Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - jslyder

Pages: [1] 2
Hi Dr_C,

I use Metashape on a linux-based HPC cluster regularly at work.  I'll admit I'm largely ignorant on the nuts and bolts of how Metashape has been deployed in this setting, but I can comment on a few of your questions.  For licensing, it is my understanding that each individual processing node will require a license, so depending on the configuration you would likely need a few licenses to really take advantage of your HPC setup.  We have a floating license server with up to 60 licenses, so for us this is not an issue. 

As for the GUI, yes, you can access this.  I'm not sure about how it would work to send/receive calls from your windows machine, I imagine that would be possible depending on how your network was set up.  In our organization, however, Metashape is set up within the HPC cluster and we simply use a browser-based remote session to interact with this GUI. 


Thank you, that work perfectly. 

Has something changed in the Python API in version 1.7.1 that would make this solution not work for accessing the estimated marker coordinates via the console?  I was trying to run this on a project, and the estimated values from the script are very different from what I see in the reference panel.  A screenshot with the estimated values from one marker "IS9" from the reference panel and what the script prints is attached for comparison. 

The createDifferenceMask tool has been very helpful for masking background in typical three-band images.  However, it would be nice if the tuple used for the background parameter could be different lengths other than three to support one- or four-band images.  Thank you. 

Thank you, Alexey. 

Feature Requests / Have "Convert" button ignore fiducial markers
« on: May 15, 2020, 12:11:38 AM »
If I am working with a project with scanned film and I wish to convert to a different coordinate system using the convert button on the reference pane, it always messes up my project because it appears that this forces a conversion of the fiducial coordinates as well. 

Thank you

Bug Reports / Re: Generate Ortho - Big Tiff Error
« on: April 14, 2020, 06:47:03 PM »
Was there a resolution to this issue?  I'm getting a similar error, and am wondering if it may be related.  When building the orthomosaic for a pair of WorldView 3 scenes, I get the error below.  This happens for both the panchromatic images as well as for the multispectral ones.  I'm on 1.6.2 build 10247.Thanks. 

2020-04-14 09:25:10 Orthorectifying images...
2020-04-14 09:25:10 Orthorectifying 2 images
2020-04-14 09:34:18 16APR21181234-P1BS-500658906010_01_P001.TIF: 42523x51601 -> 42523x51601
2020-04-14 09:34:31 16APR21181146-P1BS-500658906010_01_P001.TIF: 42523x51601 -> 42523x51601
2020-04-14 09:35:57 libtiff error: Write error at scanline 3712
2020-04-14 09:35:57 libtiff error: Write error at scanline 3712
2020-04-14 09:36:13 libtiff error: Write error at scanline 48896
2020-04-14 09:36:13 libtiff error: Write error at scanline 48896
2020-04-14 09:36:22 Error: TIFFWriteTile: unexpected error: memory stream

Got it, thanks Alexey. 

I have many thousands of images that I wanted to organize into different calibration groups based on the date they were collected, which is reflected in the label.  I wrote the script below to accomplish this.  While the script works for this purpose, it also forces the resolution of all images to 0x0 (screenshot attached).  If I try to align the images, it goes through one batch where 0 points are detected for each image, and I get the error "Error: Can't select tile size."  If I click on the image in the workspace, it still shows the appropriate dimensions,17588 x 16039. 

Is this a bug, or am I doing something wrong in the script? 
Thank you

Code: [Select]

chunk =

# Make list of dates to create calibration groups
calGroups = []
for camera in chunk.cameras:
    yr = camera.label[:2]
    if yr not in calGroups:

#Create the calibration groups
for i in calGroups:
    sensor = chunk.addSensor()
    sensor.label = i
    sensor.type = Metashape.Sensor.Type.Frame
    sensor.focal_length = 208.047
    sensor.pixel_height = 0.0125
    sensor.pixel_width = 0.0125

#Assign cameras to calibration groups. 
for camera in chunk.cameras:
    for s in chunk.sensors:
        if s.label == camera.label[:2]:
            camera.sensor = s

General / Re: Sentinel-2 satellite processing
« on: March 18, 2020, 06:17:22 PM »
Sentinel-2 doesn't collect stereo imagery... seems like a better bet would be to load a scene into any GIS or image processing software, (e.g., QGIS, Arc, Imagine) and use  a raster calculator function.   

Feature Requests / Support for NITF imagery
« on: March 05, 2020, 12:18:23 AM »
Now that the RPC sensor model has been added, it would be nice if there were support for NITF (.NTF) images.  Thanks! 

General / Re: All tie points are invalid for WorldView Scenes
« on: February 14, 2020, 05:17:19 PM »
Great, thank you!

General / Re: All tie points are invalid for WorldView Scenes
« on: February 13, 2020, 11:58:25 PM »
Hi Alexay,

Camera type is set to RPC, it is actually set to that by default.  I've attached a few screenshots showing that there are many matches found but they are all treated as invalid.  It's unclear whether I can share these images based on our license, but I'm trying to track down that information now.


General / All tie points are invalid for WorldView Scenes
« on: February 10, 2020, 10:39:38 PM »

I am attempting to process WorldView 2 and 3 scense in version 1.6.1.  After alignment, all my images say that they are aligned, but there are zero points in my point cloud.  When I look at view matches, I see several thousand matches between the pairs of images, but all of them are invalid.  I'm having this same issue across a number of tests, from 2-4 scenes, both multispectral and panchromatic, with and without guided matching, as well as both WV2 and WV3.  Any thoughts on what might be going wrong here?  I followed the steps in the tutorial section, but did not always build optimized TIFFs. 

As an aside, every time I use the "Viiew Matches" pane Metashape crashes after about 30 seconds. 

Thank you for your help. 

General / Capture distance in reference settings.
« on: January 15, 2020, 06:16:23 PM »
I recently noticed that "Ground altitude (m)" has disappeared from the reference settings and appears to be replaced by "Capture distance (m)."  Does this serve the same purpose as ground altitude did, i.e. to allow proper preselection with oblique imagery? 

Thank hyou

Pages: [1] 2