I do a lot of work with lidar and meshing. Usually the process goes like this:
- Register the scans together
- Export out a very large point cloud
- Mesh the point cloud at the minimum resolution possible while keeping needed details
- Bring that mesh into a 3d package
- Build a new low res version using the high res as a guide
The issue with this process is dealing with such a high poly count, the alembic for this mesh was around 4.6gb, here is the wireframe on shaded view:
The way I usually quickly retopologise things like this is with Maya’s modelling toolkit, mainly the ‘quad draw’ tool which allows you to make a mesh ‘live’ and then draw vertices on that mesh and basically connect the dots to make polygons. Once you get used to using it its really fast!
But one thing you’ll find when you bring geo like this into Maya is that using it as a ‘live’ mesh is not fun! Snapping things to this mesh can take 20-30 seconds, so drawing vertices on it, where maya pretty much snapping to the mesh, is pointless!
So the solution is to export this lidar mesh as a GPU cache, what this will do is remove this mesh out of your scene and then load it into the GPU RAM. Now not everyone can do this as not every machine will have dedicated graphics memory, luckily my work machine has a Titan X with 12gb dedicated RAM so this wasn’t a problem.
Its simple to do, here are the steps:
- Go to Cache > GPU Cache > Export selection and export out your mesh (if its a static mesh then just use current frame)
- Once its exported delete the mesh out of your scene
- Go back to the GPU Cache menu and import it back in
This will now load a single item into your outliner with a green thunderbolt symbol. You wont see anything in your viewport, yet! Give it some time as maya is now loading that mesh onto your GPU. Once its done it will displace everything and you can navigate and use it a whole lot faster than before!