Skip to content

Latest commit

 

History

History
106 lines (73 loc) · 3.79 KB

README.md

File metadata and controls

106 lines (73 loc) · 3.79 KB

blender-kitti

License: MIT

KITTI Point Cloud KITTI Point Cloud
KITTI Point Cloud KITTI Point Cloud
KITTI Scene Flow KITTI Scene Flow

About

This repository contains some code to create large particle collections (voxel grids, point clouds) together with color information in Blender. blender-kitti has two goals in mind:

  • The created objects are exact, meaning that all particles are created at their defined location and all colors have the exact RGB-value as specified. All particles can be colored individually.
  • Performance of the scrips is acceptable. It should not take much longer than a second to create a 100k point cloud.

Together, these qualities enable blender-kitti to render large scale data from the KITTI dataset (hence the name) or related datasets.

When it comes to visualization, everyone has a different usecase. So this is not a one-fits-all solution but rather a collection of techniques that can be adapted to individual usecases.

There is example code that renders the demo images above. Use this to verify that your installation works and as a starting point for your modifications.

Installation into Blender's bundled python

# Wherever your Blender installation is located. E.g. cd /opt/blender-2.90.1-linux64/2.90/python
$ cd <blender_directory>/<blender_version>/python

# make pip available
$ ./bin/python3.Xm lib/python3.X/ensurepip

# install
$ ./bin/pip3 install -e <path_to_blender_kitti>

Render demo images

Render the bundled KITTI point cloud with semantic coloring from two different camera perspectives. This writes two image files to the /tmp folder.

$ blender --background --python-console

>>> import blender_kitti_examples
>>> blender_kitti_examples.render_kitti_point_cloud()

Render the bundled Semantic KITTI voxel grid as top view and close-up image. This writes two image files to the /tmp folder.

$ blender --background --python-console

>>> import blender_kitti_examples
>>> blender_kitti_examples.render_kitti_voxels()

Render the bundled KITTI point cloud with pseudo odometry for hsv colored scene flow from two different camera perspectives. This writes two image files to the /tmp folder.

$ blender --background --python-console

>>> import blender_kitti_examples
>>> blender_kitti_examples.render_kitti_scene_flow()

Work on a scene in Blender

You can import and use blender-kitti in the python console window in the Blender-GUI itself to work on a given scene.

# Create a random [Nx3] numpy array and add as point cloud to a scene in blender.
import bpy
import numpy as np
from blender_kitti import add_point_cloud

# create some points
points = np.random.normal(loc=0.0, scale=2.0, size=(100, 3))

# get current scene
scene = bpy.context.scene

# create point cloud object and link to scene
add_point_cloud(points=points, scene=scene, particle_radius=0.2)

Result:

KITTI Point Cloud

Ideas for future development

  • Track all created objects/meshes/images and be able to completely remove them later
  • Handle name clashes or overwrite existing objects
  • Define the rotation/scale of individual particles
  • Create a useful, small API
  • Finally be able to build Blender-bpy as module (Dockerfile)