1
Bug Reports / Re: Deleted tie points from all cameras re-appear, when realigning any one camera
« Last post by Iluvathar on Today at 10:54:42 AM »Do you have the keep key points option enabled ?
Recent Posts
2
Agisoft Viewer / Re: Colored polylines
« Last post by Craig M on Today at 01:40:14 AM »Thanks for the response.
I have sent an email to the listed email address.
I have sent an email to the listed email address.
3
Bug Reports / Deleted tie points from all cameras re-appear, when realigning any one camera
« Last post by Mariusz_M on January 23, 2025, 10:51:45 PM »There is a bug in 2.2.0 that makes all deleted tie points re-appear when I align one camera, even if it this camera was previously not even aligned.
Steps to recreate.
1. Align cameras on a photoset, where you know not everything will align.
2. Make note how many tie points are in total.
3. Select and delete a lot of tie points, like a half, so you can clearly see that is missing.
4. Right click on a camera that is not aligned and choose "Align selected cameras". If it manages to align, it will show not only the points belonging to this camera, but also all deleted points. The total number of the tie points will equal what you noted in Step 2.
Plan B In case the camera does not want to align:
5. Make note how many points you have after Step 3, since you could not complete Step 4.
6. Reset one of the cameras that is already aligned and once it is reset, the points belonging to this camera will disappear.
7. Re-align the camera using Right Click/Align Selected Cameras. You expect only to bring back the points belonging to this camera. And this is how it has been forever. But with verison 2.2.0 you get all the points back from all the cameras, even the previously deleted half of the points in Step 3. So the total numer of points is again what you noted in Step 2.
It looks like all the tie points from the initial cloud are saved and when a camera is aligned, instead of bringing only the points that belong to it, the software brings all the stored points of the initial cloud. This makes optimizing cameras difficult, because the points delted manually or with gradual selection are back and still taken into consideration for re-optimization.
Steps to recreate.
1. Align cameras on a photoset, where you know not everything will align.
2. Make note how many tie points are in total.
3. Select and delete a lot of tie points, like a half, so you can clearly see that is missing.
4. Right click on a camera that is not aligned and choose "Align selected cameras". If it manages to align, it will show not only the points belonging to this camera, but also all deleted points. The total number of the tie points will equal what you noted in Step 2.
Plan B In case the camera does not want to align:
5. Make note how many points you have after Step 3, since you could not complete Step 4.
6. Reset one of the cameras that is already aligned and once it is reset, the points belonging to this camera will disappear.
7. Re-align the camera using Right Click/Align Selected Cameras. You expect only to bring back the points belonging to this camera. And this is how it has been forever. But with verison 2.2.0 you get all the points back from all the cameras, even the previously deleted half of the points in Step 3. So the total numer of points is again what you noted in Step 2.
It looks like all the tie points from the initial cloud are saved and when a camera is aligned, instead of bringing only the points that belong to it, the software brings all the stored points of the initial cloud. This makes optimizing cameras difficult, because the points delted manually or with gradual selection are back and still taken into consideration for re-optimization.
4
Bug Reports / Re: Look through shows the right view, but zoomed in
« Last post by Paulo on January 23, 2025, 10:09:03 PM »Hi Mariusz.
the problem is the view magnification has to change according to view height seen in meters vs view height in pixels
So a viewpoint.mag of 1 means that a view point with 512 pixels height you a seeing a height of 512 m... So you must adjust according to what your 60 deg FOV camara what the hei=ght in me ter should be...
A script can probable be developped...
the problem is the view magnification has to change according to view height seen in meters vs view height in pixels
So a viewpoint.mag of 1 means that a view point with 512 pixels height you a seeing a height of 512 m... So you must adjust according to what your 60 deg FOV camara what the hei=ght in me ter should be...
A script can probable be developped...
5
Bug Reports / Re: Look through shows the right view, but zoomed in
« Last post by Mariusz_M on January 23, 2025, 08:08:49 PM »Any
6
Feature Requests / Re: ECW export
« Last post by Atmospheric on January 23, 2025, 08:03:18 PM »+1
7
General / Re: Different camera, same chunk
« Last post by Alexey Pasumansky on January 23, 2025, 07:07:57 PM »Hello HakunaMatata,
Metashape will automatically split images taken by different cameras into separate calibration groups, but it is better to check that it was done correctly (via Tools Menu -> Camera Calibration dialog) before starting the alignment.
