About Control Points 'Distance' in Hugin scripting

[Álvaro Huertas]

Hi!

Im currently working with the internals tools that Hugin puts together (pto_gen, pto_mask, cpfind, etc) for solar panel stitching purposes.

I've noticed that the 'Distance' parameter in the control point tab (in Hugin interface) is playing an important role in the overall stitching quality (i don't really understand how hugin manage this parameter or how it's calculated, but it seems to work fine).

I would like to know how i can interact with this parameter from the individual tools like cpfind (since im programming a shell script) or how can i calculate it in order to improve the overall performance of my software.

Thanks in advance!

[T. Modes]

Hi,

alvaroh...@gmail.com schrieb am Mittwoch, 27. Januar 2021 um 12:39:39 UTC+1:

I've noticed that the 'Distance' parameter in the control point tab (in Hugin interface) is playing an important role in the overall stitching quality (i don't really understand how hugin manage this parameter or how it's calculated, but it seems to work fine).

I would like to know how i can interact with this parameter from the individual tools like cpfind (since im programming a shell script) or how can i calculate it in order to improve the overall performance of my software.

Take both points of the control points and remap these points from the input images to the output panorama. The error is the distance of these remapped points in the output pano. If the distance is zero, the points matches perfectly. The higher the distance the higher the points are separated in the output.

You can't manipulate the distance directly. It is calculated from the control points, the current geometric parameters of all images (yaw, pitch, roll, horizontal field of view, distortion parameters,…) and the selected output parameters (projection, field of view, canvas size).

Control points can be found cpfind, the geometric parameters can be optimized by autooptimiser  (maybe with pto_var to select which parameters to optimise) and the output parameters can be set by pano_modify.

Some additional information:

* autooptimiser optimise the geometric parameter in such a way the the mean error distance becomes as small as possible.

* The mean error distance can be calculated by checkpto.

[Álvaro Huertas]

Thank you so much!

Finally I could solve my problem with 2 iterations of autooptimiser + cpclean in order to delete the outliers cpo.