Agisoft Metashape
Agisoft Metashape => General => Topic started by: Dmitry Semyonov on September 19, 2021, 04:01:28 PM
-
Pre-release version of Agisoft Metashape 1.8.0 is available for download.
This pre-release is considered as unstable and is not recommended for production use.
Please use the links below for download:
Agisoft Metashape Standard edition
Windows 64 bit (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/metashape_1_8_0_x64.msi)
Mac OS X (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/metashape_1_8_0.dmg)
Linux 64 bit (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/metashape_1_8_0_amd64.tar.gz)
Agisoft Metashape Professional edition
Windows 64 bit (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/metashape-pro_1_8_0_x64.msi)
Mac OS X (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/metashape-pro_1_8_0.dmg)
Linux 64 bit (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/metashape-pro_1_8_0_amd64.tar.gz)
Agisoft Metashape Python 3 Module (requires Professional edition license)
Windows 64 bit (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/Metashape-1.8.0-cp35.cp36.cp37.cp38-none-win_amd64.whl)
Mac OS X (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/Metashape-1.8.0-cp35.cp36.cp37.cp38-abi3-macosx_10_13_x86_64.whl)
Linux 64 bit (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/Metashape-1.8.0-cp35.cp36.cp37.cp38-abi3-linux_x86_64.whl)
Agisoft Metashape Java Library (requires Professional edition license)
Download (https://s3-eu-west-1.amazonaws.com/download.agisoft.com/metashape-java-api-1.8.0.zip)
Metashape 1.8.0 Change Log (https://www.agisoft.com/pdf/metashape_changelog.pdf)
Metashape 1.8.0 Python API Reference (https://www.agisoft.com/pdf/metashape_python_api_1_8_0.pdf)
Please let us know if you have any problems using this version.
Please post problems and/or questions concerning other Metashape version in separate topics.
This topic will be removed when the final 1.8.0 version is released.
Note
Project files created with 1.8.0 version can't be opened with Metashape 1.6 and earlier versions.
EDIT:
Metashape 1.8.0 build 13794 was finally released. Thank you for useful feedback.
Please use the links on the main Metashape page for download:
Download Installers (http://www.agisoft.com/downloads/installer/)
-
I know that it's still very early pre-release but here's a first test. Same settings with Mesh generated from depthmaps:
1.7.4
(https://i.postimg.cc/ncMpgWjc/Screenshot-2021-09-20-171415.jpg)
1.8
(https://i.postimg.cc/xj69sQd4/Screenshot-2021-09-20-171442.jpg)
1.7.4
(https://i.postimg.cc/gk5mWK8y/Screenshot-2021-09-20-171522.jpg)
1.8
(https://i.postimg.cc/brV8dxMh/Screenshot-2021-09-20-171540.jpg)
On the plus side this issue which was introduced with 1.7 seems to have been mitigated:
1.7.4
(https://i.postimg.cc/nhydpX3w/Screenshot-2021-09-20-172549.jpg)
1.8
(https://i.postimg.cc/mgzdvYyH/Screenshot-2021-09-20-172701.jpg)
Mak
-
Apple Silicon support! Yay! I’m really excited to see how this improves performance. I’ll be getting the new MacBook Pro when it comes out soon, so this is great!
-
Apple Silicon support! Yay! I’m really excited to see how this improves performance. I’ll be getting the new MacBook Pro when it comes out soon, so this is great!
please let us know how it works !
-
I'm getting similar results to Mak11:
Some areas are vastly improved while others suffer quite dramatically.
-
Hello Mak11, hello wojtek,
For us as usual it will be very helpful to get a sample dataset that we can use for internal testing and comparison of the version results. So if you can share such data, please send the download link to support@agisoft.com.
Meanwhile, we already have some improvements related to the depth maps generation.
-
Hi Alexey
It says in the change log that this new beta version (1.8x) supports JPEG-R images. However, these files are imported as normal JPEG RGB files. What is the correct procedure for using them as radiometric files?
