time since I built it, but I think I just did build it by defining OCE=/opt/oce and OCE_DIR=/opt/oce/lib/oce-0.16 for cmake. I am using the oce package from the AUR.