Topics

1

General / point cloud precission degraded during export

« on: May 17, 2025, 03:17:26 PM »

Here is what I find.
Build a dense cloud within Metashape and pick a random collection of points. (I have done by means of the point drawing tool and then exported shape layer as survey point)
what I get is a list of points with coordinates like this. I have forced 4 decimal places on purpose.

469198.0893   4625535.6525   1020.9728
469200.9825   4625551.8803   1020.9609
469216.0177   4625558.8820   1020.7016
469235.0625   4625544.2959   1020.5023
469259.3013   4625557.6343   1020.7807
469272.4699   4625569.4271   1020.6212
469253.4956   4625577.6966   1020.1291
469223.2505   4625592.8820   1020.1857
469201.0821   4625580.1614   1020.6648
469180.5939   4625594.9720   1020.4086
469200.7915   4625598.4140   1020.4623
469222.4491   4625573.0174   1020.4413
469205.2609   4625563.3425   1020.8171
469234.5305   4625580.4200   1020.2658
469238.0023   4625565.8765   1020.3454
469228.7112   4625549.2787   1020.5605
469224.8822   4625529.2167   1021.2488
469212.8580   4625520.4667   1021.3260
469235.9866   4625517.7975   1020.7513
469262.2400   4625536.9397   1020.6720
469289.2738   4625544.9382   1020.3045
469297.8275   4625565.9952   1019.9460
469290.8024   4625580.5749   1019.9231
469257.4640   4625590.1000   1020.0492
469233.1867   4625595.1511   1020.1426
469250.0749   4625603.2549   1020.0406
469212.4016   4625582.2603   1020.4655
469178.2591   4625560.1662   1018.1213
469176.1339   4625535.1855   1020.6418
469205.2949   4625525.7642   1021.1719


Then I have exported the dense cloud and open into CloudCompare and picked a similar collection (not the same ones, I said its a random sample)
Here is what I have

469212.8600   4625596.7200   1020.3900
469193.0000   4625609.5200   1020.4000
469273.4700   4625628.8600   1019.9400
469236.1300   4625634.8300   1020.1400
469365.8400   4625617.5800   1019.4100
469335.3000   4625558.8000   1020.1400
469269.9900   4625564.1500   1020.6700
469238.5000   4625556.9900   1020.4500
469196.2000   4625576.6500   1020.9900
469174.4100   4625563.3200   1020.8000
469176.0100   4625538.6600   1020.8900
469233.8900   4625493.7500   1020.5500
469287.3700   4625511.6400   1020.3300
469304.8600   4625540.0500   1020.0500
469350.1100   4625617.9000   1019.1800
469353.5000   4625662.0400   1019.3200
469205.2900   4625511.9200   1021.3300
469201.3000   4625540.5200   1021.0300
469219.8300   4625568.7100   1020.5900
469250.2700   4625592.5300   1020.0800
469282.8000   4625578.4500   1020.4100
469244.4800   4625579.8600   1020.2400
469235.8800   4625597.7900   1020.1500
469224.9600   4625581.8400   1020.3300
469193.8700   4625594.0400   1020.5600
469203.3101   4625555.5700   1020.9900
469224.1200   4625544.0600   1020.7800
469235.2500   4625526.6700   1021.0000
469273.4900   4625524.4900   1020.7700
469287.3300   4625542.0800   1020.3800

all milimetres have gone! Why?

Same happens within Global Mapper or any program, so is it maybe Metashape degrading the point cloud accuracy during export?

Regards

2

Feature Requests / Obect vs terrain modes

« on: May 16, 2025, 09:51:36 PM »

I'll be frank. There are a number of things about the way we navigate in the model view that truly demonstrate how pathologically lazy our Agisoft heroes are when it comes to improve the GUI user experience.

We've spent years and years getting used to a stone in our shoe, but the callus on my foot hasn't made me forget about that damn stone.
Object mode is the only one I use because switching from it to terrain mode is so much painful as telling a violinist to switch the bow and violin to play Beethoven and use the other mode to play Vivaldi. They aren't two modes that can coexist in the same standard brain. IMHO

Since mostly all users I know prefer object mode, couldn't a toggle be added so that in referenced models, all movements could be forced to always keep Z up, but without changing anything else in that mode?

Is there anyone here, leaving aside those who feel comfortable watching the Tower of Pisa, who likes not being able to prevent the view from rolling to one side or the other when using free rotation without  fighting to pick the trackball's equator?