And generally it should not be a problem to process the images taken by different cameras in the same chunk, providing that level of detail for overlapping images from different cameras is more or less the same.
Would be a problem, if you are trying to align nadir only images taken with the drone from 300+ meters with the hand-help camera images of the building walls - they have very different effective resolution and also most likely lack overlap, as walls are not properly seen from the drone and from the ground you would not see the roof, as it is seen from the drone.
Metashape will automatically split images taken by different cameras into separate calibration groups, but it is better to check that it was done correctly (via Tools Menu -> Camera Calibration dialog) before starting the alignment.
And generally it should not be a problem to process the images taken by different cameras in the same chunk, providing that level of detail for overlapping images from different cameras is more or less the same.
Would be a problem, if you are trying to align nadir only images taken with the drone from 300+ meters with the hand-help camera images of the building walls - they have very different effective resolution and also most likely lack overlap, as walls are not properly seen from the drone and from the ground you would not see the roof, as it is seen from the drone.
8
Bug Reports / Re: Decimal places error in processing and shape reports
« Last post by Alexey Pasumansky on January 23, 2025, 07:01:26 PM »Hello wojtas,
Thank you for reporting. We'll fix it in 2.2.1 update.
Thank you for reporting. We'll fix it in 2.2.1 update.
9
Camera Calibration / Re: Bright and Cloudy Dark Spots in Orthomosaic After Reflectance Calibration
« Last post by Geomet on January 23, 2025, 06:54:19 PM »Here are screenshots of the issues.
10
Camera Calibration / Bright and Cloudy Dark Spots in Orthomosaic After Reflectance Calibration
« Last post by Geomet on January 23, 2025, 06:48:43 PM »Hello,
I am experiencing an issue with bright and cloudy dark spots appearing in my orthomosaic after processing. The problem seems to originate from reflectance calibration, as I notice similar brightness/darkness inconsistencies in the individual images immediately after calibration.
When I zoom into the orthomosaic, I can see that the shadowy areas appear as polygonal-shaped regions. I am skeptical that this issue is related to cloud cover, as all my other surveys have been conducted under a range of cloud conditions without any issues.
I am using a DJI Mavic 3M multispectral drone and a MicaSense reflectance panel. I follow the same workflow for all 20 surveys I’ve processed, but this issue is only appearing in three of them.
I manually Calibrate reflectance by placing the panel images in the separate image group (Note: The Mavic 3M has a sun sensor, but it fails to calibrate irradiance on about 1/3 of the images for every survey I’ve done. Enabling or disabling the sun sensor does not seem to affect this issue).
The same workflow works fine for other surveys, but for the three problematic ones, I see this issue persist despite consistent settings and processing steps.
I’ve attached screenshots of the shadowy polygons in a reply to this message
Does anyone have insight into what could be causing these bright and dark spots in the orthomosaic? I’d appreciate any suggestions or recommendations! Thanks!
I am experiencing an issue with bright and cloudy dark spots appearing in my orthomosaic after processing. The problem seems to originate from reflectance calibration, as I notice similar brightness/darkness inconsistencies in the individual images immediately after calibration.
When I zoom into the orthomosaic, I can see that the shadowy areas appear as polygonal-shaped regions. I am skeptical that this issue is related to cloud cover, as all my other surveys have been conducted under a range of cloud conditions without any issues.
I am using a DJI Mavic 3M multispectral drone and a MicaSense reflectance panel. I follow the same workflow for all 20 surveys I’ve processed, but this issue is only appearing in three of them.
I manually Calibrate reflectance by placing the panel images in the separate image group (Note: The Mavic 3M has a sun sensor, but it fails to calibrate irradiance on about 1/3 of the images for every survey I’ve done. Enabling or disabling the sun sensor does not seem to affect this issue).
The same workflow works fine for other surveys, but for the three problematic ones, I see this issue persist despite consistent settings and processing steps.
I’ve attached screenshots of the shadowy polygons in a reply to this message
Does anyone have insight into what could be causing these bright and dark spots in the orthomosaic? I’d appreciate any suggestions or recommendations! Thanks!
