Downsample vertices when creating vtks

bcipolli commented 9 years ago

Currently, performance isn't great:

When viewing a single hemisphere with the destrieux parcellation causes my rather new macbook pro to load the CPU--enough where the machine gets very hot and runs the fan.
With megabytes of vtk data to download, load time is non-trivial (1-5 seconds or longer).
When embedding these figures in other pages, performance can tank (I tried this with impress.js last night, for example.

It would be great to have simple downsampling of vertices when creating VTKs for visualizations, and to manually test which downsamplings would still look good in the demo.

@binarybottle any suggestions on downsampling schemes? Perhaps sequential elimination: every N vertices, eliminate the last M of them (50% would be N=2, M=1).

bcipolli commented 9 years ago

Eliminating some % of faces was a bad idea lol:

binarybottle commented 9 years ago

Please try mindboggle.guts.mesh.py's decimate() or decimate_file().

bcipolli commented 9 years ago

Fixed in https://github.com/binarybottle/roygbiv/commit/49157b300cfcaf209b639d063706ab419f940baa

binarybottle / roygbiv

Downsample vertices when creating vtks #6