Cheers
-
It says in the change log that this new beta version (1.8x) supports JPEG-R images. However, these files are imported as normal JPEG RGB files. What is the correct procedure for using them as radiometric files?
Hello Renato,
R-JPEG files with embed temperature are loaded to Metashape Pro as multi-camera systems (you should see that option, when adding R-JPEG images to Metashape 1.8 ). After adding such images to the project, process as usual and to see the thermal orthomosaic use Raster Calculator - define thermal formula for output index (likely should be the following LWIR / 100 - 273.15), enable transformation and adjust the palette display.
-
It says in the change log that this new beta version (1.8x) supports JPEG-R images. However, these files are imported as normal JPEG RGB files. What is the correct procedure for using them as radiometric files?
Hello Renato,
R-JPEG files with embed temperature are loaded to Metashape Pro as multi-camera systems (you should see that option, when adding R-JPEG images to Metashape 1.8). After adding such images to the project, process as usual and to see the thermal orthomosaic use Raster Calculator - define thermal formula for output index (likely should be the following LWIR / 100 - 273.15), enable transformation and adjust the palette display.
Only for flir format? non dji format?
-
Hi Alexey
fjgarciam made the question I was going to ask too. I have thermal images from H20T and from Mavic 2 Enterprise Advanced and, in both cases, are not recognized as R-JPEG images.
They are, indeed, mounted as multi camera images but both RGB and Thermal images appear as regular JPEG's.
Cheers
-
Hello,
If you could share a few R-JPEG images that are not automatically treated by Metashape as multi-camera system upload adding photos, it would help us to improve R-JPEG data support.
-
Hi Alexey
Just sent you a small set by email from Mavic 2 Enterprise Advanced.
Best Regards
-
Hello fjgarciam,
Only for flir format? non dji format?
Yes, currently only FLIR format for R-JPEG images is supported.
-
I know we can shuffle the old version to a new folder and install the new version in the standard folder, so that we can have both installed, for testing... Even the licence works without copying over any files now i think...
But could we have an option in the Installer, on the 'Destination Folder' page, so that it installs it in a separate folder, and has it's own preferences storage? Perhaps with with the option to import your production version preferences over during install.
It could be installed installed in the "C:\Program Files\Agisoft\Metashape 1.8.0 Build 13257 Pre-Release" folder, say... Automatically including the version and build number perhaps. And including these versions/build number in the Start Menu shortcuts, etc. So dead easy to find, differentiate from our production version we have installed.
Later, it'd also be easy to cleanly uninstall these test versions from Windows' Apps & Features list.
Then when the new version is formerly released, we just install that as normal, and it updates our installation as it usually does.
We could then have as many old and newer versions installed as we wanted, really easily, making testing and benchmarking dead easy too.
-
Hi!
This dataset (https://drive.google.com/drive/folders/1xdqbvDTD2uIzJqweTw0BJ-gd8nHvxEG8?usp=sharing) was working before but is now crashing in build 13257 when generating a dense cloud.
Steps:
Import all photos.
Calibrate cameras - Spherical
Align cameras
Build dense cloud
Crash!
-
Hello noodle,
The issue will be fixed in the next 1.8.0 pre-release update, thank you for reporting and sharing the problematic dataset.
-
I know that it's still very early pre-release but here's a first test. Same settings with Mesh generated from depthmaps:
1.7.4
(https://i.postimg.cc/ncMpgWjc/Screenshot-2021-09-20-171415.jpg)
1.8
(https://i.postimg.cc/xj69sQd4/Screenshot-2021-09-20-171442.jpg)
1.7.4
(https://i.postimg.cc/gk5mWK8y/Screenshot-2021-09-20-171522.jpg)
1.8
(https://i.postimg.cc/brV8dxMh/Screenshot-2021-09-20-171540.jpg)
On the plus side this issue which was introduced with 1.7 seems to have been mitigated:
1.7.4
(https://i.postimg.cc/nhydpX3w/Screenshot-2021-09-20-172549.jpg)
1.8
(https://i.postimg.cc/mgzdvYyH/Screenshot-2021-09-20-172701.jpg)
Mak
Just wanted to point out that things are even worst in version 1.8.0 build 13354 :(
Mak
-
Yeah, i'm not getting very good results either. Is there a tweak to revert back to depthmap version from 1.7.3?
edit: I will upload a test dataset and email it to support.
edit2: email sent
-
This version (1.8.0 build 13354) crashes while importing a points file in .las format, exported from a dense point cloud created by itself but segmented in tiles. Prior versions work without any problem. Submitted the crash report using the build in crash reporter app.
