mesh-to-depth

Depth map generator written from scratch in C++ and Python. CPU-only, self-contained, no external dependencies. Ideal for off-screen online depth data generation from meshes in Python scripts.

Multi-layer depth map implementation comping soon (see 3D Scene Reconstruction with Multi-layer Depth and Epipolar Transformers, ICCV 2019).

Install

Only tested on Python 3.6-3.8 on Ubuntu 18.04.

# Make sure pip is up to date: pip install -U pip
pip install mesh-to-depth

For other python versions and distributions, you may have to compile from source. See this page for supported precompiled binaries.

Example

import numpy as np
import mesh_to_depth as m2d

params = []

params.append({
    'cam_pos': [1, 1, 1], 'cam_lookat': [0, 0, 0], 'cam_up': [0, 1, 0],
    'x_fov': 0.349,  # End-to-end field of view in radians
    'near': 0.1, 'far': 10,
    'height': 240, 'width': 320,
    'is_depth': True,  # If false, output a ray displacement map, i.e. from the mesh surface to the camera center.
})
# Append more camera parameters if you want batch processing.

# Load triangle mesh data. See python/resources/airplane/models/model_normalized.obj
vertices = ...  # An array of shape (num_vertices, 3) and type np.float32.
faces = ...  # An array of shape (num_faces, 3) and type np.uint32.

depth_maps = m2d.mesh2depth(vertices, faces, params, empty_pixel_value=np.nan)

Visualize:

import matplotlib.pyplot as pt

pt.imshow(depth_maps[0], interpolation='none')
pt.colorbar()

See also: python/tests

Test

Manual

Requires pytest.

git clone --recurse-submodules https://github.com/daeyun/mesh_to_depth
cd mesh_to_depth

Then either build the .so files with bash ./cpp/scripts/build_all.sh or do pip install -e .. Then run rests:

cd python
pytest ./tests

Automatic

pip install tox
git clone --recurse-submodules https://github.com/daeyun/mesh_to_depth
cd mesh_to_depth
tox