Mostly looking for options here since remote working is pretty much the norm these days. I’m trying to configure a solution where I have Paraview compiled with Mesa (software rendering, and no GPU) on a remote Linux computer. I connect to this machine from Windows via VNC (which doesn’t have OpenGL support), and I launch Paraview (i.e., not client / server based). I’m aware of the VirtualGL approach, but would really prefer this solution instead. I’m not too concerned about performance since the models that I intend to display / interact are fairly small. Has anyone tried this out successfully, and if so, what are the compile / configuration options needed to get this working?
I recommend you build your Linux/server side via the ParaView superbuild project. Enable osmesa and disable Qt. Superbuild has all of the magic configure settings for ParaView, Mesa and the rest of the dependencies preset.
This is how ParaView was built and installed for remote access on a number of clusters and supercomputers to support remote access much as you describe.
Have you tried connecting with NoMachine?