Lens Calibration

Answered

Follow

Leif

• October 20, 2016 10:14

Hi,

is it possible and recommended to import lens undistortion data for the Brown 3 model.
We have checkerboard lens grids pictures for the lens.
We have used *********** in the past to get values for the distortion model.
Are these values compatible to the Brown 3 model.
First tests here show that something does not quite seem to match.

Is there any recommended workflow to handle lens undistortion without relying on the automatic feature of RC?
We would rather rely on proper checkerboard lens grids.

Cheers,

Leif

0

• 

  martinb

  • October 20, 2016 11:56

  Deer Leif,
  
  RC uses lens distortion coefficients with respect to the unit size image.
  
  image coordinate = camera calibration matrix * distortion ( model, X in camera coordinate system )
  
  Therefore depending on a tool which you are using you should first recalculate coefficients to the unit size image.
  
  

  Is there any recommended workflow to handle lens undistortion without relying on the automatic feature of RC?

  
  
  You can switch lens/calibration mode to fixed and input your values. You can do this using XMP files automatically.
  The easiest way to do that is to align few images, export XML (using export metadata) and to edit XMP files.
  RC automatically loads XMP files with the same name as loaded image files, e.g. image01.jpg and image01.xmp.
  
  Another option, if your camera / lens model is somewhat distinguished, is to edit sensors xml file and add entry which automatically assigns lens/calibration parameters for all images which contain this camera/lens model.
  
  

  Code:

  We would rather rely on proper checkerboard lens grids.

  
  This is always question. If you stay in lab conditions then yes. If your camera is high res then pre-calibration before moving the camera is risky and calibration will change.

  0

  Comment actions Permalink
• 

  Leif

  • October 20, 2016 16:10

  Hi MartinB,
  
  thank you for your answer.
  We are using a Canon 5dR S, so pretty highres
  If I get it right you are saying the automatic calibration of RC is probably better than using a precalibration?
  So just leaving it to solve automatically every time would be the recommended way?
  Did you get the best results with this approach or is a calibration recommended?
  
  Cheers,
  
  Leif

  0

  Comment actions Permalink
• 

  Annukka Debenjak

  • December 17, 2019 15:13

  Hi MartinB
  
  I'm wondering about the same and would be thankful for a recommendation. I'm aiming to create very accurate models so calibrating the camera and lens seems right, but I would like to hear if it's actually not improving the accuracy?

  Br,

  Annukka

  0

  Comment actions Permalink

Please sign in to leave a comment.