Best regards
-
Hello Renato,
The issue with the point cloud import will be fixed in the next update. Thank you for the reporting and submitting the report.
-
Accidentally ran a batch in 1.8.0 the build (trying to figure out what went wrong exactly)
EDIT: looks like that happened at the mesh generation stage, i retried that part using the 1.8.0 depth maps and it worked ok in 1.7.3. So it is not all wasted :)
-
Hello all,
We have published a pre-release update (build 13597) which includes additional depth maps de-noising filter step. Would be interesting to see, if you observe considerable improvement in the mesh reconstruction for the projects, where the issues have been observed in the previous pre-release versions. (depth maps should be re-calculated in the new version).
-
Hello support,
in latest 1.8 pre release, I notice that shape object no longer has type and vertices attributes... Is this permanent? It is concerning as I have quite a few scripts using shape.vertices.....
see attached screen copy
-
Hello,
When I load a DTM and shape in Viewer, they do not display while they are perfectly viewable in Global mapper,
see following
-
Hello Paul,
For shape properties you should now access shape.geometry (actually, should be available in 1.7 API as well).
For example, to add a point shape:
point = chunk.shapes.addShape()
point.geometry = Metashape.Geometry.Point(Metashape.Vector([x,y,z]))
-
I have a project with 7 chunks. I ran the Batch Process - Build Mesh to process all the chunks. When it was finished it had only built part of the mesh for several of the chunks:
(https://digital-mapping.net/forums/Agisoft/version_180.jpg)
Version 1.8.0 build 13354
-
When I load a DTM and shape in Viewer, they do not display while they are perfectly viewable in Global mapper,
As for the DEM/Orthomosaic display in Agisoft Viewer, they can be only shown if Show Basemap layer option is enabled and these rasters are georeferenced in the coordinate system that can be transformed to WGS84.
-
Hello bisenberger,
Are there any disabled cameras in the chunk? Please also check, if the depth maps thumbnails show adequate information in the Photos pane or Photo view mode for all the cameras, or whether some depth maps are completely black?
P.S. the latest pre-release build number is 13597.
-
Hello Paul,
For shape properties you should now access shape.geometry (actually, should be available in 1.7 API as well).
For example, to add a point shape:
point = chunk.shapes.addShape()
point.geometry = Metashape.Geometry.Point(Metashape.Vector([x,y,z]))
Hi Alexey,
thanks for this info. However, I have many scripts where I access the existing shape vertices with shape.vertices attribute. Now in 1.8, how do I access these vertices coordinates?
-
Hello bisenberger,
Are there any disabled cameras in the chunk? Please also check, if the depth maps thumbnails show adequate information in the Photos pane or Photo view mode for all the cameras, or whether some depth maps are completely black?
P.S. the latest pre-release build number is 13597.
All cameras are enabled.
I had already uninstalled 1.8.0 before I seen your reply. I re-installed version 1.7.5, but get this message:
(https://digital-mapping.net/forums/Agisoft/error_message.jpg)
I will try installing the latest 1.8.0 build and check the depth maps.
-
Hello Paul,
You can access the coordinates of the shape vertices via shape.geometry.coordinates.
-
Hello Bill,
To avoid VCRUNTIME140_1.dll error you may need to download and install Microsoft Visual Studio Redistributable Package.
-
When I load a DTM and shape in Viewer, they do not display while they are perfectly viewable in Global mapper,
As for the DEM/Orthomosaic display in Agisoft Viewer, they can be only shown if Show Basemap layer option is enabled and these rasters are georeferenced in the coordinate system that can be transformed to WGS84.