Only if the model isn't referenced, or hasn't yet been transformed into a world with Z up, should the current mode's crazy rotation be the default. This current "zero gravity" mode may be desirable in very specific cases and yes, keep it for such very specific operations, but for my taste, it's exceptional, not universal.

But if what I suggest doesn't seem like a sensible request, why not just add special screen zones near the window margins  that can be made visible and active by holding certain key to constrain the rotations of the current mode to rotations about camera view axis? Something like this:


Best regards.

3

Bug Reports / merge chunks + masks fails

« on: April 15, 2025, 04:03:25 PM »

Ive got 2 chunks both with masks, when I merge with "merge masks" checked, the merged chunk comes out with masks only in one set of images.

Regards
Geobit

4

General / Filter photos by shape

« on: April 14, 2025, 10:59:45 AM »

Filter photos by shape works only for the vertices of the given shape. There should be an option to select all the photos that can see the shape along the edges and optionally also inside a closed shape.

Agisoft, please, give us a proper polygon selection tool, (or at least let us use any closed shape for selections as serves for dense points), you can't keep ignoring this simple tool requested for years, freehand is fast but is very imprecise.
Let us select photos by polygon, select triangles by polygon, select tie points by polygon, PLEASE

5

Feature Requests / larger recent project list

« on: March 07, 2025, 08:56:33 PM »

I wish the recent projects list was longer, why not at least 12 rather than only 5?
Thanks

6

Bug Reports / Ambient occlussion messed up by block model

« on: February 18, 2025, 10:49:03 PM »

I had to build my mesh partitioned into blocks due to bad allocation error (not enough memory?). After that, I merged the blocks and got one single mesh model that appartently is clean and appears to be gapless in the seams.
However, the mesh topology in the seams is faulty and that causes these ugly black strips on the occlussion map.
The merge blocks operation needs some rework in my opinion
Regards
Geobit

7

Bug Reports / Mask from model has gone?

« on: February 01, 2025, 03:03:09 PM »

Edit: No, it is now under tools menu.

Regards

8

General / importing E57 scans. I'm lost

« on: September 24, 2024, 04:04:41 PM »

Hi, Agisofties
Why some e57 exported from a terrestria laser scanner turns into an equirectangular depth panorama and sometimes as a set of cube faces, and sometimes there is no image version at all. Is there a way to choose how Metashape reads a scan, and more importantly, is there any reason to consider one format result comes with any particular benefit or is better suited for any workflow.
The following image shows the result of importing three E57 scans, and ech behaves differently even if I use same settings at import dialog.

Best regards

JMR

9

Feature Requests / Align chunks by laser scans

« on: April 30, 2024, 11:33:49 AM »

I think it would be useful and I'd love to see this added to the methods. If I have two chunks connected by a common laser scan or point cloud, it would be super easy and fast to use it as reference source for transforming one chunk to the other.
Cheers

GEOBIT

10

Bug Reports / invisible boundary shapes

« on: December 01, 2023, 10:24:19 PM »

Dear Agisoft team.
Boundary shapes remain active in spite of the layer status. I wished to be able to keep different sets of boundaries stored in layers with different purposes, and thought I would be allowed to choose the active set of boundaries to produce desired deliverables with different extents according to the active layer while the remaining layers were disabled... but nope, it does not work. Boundary action override the layer status and the clipping was always the same. I think this is not the right behaviour.
Regards

Geobit

11

Bug Reports / Snaps...

« on: November 29, 2023, 02:56:08 AM »

Snap tool was eagerly welcome, but after a few years it's still terrible, one can hardly see what is snapping to the tiny red dot and the snapped feature is not highlighted at all. It is the subtlest snap feedback I have ever seen. It is truely hard to notice when not even totally invisible if one is trying to snap to some vertex under another feature. The two modes available are not enough (what you call axis snap, I would not call it a snap mode, ibut rather ortho drawing mode)

When we are editing a shape A and we drag a vertex that was created with a snap on another vertex of a shape B, they move together even if we are not necesarily wanting to edit both features. is there a way to prevent this behavior?  can we de-snap a node on a feature so it can be dragged alone?

I have reported by mail that snap isn't working at all during rectangle creation... the bug remains the same in latest stable release and so in the 2.1 beta. Please solve.

By the way if you give us a rectangle tool, please let us have persistent rectangles. Allow us to draw them rotated, let us modify height and width with a gizmo in the midpoint of each side, or let us drag two adjacent sides consistently orthogonal when we drag a corner.

