Forum

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 - WaltDixie

Pages: [1]
1
General / Re: Geoid loaded but not recognized in custom EPSG crs
« on: March 22, 2023, 12:24:06 AM »
Anybody that can help with this issue?

2
General / Geoid loaded but not recognized in custom EPSG crs
« on: December 23, 2022, 02:02:23 PM »
The vertical datum "Ostend Height" is predefined in the following CRS predefined in Metashape:

Code: [Select]
crs=Metashape.CoordinateSystem('EPSG::8370')
crs_equivalent=Metashape.CoordinateSystem('EPSG::3812+5710')
The vertical datum EPSG code is 5110, the associated horizontal & vertical CS EPSG code is 3812/5710. So the above two crs are equivalent.

I downloaded the geoid from https://github.com/OSGeo/PROJ-data/tree/master/be_ign
I adapted the TIFFTAG_IMAGEDESCRIPTION so that MEtashape recognizes the geoid as a vertical datum with epsg code 5110.
I'm able to do a coordinate transformation from lonlat towards this system.

Code: [Select]
crsref=Metashape.CoordinateSystem('EPSG::4326')
coord = Metashape.Vector([2,50,10]) #lonlatz
crsref.transform(coord,crsref,crs) #works
crsref.transform(coord,crsref,crs_equivalent) #works, same output

However, if I define a custom crs with the same vertical crs EPSG code (5710), running a similar transformation returns the error "Vertical datum missing".

Code: [Select]
crscustom=Metashape.CoordinateSystem('EPSG::32631+5710')
crsref.transform(coord,crsref,crscustom) #fails, reports 'Vertical datum missing'

My expectation is that it would not matter which horizontal CRS is used for the vertical datum to be recognized.
I can't detect any difference in the wkt string.
So, can you explain what is missing here?



3
The raster calculator seems to auto-convert raster data to float32 format.
Is there any way to avoid this? Eg. 16-bit micasense data and a raster transform to export the first 4 bands.
I've tried entering "int(B1)" in the calculator, but this is apparently not allowed.

(See also https://www.agisoft.com/forum/index.php?topic=7271.0 , where it seems to be confirmed that raster calculator approach always converts to float32.)

4
Feature Requests / Export orthomosaic as band interleaved
« on: November 07, 2018, 01:44:54 PM »
I would prefer to export orthomosaic (and photos) similar to the following geotiff creation options:

tiling by 512x512 (gdal:  -co TILED=YES -co BLOCKXSIZE=512 -co BLOCKYSIZE)
compression in jpeg, using band interleaved setting (gdal: -co COMPRESS=JPEG -co JPEG_QUALITY = 99 -co INTERLEAVE=BAND)

It seems the API does not expose the option to change from pixel interleaving to band interleaving. Could this be adapted?
In band interleaving mode, the photometric RGB default setting commonly used  for RGB JPEG compressed data doesn't make sense either so should be adapted accordingly.



5
Python and Java API / Re: Changing (color etc.) style of masks in Photoscan
« on: February 14, 2017, 06:14:31 PM »
The last point can be ignored. The mask that I added had values of 0 and 1, and what was needed was 0 and 255.
That is, the 0 and 1 values seemed both black in the thumbnail view which expects mask images in [0, 255] range.

Still wonder about the other options though!

6
Python and Java API / Changing (color etc.) style of masks in Photoscan
« on: February 13, 2017, 04:04:39 PM »
Hi, I was wondering if there is a way to adjust the way masks appear within the Photoscan GUI.

At the moment, there are three style changes I'm interested in:

  • Overlay mask transparency on pictures : currently, when viewing an image, where its mask is defined the image appears darker (ie it's a 'black' transparency). For certain applications, I'd like to set it to 'green' transparency or some other colour, and adjust the level of transparency I want.
  • Overlay mask contour on pictures: I really like how the white contour is around the masked area in the standard style. Again, other colours or thicknesses would be helpful in some cases.
  • Adjust thumbnail view: The standard style shows the mask as black/white in thumbnail section. The main problem here is that I managed to add masks that don't show up in thumbnail for some reason. Is there a way to set those (the API only shows access to thumbnails for the Photo class, not the Mask or its Image class)?

For the last point, this is from masks created from an array that I converted into a single channel Image (with channel='I'), and then assigned to a mask using created_mask.setImage(img)

7
General / Re: Not Enough Memory
« on: January 15, 2017, 12:57:04 PM »
The issue will hopefully disappear if you go for medium quality or smaller size images (JPG or simply using smaller pixel regions or...).
AFAIK, Photoscan will load a higher sampled version of each image during the DPCL process, and this is the main reason that causes the memory issue. The factor x of higher sampling is determined by this medium/high/ultrahigh setting.

Perhaps check if your system settings are in line with your hardware.

Another workaround is to divide your target in chunks and then merge/combine them afterwards.
 Or start learning python scripting.

I'm still learning this. I was just planning to start another thread, where I try to list all the possible options to solve this issue.


Pages: [1]