Forum

Author Topic: Wrong roll estimation in 1.3  (Read 1581 times)

JMR

  • Sr. Member
  • ****
  • Posts: 442
    • View Profile
Wrong roll estimation in 1.3
« on: February 06, 2017, 10:25:11 PM »
Hi, Alexey:
There must be a bug in the roll estimation because in a project composed of 790photos hand held, 545 show 0.000 as estimated roll value... others show values close to -180 or to 180 but they alternate with no reason. Apparently all photos have been well oriented and pixel errors look fine.
I also have noticed that automated marker propagation stops functioning without reason after having merged two chunks.
Look carefully attached images and tell if selected images explain the shown roll estimated values in each case.
Yaw doe not seem consitent either. the last example with three images that were taken sensibly perpendicular to the fa├žade, appear to have very diferent yaw angles.
Best regards!

GEOBIT
« Last Edit: February 06, 2017, 10:35:25 PM by JMR »

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 13332
    • View Profile
Re: Wrong roll estimation in 1.3
« Reply #1 on: February 07, 2017, 07:00:32 PM »
Hello JMR,

It looks like so-called Gimbal Lock effect appears when the Pitch angle is close to 90 degrees.
Best regards,
Alexey Pasumansky,
Agisoft LLC