Messages

1

Camera Calibration / Re: Calibrate cameras only on markers

« on: March 09, 2022, 11:42:49 AM »

Hi there,
sorry that was really not very clear. In the model view you have to select all points e.g. rectangle selection or select one point and invert the selection and then use the delete button on your keyboard. make sure you don't accidentally delete your markers. the idea is deleting every individual point of the sparse point cloud instead of deleting the point cloud.
cheers

2

Camera Calibration / Re: format conversion - Parameter consistency

« on: December 29, 2021, 12:16:45 PM »

In addition I dug out this old post, that helped me to approximately calculate the radial parameters.

Hi Alexey/All,

I apologize up front if this has been answered elsewhere, but can you point me to the algorithm/papers that describe the conversion  between the australis calibration coefficients and the defaults shown in the Photoscan GUI?

For the radial terms, It appears that the conversion is roughly (but certainly not exact):

k1_photoscan = f^2 * k1_australis
k2_photoscan = f^4 * k2_australis
...

where f is the focal length.

Any insight greatly appreciated.

Thanks!

3

Camera Calibration / Re: format conversion - Parameter consistency

« on: December 29, 2021, 11:55:28 AM »

Hi Alexey,
thanks again. The appendix was indeed helpful for calculating distortion profiles. However it does not say anything about parameter conversion. What are the equations used when exporting camera calibration in the australis format? From this I should be able to reverse-engineer the equations to convert my parameters from VMS to Metashape.
Thanks,
Johannes

4

Camera Calibration / Re: format conversion - Parameter consistency

« on: December 28, 2021, 06:41:09 PM »

Hi Alexey,
thanks for your reply.

My aim is to compare the MetaShape calibration and the external calibration.

I have done the calibration in MetaShape using the same dataset and found that the K parameters show too much correlation. Additionally, the external software can only determine K1-3. Therefore, I have decided to omit K4 altogether.

I understand that MetaShape approximates the K1-4 parameters on import. Does that mean the conversion is not done by a straight forward equation? I was assuming the conversion just calculates pixel values from the mm values. As mentioned above, my parameters are generated in VMS (not Australis).
Can you provide conversion equations? I could calculate my own values and feed them directly into the camera calibration window without using the conversion tool.

Thank you,
Johannes

5

Camera Calibration / format conversion - Parameter consistency

« on: December 27, 2021, 06:44:39 PM »

Hi All,
I have calibrated my camera using an external software (VMS). I have saved the parameters in a mm format.
Is there a way to import and convert the parameters to use them in MetaShape, but without estimating additional parameters.
I.e. my calibration uses K1, K2 and K3 but after importing into MetaShape there is a value for K4. I have used the Australis Type.

Thanks for your help,
Johannes

6

Camera Calibration / Re: marker based calibration - "optimize cameras" greyed out

« on: December 20, 2021, 04:50:04 PM »

Hi JohnH1,
thanks for your reply. I have found that you can get rid of tie points by selecting all tie points and deleting them. If you delete the sparse point cloud from the workspace you will end up with a greyed out "optimize camera" button. This way you can trick MetaShape into performing a marker-based camera calibration.

I now believe the problem described above is not related to the software version but the method of deleting the tie points.
Cheers,
Johannes

7

Camera Calibration / marker based calibration - "optimize cameras" greyed out

« on: March 18, 2021, 03:05:33 PM »

Hi All,
I am trying to use Metashape for camera pre-calibration. I want to calibrate the camera based on markers only and therefore remove all the tie-points after the bundle adjustment.  The reference data (marker coordinates) are in a local coordinate system. We have published a paper before using this workflow and the old project still works fine.
Now I am trying to do the same workflow, only with a new set of cameras and marker coordinates, but the "optimize camera" button is greyed out (as well as generate report etc).

I have downgraded the to several previous versions already, but the issue does not appear to be a new "feature". The old project allows camera calibration without tie points and the new project does not.

I'd appreciate your ideas.

