Messages

1

Python and Java API / Duplicate chunk but only selected cameras

« on: October 27, 2020, 04:07:50 PM »

hello,
I'm looking for duplicating the current chunk several times via scripting, I would like to copy only certain images of the original chunk each time.
can I submit a list of cameras of the original chunk to be copied to the chunk.copy() method?
thanks

2

General / Re: Skip the alignment

« on: November 03, 2017, 07:58:33 PM »

+1

Hi Martin,
we are probably sharing a similar problem, I've just posted these two questions:

http://www.agisoft.com/forum/index.php?topic=7901.0

http://www.agisoft.com/forum/index.php?topic=7949.0

to SAV,
I don't know about Martin, but I cannot perform the align photos process because my cameras are in panoramic configuration

I would like to perform a points-based chunks alignment, each chunk containing 4 photos (panoramic configuration) of which I know relative position, relative orientation and the camera calibration

3

General / Point-based chunks alignment with "non-aligned" photos?

« on: November 03, 2017, 07:24:45 PM »

Hello all,

I would like to align a set of chunks without first having performed the alignment of the cameras.
my goal instead, is to load the camera's location and orientation from a file
-BUT-
the coordinates I want to load does not represent the absolute locations of the cameras, but only their relative positions (within the chunk).
each chunk contains 4 cameras with their calibration, location and orientation fixed
the problem is that those 4 cameras are in a panoramic configuration and therefore I cannot perform the alignment

How can I align the set of chunks?
it seems like without the align photos process Photoscan does not "trust" my input coordinates.
Is there a way around this problem?

4

Python and Java API / Constraints for a multi-camera system

« on: October 23, 2017, 12:40:28 PM »

Hello everyone,

I'm currently testing the possibility of using python scripting to constrain the relative positions and orientations of a multi-camera system.
My goal is to acquire images with multiple cameras (let's say four) rigidly mounted on a hand-carryable device. During the alignment process, I would like Photoscan to orient the whole set considering the relative distances between the cameras and their relative orientations as constraints.
is there a way in Photoscan that allows me to implement such constraints both initially and during the bundle adjustment?

for now, after having calibrated the different cameras and their relative distances, I managed to impose them using scalebars but I'm struggling in constraining the relative orientations.
Furthermore, I'm not sure that Photoscan can consider those constraints "dynamically" during the bundle adjustment.

Any suggestion?

I thought I could impose the camera's coordinates and normals by creating a chunk for each shooting position of the multi-camera system, in this way I would have ended up with a bunch of chunks containing four already oriented images with their own calibration already loaded. I would then have run a point-based chunks alignment.
But my problem here is that Photoscan only accepts my coordinates as initials and doesn't let me run the chunks alignment without first having performed the photos alignment (which I cannot run because, for now, my cameras are in panoramic configuration).


Thank you all in advance,
Luca