Topics

1

Python and Java API / Some user interaction / dialog box basics

« on: February 03, 2021, 08:03:33 PM »

Two questions:
1. I have a custom dialog box that runs from a menu item. When the user adjusts values in the dialog and runs a routine, I want their adjusted values to show up as default the next time they pull it up from the menu (within the session is fine - I don't need to save between sessions).  I tried using global variables within the python script: defined at the top of the script and then set to new values as the routine runs.  The original values set at the top of the script do appear within my dialog class, but setting the new values does not allow them to persist into the next run of the dialog.  How is this best handled?

2. My dialog is modeled primarily from the "split in chunks" script, where the dialog does not close until the user clicks on the quit button.  I would like to automatically close the dialog after the routine runs.  Simply calling QtCore.SLOT("reject()") does not seem to do the job.  How should this be done?

Thanks!

2

Feature Requests / Constrained camera altitude but not horizontal position

« on: February 03, 2021, 05:28:27 AM »

I have a project with no GPS camera positions, but I do have good 3D GCP positions.  The project is for underwater topography and there is a lot of noise in the tie points due to light ripples on the ground surface.  Modeling camera positions is going ok, but I often have camera positions modeled too high or too low.  All camera positions were taken from just below the surface of the water, so I do know the altitude +/- about 20 cm.  It would be great to specify the camera Z-values and an associated accuracy without needing to specify X and Y.

3

Python and Java API / Removing markers from selected cameras (python)

« on: February 03, 2021, 04:59:16 AM »

I'm trying to script the removal of markers that have been placed on selected photos.  I can get them to go from green to blue, and I've seen other scripts for removing blue markers, so I thought it should be simple to combine the two concepts.  But it isn't working.  The following shifts them to blue but does not remove them.  Any suggestions?

Code: [Select]

def cameraMarkerClear():
        print("Clearing markers from selected cameras...", flush=True)
        chunk = doc.chunk
        countCams = 0
        countMarks = 0
        for camera in chunk.cameras:
                if camera.selected:
                        countCams = countCams + 1
                        for marker in chunk.markers:
                                proj = marker.projections[camera]
                                if proj is not None:
                                        if proj.pinned:
                                                countMarks = countMarks + 1
                                                proj.pinned = False  ### this turns them from green to blue
                                                proj = None  ### this should be what other scripts use to remove blue markers
                                                print("removing " + marker.label + " from " + camera.label, flush = True)
                       
        print("Removed " + str(countMarks) + " markers from " + str(countCams) + " cameras.")

app.addMenuItem("Testing/Clear Markers From Selected Cameras", cameraMarkerClear)

Thanks!

4

Feature Requests / Reference Setting or Gradual Selection (filter) by distance from camera?

« on: April 23, 2020, 10:44:08 PM »

I do a lot of topography with oblique photos (walking around with the camera).  My feature request is to be able to remove (or prevent) photos from being used for tie-points based on distance.  It is really two requests, because I'd like to be able to remove both far away, and very close tie points.  Those far away have both poor resolution in the photograph and typically poor base-height ratios with other photos (yes, the Reconstruction Uncertainty get at this to some degree).  Tie points from the immediate foreground of my oblique photos can be very sensitive to lens model imperfections and often cause problems in optimizing my models.  So for my work, I would probably remove tie-point projections from a particular camera if they are closer than 5 meters or further than 100 meters.

I don't know the math behind the scenes well enough to know exactly what is best, but I can imagine two possibilities:

1. A pair of Reference Settings would be available that specify a minimum and maximum estimated distance from the camera for that camera (photo) to be used for a tie point.  This may not be possible with initial alignment if the scale is not yet available, but could go into force once the distance is known.

2. A pair of Gradual Selection filters for removing tie points based on camera distance.  Since these filters act on tie points that have multiple cameras involved, perhaps it would need to work on average distance from the cameras.  I think that would still do an OK job for what I'm thinking of.  It would be ideal if the measure was in meters.

Thanks!

5

Feature Requests / Setting for marker size

« on: January 28, 2018, 06:35:44 AM »

Much like discussed in this post (http://www.agisoft.com/forum/index.php?topic=6788.msg32781#msg32781), I now have a laptop with a high resolution screen.  But the settings available in windows for increasing the size of icons and such has no effect on the marker symbol that PhotoScan overlays onto photos.  I'm now squinting at my screen to see where I placed markers... the symbol (both circle and flag) now covers little more than 1 mm on my screen - the circle itself less than half a mm.

A setting in preferences for the icon size would be very helpful.

6

Feature Requests / Alignment indicator in photo pane... maybe with some numbers too

« on: February 12, 2016, 08:14:42 PM »

Presently (v. 1.2.3) the Worksplace provides a list of all photos in the project with an indicator of "NA" for non-aligned photos, but you only see the photo thumbnail one at a time as you click on them.

Conversely, the Photos pane shows thumbnails of everything, with indicators for markers placed in the photos, but no indication of whether or not they are aligned.  It would be great to add an icon indicator, much like the marker flags, that shows if the photo is aligned, or better yet, maybe have a gray frame around the image (similarly, disabled photos currently indicated with a red icon could have a red frame).

To make this even better, perhaps a couple numerical values could be displayed next to the icons?  I'd suggest [Projections] / [Total Key Points Available] : [Error (pix)]

Why do this?  In my case, I work primarily with very oblique photos for natural topography.  I often need to go through a manual process of resetting alignment of photos to get an optimal model.  Having the thumbnail and the alignment indicator right together, particularly with adding the Pixel Error and some Key Point details would increase my work efficiency quite a bit.

Thanks!
-Eric

7

General / Meaning of ! (exclaimation point) when optimizing cameras?

« on: November 24, 2015, 02:56:56 AM »

I'm sure it indicates some sort of problem in my model, but it might help to know what problem to look for... so just what does the exclamation point mean in

Optimizing point cloud...
adjusting: xxxxxx!xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx 0.569732 -> 0.517

also I think I've seen cases where the last x was followed by a '-' Does that have some meaning too?

Thanks!

8

Python and Java API / Split in Chunks script yielding empty chunks

« on: August 18, 2015, 08:01:43 PM »

I'm doing some work with the "Split in Chunks" script (http://wiki.agisoft.com/wiki/Python), which generally works fantastically!  However, I have found some cases where the final merged dense point cloud is missing a number of the chunks, or the dense cloud is missing entirely.  If I don't have the script merge the chunks back together, I find some of the individual chunks are missing the point clouds. 

I now have a small demo project of 8 photos that just running Build Dense Cloud in the Workflow menu yields a dense cloud with a little over 1 million points.  Using the script, slightly modified (attached) to provide settings (using medium quality, filtering disabled), splitting to 4 chunks, then merging, results in a merged chunk that has only the photos (no point cloud or tie points at all).  Running the same thing, except without merging, results in 4 chunks, 3 with tie point clouds and dense clouds, and 1 with only tie point clouds.

I'm wondering if something is going wrong with the script?  Or is there some other problem occurring?  What can I do to fix it?

Thanks!!!