Best regards.
Geobit & Accupixel

12

Feature Requests / desirable save/load features

« on: November 29, 2023, 01:58:00 AM »

Here you are three
Please enable a save/load region.
Please enable save/load view
please enable a save/load shown/hidden items setting. (or just add a  revert "hide all")
please anable a save/load layers structure and symbology

Regards

13

Bug Reports / Honey, I shrunk the kids...

« on: September 24, 2023, 10:21:19 PM »

Hello. I have experienced one of the weirdest things ever with Metashape 2.0.3
I'm processing a large project with a some thousands of photos, and after finished alignment and some other pretty standard operations, I closed the project until next day.
When I open, I see the sparse cloud has been dramatically downscaled for no reason...

Any thoughts?.

José Martínez

14

Feature Requests / Allow saving bounding region

« on: January 23, 2023, 03:01:53 PM »

I know we can use script to copy region from one chunk to another, but it would be useful to save it so it can be recalled easier.
Regards

15

Bug Reports / ... error reading pixel

« on: January 13, 2023, 04:06:54 PM »

What can I do?

[...]
2023-01-13 13:55:52   2.83981/4~71% avg groups per camera
2023-01-13 13:55:52   146.25/206~70% avg cameras per group
2023-01-13 13:55:52   0% leaves avg camera intersects in group
2023-01-13 13:55:52   1% big leaves avg camera intersects in group
2023-01-13 13:55:52   29% of leaves are big
2023-01-13 13:55:52   100% of voxels in group are from big leaves
2023-01-13 13:55:52 saved 4 groups (0.03125 KB) for 344 leaves
2023-01-13 13:55:52 saved 2.28516 KB of affected cameras and 0.804688 KB of exclusive cameras for all leaves
2023-01-13 13:55:52 Scheduling octree 344 leaves hist initing in 4 groups (86 avg leaves per group)
2023-01-13 13:55:52 Initing hists in leaves group #1/4: [0; 60)...
2023-01-13 13:56:02 total cubes: 105221949 cubes, 1605.56 MB, 163.966 MB/s
2023-01-13 13:56:02 Data: 2408.34 MB hists + 0 MB codes, 117.297 MB pyramid
2023-01-13 13:56:02 Found 2 GPUs in 0 sec (CUDA: 0 sec, OpenCL: 0 sec)
2023-01-13 13:56:03 [GPU 1] Using device: NVIDIA GeForce GTX 1070 Ti, 19 compute units, free memory: 7104/8191 MB, compute capability 6.1
2023-01-13 13:56:03   driver/runtime CUDA: 12000/10010
2023-01-13 13:56:03   max work group size 1024
2023-01-13 13:56:03   max work item sizes [1024, 1024, 64]
2023-01-13 13:56:03 [GPU 2] Using device: NVIDIA GeForce GTX 1070 Ti, 19 compute units, free memory: 7197/8191 MB, compute capability 6.1
2023-01-13 13:56:03   driver/runtime CUDA: 12000/10010
2023-01-13 13:56:03   max work group size 1024
2023-01-13 13:56:03   max work item sizes [1024, 1024, 64]
2023-01-13 13:56:08 openexr error: Error reading pixel data from image file "openexr". Data decompression (zlib) failed.
2023-01-13 13:56:08 openexr error: Error reading pixel data from image file "openexr". Data decompression (zlib) failed.
2023-01-13 13:56:08 openexr error: Error reading pixel data from image file "openexr". Data decompression (zlib) failed.
2023-01-13 13:56:09 openexr error: Error reading pixel data from image file "openexr". Data decompression (zlib) failed.
2023-01-13 13:56:09 openexr error: Error reading pixel data from image file "openexr". Data decompression (zlib) failed.
2023-01-13 13:56:09 openexr error: Error reading pixel data from image file "openexr". Data decompression (zlib) failed.
2023-01-13 13:56:09 [GPU 1] Exception
2023-01-13 13:56:09 [GPU 2] Exception
2023-01-13 13:56:09 Peak memory used: 8.44 GB at 2023-01-13 13:56:06
2023-01-13 13:56:09 deleting all temporary files...
2023-01-13 13:56:10 done in 0.734 s
2023-01-13 13:56:10 Finished processing in 2280.18 sec (exit code 0)
2023-01-13 13:56:10 Error: Can't decompress depth