I've been working lately with the devs at plotly to optimize their trisurf function in order to make it possible to do interactive 3d plotting in python. Plotly uses WebGL as a backend, so it's generally much smoother and more stable than Mayavi, with the +1 of being able to render in a browser.
With some recent changes to their codebase, I got the plotting time for an fsaverage hemisphere down to about 10 seconds, which isn't too bad. I put together a little proof-of-concept package to do this kind of plotting here:
I was going to turn that demo into a binder so people could live-demo this, but I'm worried about overloading the memory in the binder instance (I think there might be a memory leak in the plotly JS code somewhere). It basically sends all the contents of a trisurf plot over to javascript, so this can slow things down if there are too many files.
So, I was wondering what kinds of memory / CPU / GPU limitations there are in binder, and if it'd be appropriate to do this kind of thing. Alternatively it'd be awesome if somebody could just demo plotting a brain using the same trisurf code in the example, and tell me if that's too heavy to host on a binder.
(also, any suggestions for optimizing things would be welcome. I've been wanting to plot 3d brains in python for a long time :) )
I've been working lately with the devs at
plotlyto optimize their trisurf function in order to make it possible to do interactive 3d plotting in python. Plotly uses WebGL as a backend, so it's generally much smoother and more stable than Mayavi, with the +1 of being able to render in a browser.With some recent changes to their codebase, I got the plotting time for an
fsaveragehemisphere down to about 10 seconds, which isn't too bad. I put together a little proof-of-concept package to do this kind of plotting here:https://github.com/choldgraf/ecogtools/blob/master/examples/3d_plots.ipynb
I was going to turn that demo into a binder so people could live-demo this, but I'm worried about overloading the memory in the binder instance (I think there might be a memory leak in the plotly JS code somewhere). It basically sends all the contents of a trisurf plot over to javascript, so this can slow things down if there are too many files.
So, I was wondering what kinds of memory / CPU / GPU limitations there are in binder, and if it'd be appropriate to do this kind of thing. Alternatively it'd be awesome if somebody could just demo plotting a brain using the same trisurf code in the example, and tell me if that's too heavy to host on a binder.
(also, any suggestions for optimizing things would be welcome. I've been wanting to plot 3d brains in python for a long time :) )