Agisoft Metashape
Agisoft Metashape => General => Topic started by: lunar on June 17, 2016, 08:30:49 PM
-
what is Adaptive model fitting option in Align Photos dialog ?
-
Good question, I was wondering about this too.
-
I would like to know more too. Alignment seems to be better out of the box with this option. For non GCP's aerials, almost skips the need for the "optimize alignment" option. Is it possible to know a little bit more bout this option. It's not in the manual yet :-)
Cheers
-
There is a new check box in align photos now called adaptive camera model fitting. What does it do?
-
Adaptive camera model fitting option enables automatic selection of camera parameters to be included into adjustment based on their reliability estimates. For data sets with strong geometry it helps to adjust more parameters during initial camera alignment. For data sets with weak camera geometry it helps to prevent divergence of some parameters. For example, estimation of radial distortion parameters for data sets with only small central parts covered by the object is very unreliable. In the same way estimation of principle point coordinates can produce unreasonable results in case of strictly parallel camera movement.
-
Thanks for the reply! It seems it helps with alignment accuracy in both data sets with weak and strong geometry. In what scenarios should we use this feature and when should it be avoided?
-
Thank you for the explanation Dmitry. I've rerun projects with this option and results get always better. I can conclude that, in most aerial surveys done by UAV, this option must be on.
Cheers
-
Hello RHenriques,
The exception for the aerial flights for this option is usually related to the very stable flight (usually with the gimbal) on rotor-wing aircraft when the camera orientation doesn't change at all (for example, image up is always looking in the same direction irrespectively to the flight direction).
Hello igor73,
Adaptive camera model fitting might not work when there are too few matching points between the images and therefore the image alignment is not stable (one of the examples of such cases have been discussed on the forum not so long ago and has been related to the full body scanning).
-
Very interesting!
Does it have any impact on the overall time needed for computation? Does it work with preselections as well?
-
Dear all,
i'm getting back to your old post related to "adaptive camera model fitting".
I have a weak set of 350 photos from an indoor flight in bad lighting conditions. As expected i experience problems in alignement.
But I get all of the photos aligned when "adaptive camera model fitting" is not selected and alignment is at "low".
What is your experience with such settings?
Accuracy of each point in the later pointcloud will be lower, right?
Greetings and thanks a lot
Alex
-
Hi Alexey Pasumansky,
I would like to know how to interpret the errors of my check points (Usango GCPs) that I have obtained using and not using Adaptive camera model fitting.
When I don't use them, I get an error in altitude (Z) of 1.5 meters. On the contrary, using Adaptive camera model fitting, this error is reduced to 55 cm. Which is quite positive.
On the other hand, when I use "Adaptive camera model fitting", I get also a very high value of the B1 parameter : -34.8111. And according to a publication of yours titled "What does camera calibration results mean in Metashape?" the value of B1, B2 should not be more than 10.
So how should I interpret my results?
Should I use or not: Adaptive camera model fitting, even if this is going to modify greatly my B1 parameter (affinity coefficients)?
Link to the publication: https://agisoft.freshdesk.com/support/solutions/articles/31000158119-what-does-camera-calibration-results-mean-in-metashape-
Regards,
Jean Pierre