Messages

1

Bug Reports / Re: Geoid heights and tiled model

« on: April 07, 2023, 10:50:22 AM »

Didn't actually noticed that. I agree it should be clear when it comes to export to TLS (a warning would help just not to waste time checking or miss that when it goes to customer).

From my experience there are a lot of conversion problems regarding coordinate systems with custom geoid in a lot of GIS software. Sometimes to avoid such problems in 3rd party software I switch Agisoft project definition from CS with geoid height to standard CS without geoid (agisoft assumes elipsoidal by default - this can be done just before exporting) with no CS height conversion/recalculation. So the CS definition (in agisoft and in exports) will say this is ellipsoidal height, but the actual height value will be geoid height. It is always better to check the result, but with small scale project and some GCP's potencial height errors that come from this "fraud" are not noticable in the products.

So use Geoid coordinates, but you can deceive agisoft that this is ellipsoid height he should use

Not a full solution but it can get the work done.

2

Bug Reports / Re: Assertion "3478972301441" failed at line 720

« on: April 05, 2023, 08:28:57 AM »

Hello Everyone
I just encountered very similar problem.
I have a bunch of scans from leica scanner orinted in Cyclone and exported to e57 separately. I can process all of them in ContextCapture, but some of them (3 out of 20 scans) fail to process to mesh in Agisoft giving error: failed #44 BuildModel.prepairDepthMaps (1/1): Assertion "23671468921001171" failed at line 1149!
It is curious, that same 3 scans in e57 fail to convert to RCP in recap...
Have proceed with some small investigation...
1. changing region in Agisoft does not help (smaller, bigger, different parts of the scan).
2. deleting points not in range, updating, compressing point cloud in Agisoft does not help.
3. eksporting without color or intensity does not help
4. filtering point cloud containing 700.000.000 points giving 0.5mm resolution gives point cloud with 1.700.000.000 :O (something is not right... should be smaller if i understand correctly)
5. importing point cloud with different coordinate precision settings didnt help.
6. using 2.0.2 or 2.0.1 - no difference - fresh projects each time.
7. Two desktops (64 GB RAM and 128 GB ram, different hardware) - same error.
Two of them are in scanners high resolution giving 700.000.000 points, but one of them is in medium resolution with about 200.000.000 points (as all of working scans) so I guess this is not the size problem.
What we noticed is that the scan is processing in Agisoft and Recap if we delete points close to the scanner position in Cyclone before exporting. Is it possible that the problem is too dense point cloud on objects near the scanner??? If i delete same points inside Agisoft (after import) it does not solve the problem - scan is still "faulty" in some way and does not want to process to mesh.
Anyway, the workflow would require to load the scans and just process. Identifing 3 out of 20 scans which is cousing the problem is a bit time consuming (assertation error does not give strait information about the name of scan/scans in project, so I guess it would be best if you could check and solve the problem on importing clouds.
I can send you a link with a sample scan or two. Please provide me an email (or other private masseging system) I can use for passing it.
Best regards

3

Bug Reports / Re: Point Cloud export - split in blocks

« on: January 20, 2020, 01:35:55 PM »

1.6.1 works fine now.

Thank you for quick response and fix

4

Bug Reports / Re: Point Cloud export - split in blocks

« on: January 09, 2020, 04:09:05 PM »

Yes I am using Metashape Pro 1.6.0 release version (build 9925).

5

Bug Reports / Point Cloud export - split in blocks

« on: January 09, 2020, 07:21:49 AM »

Hello everybody.

I noticed a bug in 1.6 and 1.5 versions of metashape when exporting point cloud with use of "Split in blocks" function.
I have a  project that uses Poland CS2000 zone 5 CS (EPSG:2176). when I export point cloud using "Split in blocks" function (1000x1000) then first tile is correct but every next tile contains complete data (without any tileing). I have checked that using las and pts - same problem in both cases. U can see that in sizes of resulting las tiles:

test-0-0.las   71 035 822   2020-01-09 05:19   -a--
test-1-0.las   406 845 114   2020-01-09 05:19   -a--
test-2-0.las   406 845 114   2020-01-09 05:19   -a--
test-3-0.las   406 845 114   2020-01-09 05:19   -a--
test-4-0.las   406 845 114   2020-01-09 05:19   -a--

When I take same project and use Photoscan 1.2.6 then the export splits point cloud correctly, so I guess this is a bug.

Using Windows version.

Best regards