Create RGBD Image from Mesh in Open3D

Question

I'm trying to create a point cloud from a mesh. I haven't had any success. The compiler complains at the line actor = o3d.geometry.PointCloud.create_from_rgbd_image(rgbd, cam) with the following error: [CreatePointCloudFromRGBDImage] Unsupported image format.

When checking rgbd by using print(rgbd) I get the following info:

RGBDImage of size 
Color image : 1920x1061, with 3 channels.
Depth image : 1920x1061, with 1 channels.
Use numpy.asarray to access buffer data.

So it makes me think that the image is valid. Any thoughts on what the issue might be?

import open3D as o3d

def render_depth(cam_intrinsic, model):
    # load model
    actor = o3d.io.read_triangle_mesh(str(model))
    actor.compute_vertex_normals()

    # create visualizer object
    vis = o3d.visualization.Visualizer()
    vis.create_window()
    vis.add_geometry(actor)
    ctr = vis.get_view_control()

    # retrieve intrinsic camera settings
    parameters = o3d.io.read_pinhole_camera_parameters(cam_intrinsic)
    ctr.convert_from_pinhole_camera_parameters(parameters)
    depth = vis.capture_depth_float_buffer(False)
    image = vis.capture_screen_float_buffer(False)
    vis.run()
    vis.destroy_window()

    return depth, image

def main():

    # produce depth and color image, for creating RGBD image
    depth, color = render_depth('directory_to_camera_properties.json', 'model_directory.obj')

    # create RGBD image
    rgbd = o3d.geometry.RGBDImage.create_from_color_and_depth(color=color, depth=depth, 
convert_rgb_to_intensity=False)

    cam = o3d.camera.PinholeCameraIntrinsic()
    cam.intrinsic_matrix = [[500, 0.0, 0.0], [0.0, 918.8529534152894, 0.0], [959.5, 530.0, 1.0]]

    actor = o3d.geometry.PointCloud.create_from_rgbd_image(rgbd, cam)

    o3d.visualization.draw_geometries([actor])

if __name__ == '__main__':
    main()

myowncountry · Accepted Answer · 2023-02-03 11:30:51Z

0

Although I don't know what's in your directory_to_camera_properties.json file, but the vis.run() should come before you capture the depth and image in the render_depth function. Thus it should look like:

def render_depth(cam_intrinsic, model):
    # load model
    # ...

    vis.run()
    depth = vis.capture_depth_float_buffer()
    image = vis.capture_screen_float_buffer()
    vis.destroy_window()

    return depth, image

This resolved the issue for me.

answered Feb 3, 2023 at 11:30

myowncountry

12 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Create RGBD Image from Mesh in Open3D

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related