Cheers,
Johannes

8

General / Vertical control points - no visible features

« on: October 29, 2020, 12:59:12 PM »

Hi All,
my survey design contains a number of control points along a stream and thus unfortunately does not lock the model strong enough in the horizontal axis perpendicular to the stream. I.E. the model has a tendency to rotate slightly along the river axis.
Now there is a number of additional GNSS measurments available that were intended for the validation of the Z-accuracy of the DEM. Some are located beyond the control points and could extend the model. Unfortunately they are not related to visible features. Is there a way to include them as z-only control points?
Thanks for your help, it's much appreciated.
Johannes

9

General / Re: Forward and Backward Projection: Project global mask on scenes

« on: October 20, 2020, 04:10:45 PM »

Hi All,
I've found a solution for projecting mask from the model to my scenes in this post: https://www.agisoft.com/forum/index.php?topic=12640.msg56180#msg56180

I am still desperate to find a way of backward and forward projecting actual data. especially between sensors.

Please let me know if you have ideas.

10

General / Re: Is there a way to put mask from sparse point cloud to image?

« on: October 13, 2020, 10:47:03 AM »

Hi Phogi,
I would be really interested in a global mask function, too. Please let us know if you figure out how to do it.

Concerning your AOI: You could create a copy of your chunk and remove all cameras that do not cover your AOI. Should make the processing faster.

11

Bug Reports / Re: crashing when aligning images

« on: October 13, 2020, 10:39:04 AM »

Hello Alexey and landing.st,
for me updating the drivers instantly solved the problems. Sorry, that was an embarrassingly simple solution.

However, I still sometimes get a black screen, which I can get rid of it by changing to stereo view (and back). But some graphics remain funny: markers would still appear as black boxes and the overlay number of points is not readable. This can be fixed by re-booting, so no big deal, just fyi.

12

Bug Reports / Re: crashing when aligning images

« on: October 02, 2020, 09:49:19 AM »

Might be related or not:

I am remoting onto the workstation from home, using the remote desktop with the NVIDIA patch that allows running things on my GPU. However, every other day Metashape opens as a black window, which is apparently related to remoting in. I can get back to "normal" by changing the display mode to stereo, but the marker flags remain black boxes and text is not readable, but it works. After re-booting the workstation the interface looks fine again.

13

Bug Reports / crashing when aligning images

« on: October 01, 2020, 10:57:34 AM »

Hi All,
Metashape keeps crashing when trying to align photos. It finishes detecting points and crashes at about 75% of estimating camera locations. The dataset consists of 917 thermal images (tif) 640 x 512 pixels resolution.

I've managed to align them on high quality before and later refined the selection of photos. I've also updated Metashape. Now it would only run on low quality but crash on medium, high or highest.

please find a crash report attached and an example image behind the link (file too large).
Any ideas?

Johannes

https://newcastle-my.sharepoint.com/:i:/g/personal/b8045831_newcastle_ac_uk/EbwZnY5fMYpFisC8AhyCynsBU-YHuSEuKF6i107Nu_k7oQ?e=HfMA3T

14

General / Forward and Backward Projection: Project global mask on scenes

« on: September 17, 2020, 04:12:43 PM »

Good Morning All,
We are working with aerial RGB, multispectral and thermal imagery of a river. The datasets were not synchronously captured and thus can not be used as bands. We would like to use backward and forward projection between image and world coordinates for two tasks:

-   Mask out water: This could either be done by a global mask projected from individual scenes or a global mask projected onto the individual scenes. Ideally this could be used between all sensors.
-   Create synthetic bands: An orthophoto created from imagery of sensor A could be used to create synthetic scenes as additional bands for sensor B. This would help us to run image classification on the single scene rather than the stack of orthophotos.

To our understanding all required parameters are known: Georeferenced orthophoto, sensor locations and directions, lens calibration and depth maps. However, we couldn’t find any information if this has been done before and if there are scripts to do so.
I would be very grateful for every helpful comment.
Johannes