Last active
September 29, 2021 09:11
-
-
Save shinshiner/c91e83270ae992593d5732e117de64a8 to your computer and use it in GitHub Desktop.
Convert depth map to point cloud in Mujoco 1.50
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import numpy as np | |
def depth2pcd(depth): | |
def remap(x, in_range_l, in_range_r, out_range_l, out_range_r): | |
return (x - in_range_l) / (in_range_r - in_range_l) * (out_range_r - out_range_l) + out_range_l | |
# depth = remap(depth, depth.min(), depth.max(), 0, 1) | |
# print(depth) | |
scalingFactor = 1 | |
fovy = 60 | |
aspect = depth.shape[1] / depth.shape[0] | |
# fovx = 2 * math.atan(math.tan(fovy * 0.5 * math.pi / 360) * aspect) | |
width = depth.shape[1] | |
height = depth.shape[0] | |
fovx = 2 * math.atan(width * 0.5 / (height * 0.5 / math.tan(fovy * math.pi / 360 / 2))) / math.pi * 360 | |
fx = width / 2 / (math.tan(fovx * math.pi / 360 / 2)) | |
fy = height / 2 / (math.tan(fovy * math.pi / 360 / 2)) | |
points = [] | |
for v in range(0, height, 10): | |
for u in range(0, width, 10): | |
Z = depth[v][u] / scalingFactor | |
if Z == 0: | |
continue | |
X = (u - width / 2) * Z / fx | |
Y = (v - height / 2) * Z / fy | |
points.append([X, Y, Z]) | |
return np.array(points) |
Mayi ask what need to be the input (depth) shape ? and How to make depth from rgba or rgb image?
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Thanks for this function. It's really helpful. But I think Line 16 should have "fx = width / 2 / (math.tan(fovx * math.pi / 360 ))" and so should be Line 17. Please confirm.