cuda_image_processing.c:192 about badslam HOT 4 OPEN

This is expected, since the application requires CUDA compute capability 5.3 to run (as documented in the Readme), but the Quadro M6000 seems to support only compute capability 5.2 (according to this page). However, it should be very easy to make the application run with 5.2 as well in case you are willing to change the source code a bit and compile it yourself (unless there are other things requiring a high compute capability that I am not aware of). The reason for requiring it is that point radii are stored as 'half' numbers in a memory buffer to save space, and the corresponding conversion functions require 5.3 according to this table. The alternative would be to store them as floats, which will have no negative consequences except requiring more GPU memory. To change this, one would need to look for all occurrences of the radius_buffer which is stored in the keyframes (keyframe.h header) and change its type from u16 to float. Then drop all of the conversions that happen when reading from or writing to this buffer. For example, in cuda_depth_processing.cu#L355, simply drop the __half_as_ushort(__float2half_rn()) around the value.

from badslam.

Hello @puzzlepaint, I have a follow up question.
I have ran the program on my own data but I am now running into an issue about the scale of my recorded path as it does not match the ground truth.

It is either to way smaller or way larger than than ground truth and I think it may have to do with the parameters labeled:
Maximum depth to use, Baseline*fx, Raw to metric depth conversion factor.

Because I am new to this,I was wondering:

1)what is the meaning of this parameters

2)how do I input them to get the correct path scale (same scale as ground truth).

3)what is meant by Baseline and,

4)which fx should be used for the multiplication with the baseline value(RGB or Depth)?

Thank you very much

from badslam.

Please see here for the parameter explanation:
https://github.com/ETH3D/badslam/blob/master/applications/badslam/doc/command_line.md

Only the raw to metric depth conversion factor should be relevant for scaling.

from badslam.

@puzzlepaint , Thank you very much for you reply. I have read it and I am still getting a path that is smaller than the ground truth. I recorded my data using the intel realsense D435i and calculated that the scale factor should be 1/1000 = 0.001 but the resulting path is sill way smaller than the ground truth.

from badslam.