Skip to content

Latest commit

 

History

History
159 lines (105 loc) · 4.94 KB

README.md

File metadata and controls

159 lines (105 loc) · 4.94 KB
Raw

Colorado

Plot point-clouds, volumes and meshes in Python Notebooks.

Content:

Introduction

Install

Get Started

Introduction

Colorado plots 3D data, such as point-clouds, volumes and meshes. It uses plotly, hence can be easily integrated in a Jupyter Notebook. Colordao's draw function identifies the data type/format and choses the best representation for you.

Colorado can draw numpy arrays, or any custom object.

import colorado as cld
from soma import aims
meshR = aims.read('../docs/tutorial/data/subject01_Rhemi.mesh')
meshL = aims.read('../docs/tutorial/data/subject01_Lhemi.mesh')
cld.draw([meshL, meshR])

image

The colorado package is simply glue between numpy and plotly, but it also allows to plot aims objects (buckets volumes and meshes) in a Brainvisa environment.

Install

  1. clone this repo

  2. install the module:

$ cd colorado
$ pip install .

The following steps are necessary to plot in Jupyter:

  1. install node and npm (takes minutes). Run the instal_node.sh script:
$ ./install_node.sh
  1. Install the plotly plug-in in Jupyter notebook or Jupyter-lab (instructions in the links).

Install with Brainvisa

You might want to install colorado in an environment that already has pyAims to draw aims entites such as buckets and volumes. To do so, isntall colorado from a Brainvisa shell. In Brainvisa ()>=5), you can enter a new shell with the command bv bash.

NOTE on running pip in bv bash: To make sure you are installing python modules in the right python environment (namely that of Brainvisa) use python3 -m pip instead of pip

Resources

Get started

Colorado' draw function choses the best representation according to the type of the first calling argument. The object that can be drawn are:

  • numpy arrays
  • aims buckets volumes and meshes
  • any objects that implements the specific __draw_with_colorado__ method (returning one of the above drawable objects or a plotly graphic object)

The draw() function can plot numpy arrays, in this case the type of object is inferred by the array dimensions

Point-clouds

arrays of shape = (N,3) are interpreted as point-clouds (i.e. lists of coordinates) and plotted as scatter plots

import numpy as np

# plot a point cloud
pc = np.load("../docs/tutorial/data/numpy_pc.npy")
cld.draw(pc)

image

Volumes

Arrays of shape = (L,N,M) are interpreted as 3D volumetric images. The positive valued voxels are displayed in a scatter plot.

vol = aims.read('../docs/tutorial/data/subject01.nii')
cld.draw(vol, downsample=2, max_points=5000, th_min=950, th_max=None)
# - downsample        downsample the voxels in the volume
# - max_points        number of randomly sampled points to plot (low => fast)
# - th_min            voxels below this value will not be plotted
# - th_max            voxels above this value will not be plotted

image

Meshes

eshes are defined by sets of vertices and polygons. colorado has a handy SimpleMesh class that can be used to draw meshes

vertices = np.load('../docs/tutorial/data/mesh_vertices.npy')
polygons = np.load('../docs/tutorial/data/mesh_polygons.npy')

mesh = cld.drawables.SimpleMesh(vertices, polygons)
cld.draw(mesh)

image

multi-plot

iterables and dictionnaries can also be plot by the draw function

cld.draw([pc,mesh])

image

the draw function returns a plotly window, that can be reused after as an argument to incrementally add entites to the plot

fig = cld.draw(pc)
cld.draw(mesh, fig = fig, name = "mesh")

image

Plotting options and Makeup

Any optional argument passed to draw() which is not defined in the prototype, is directly passed to the underlying Plotly function. For example the marker argument can be used to change size, shape and color of the points in a bucket plot. See "3D Scatter Plot" in Plotly's documentation for a complete definition of all available options.

cld.draw(a, marker=dict(color='red'))

image