Forum

Author Topic: Using PhotoScan with external calibration?  (Read 3157 times)

maddin

  • Full Member
  • ***
  • Posts: 161
    • View Profile
Using PhotoScan with external calibration?
« on: May 19, 2014, 04:18:30 PM »
We run a multi-camera system for which we already have a pretty decent camera calibration, using multiple checker board images etc. This gives us intrinsic and extrinsic camera parameters which we would like to use in PhotoScan Pro 1.0.4 in order to generate a dense point cloud from our image data.

Since the .psz files can be treated like a .zip file, I thought we could place our calibration data into the doc.xml file. Doing so gives us a camera layout that looks quite reasonable, but now I am struggling getting the rest of PhotoScan to accept that data.

The project has 22 cameras, along with 22 sensor descriptions, derived from our calibration (fx, fy, cx, cy, k1 and k2). I checked 'Fix calibration' for all of them. Now I obviously cannot run 'Align Photos' again, as I do not want the camera positions and orientations to be changed again.

So I need a way to build depth maps from the existing camera data (which requires the sparse Point Cloud first, right?)

Any suggestions would be most welcome!

Thanks
Martin

PS: Is the 1.0.0 documentation for the Python  API the latest?

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 13289
    • View Profile
Re: Using PhotoScan with external calibration?
« Reply #1 on: May 19, 2014, 05:55:22 PM »
Hello Martin,

Wouldn't it be easier to use Export Cameras / Import Cameras option providing that image filenames have the same naming? In case you are using Python, than you need to use .buildPoint() function after matching photos.
Best regards,
Alexey Pasumansky,
Agisoft LLC

maddin

  • Full Member
  • ***
  • Posts: 161
    • View Profile
Re: Using PhotoScan with external calibration?
« Reply #2 on: May 19, 2014, 06:13:59 PM »
Wouldn't it be easier to use Export Cameras / Import Cameras option providing that image filenames have the same naming?

Yes, indeed, thanks for pointing that out. ;-)

Unfortunately, I still get bad results. So this probably means that our calibration parameters do not match those of PS. :-( Is there any more information available on the meaning and units of the different camera parameters?

Also, I do not get any depth maps after importing cameras. When running 'Build dense cloud', I get an error message: "Can't build dense cloud" and this output on the console:

Code: [Select]
Using device: GeForce GTX 780, 12 compute units, 3072 MB global memory
  max work group size 1024
  max work item sizes [1024, 1024, 64]
  max mem alloc size 768 MB
initializing...
selected 0 cameras from 7 in 0.001 sec
Loading photos...
Reconstructing depth...
finished depth reconstruction in 0 seconds
Device 1 performance: 0 million samples/sec (CPU)
Device 2 performance: 0 million samples/sec (GeForce GTX 780)
Generating dense point cloud...
selected 0 cameras in 0 sec
Finished processing in 0.107 sec (exit code 0)