Open muriatec opened 3 weeks ago
We don't have real-world scales in COLMAP, right?
Because the camera poses are estimated from pixels, we can not know the real-world scales of Gaussians easily. (2D images lose their z-axis)
I think you could try to parse LiDAR data instead.
Hello, thanks for adding the new feature of depth rendering! I'm trying to render depth from Gaussian splats and create point cloud from depth images, and I found the rendered depths are in pixel values (0~255), therefore I'd like to know how to convert from pixel values to real-world units (e.g. meters) given camera intrinsic?