Thanks again,
now shape file does display in Viewer but not DTM.... both were exported from Metashape in WGS84/UTM 10N
see fllowing....
-
Hello bisenberger,
Are there any disabled cameras in the chunk? Please also check, if the depth maps thumbnails show adequate information in the Photos pane or Photo view mode for all the cameras, or whether some depth maps are completely black?
P.S. the latest pre-release build number is 13597.
Here is a screen grab of the depth maps for one of the chunks that has an incomplete mesh:
(https://digital-mapping.net/forums/Agisoft/depth_maps.jpg)
-
Hello Paul,
You can access the coordinates of the shape vertices via shape.geometry.coordinates.
OK,
but the structure of shape.geometry.coordinates is different from shape.vertices as seen in:
shape
Out[6]: 2021-11-30 07:35:48 <Shape '1'>
shape.vertices
Out[7]: 2021-11-30 07:35:56 Shape.Vertices([Vector([683479.6620444183, 5618402.325884395, 349.5898219660391]), Vector([683484.8314030222, 5618401.859365016, 349.46954668612625]), Vector([683484.0623676204, 5618394.528231934, 349.49292624746954]), Vector([683478.7677394338, 5618395.108553094, 349.59959399146913]), Vector([683479.6620444183, 5618402.325884395, 349.5898219660391])])
shape.geometry.coordinates
Out[8]: 2021-11-30 07:36:54
2021-11-30 07:36:54 [[Vector([683479.6620444183, 5618402.325884395, 349.5898219660391]),
2021-11-30 07:36:54 Vector([683484.8314030222, 5618401.859365016, 349.46954668612625]),
2021-11-30 07:36:54 Vector([683484.0623676204, 5618394.528231934, 349.49292624746954]),
2021-11-30 07:36:54 Vector([683478.7677394338, 5618395.108553094, 349.59959399146913]),
2021-11-30 07:36:54 Vector([683479.6620444183, 5618402.325884395, 349.5898219660391])]]
shape.vertices[0]
Out[9]: 2021-11-30 07:38:22 Vector([683479.6620444183, 5618402.325884395, 349.5898219660391])
shape.geometry.coordinates[0]
Out[10]: 2021-11-30 07:38:51
2021-11-30 07:38:51 [Vector([683479.6620444183, 5618402.325884395, 349.5898219660391]),
2021-11-30 07:38:51 Vector([683484.8314030222, 5618401.859365016, 349.46954668612625]),
2021-11-30 07:38:51 Vector([683484.0623676204, 5618394.528231934, 349.49292624746954]),
2021-11-30 07:38:51 Vector([683478.7677394338, 5618395.108553094, 349.59959399146913]),
2021-11-30 07:38:51 Vector([683479.6620444183, 5618402.325884395, 349.5898219660391])]
How to access first vertex coordinates in shape.geometry.coordinates?
-
Hello Paul,
shapes in 1.8 version may be represented by the collection of multiple elements and will appear as list of lists via API. So if there's only single shape, then to access the coordinates of the Nth vertex you should use:
shape.geometry.coordinates[0][N]
-
I just ran two tests with build 13597 and I'm happy to report that the blobby mess on the underside of the geometry has been almost entirely fixed. It's still not quite as clean as 1.7.3, but it's MUCH better than any version from 1.7.4 to the previous beta. In my first test the geometry was 100% clean with no blobs. In my second test I had some very tiny blobs, but they were disconnected from the main geometry and far away, so very easy to clean up in Blender. I have attached a screenshot showing this.
I did see one error in the Log during Texture map generation: "Failed to initialize context. Reason: VK Error: VkResult is "ERROR_INITIALIZATION_FAILED" at line 66". However, it seemed to have created the texture just fine despite this error message. I have uploaded a screenshot of that as well.
The last observation I have that it is now showing both [GPU1] and [GPU2] in the log. With 1.7.3 it only showed one GPU on my MacBook Pro M1 Max. In the Log you can see that it says "Found 1 GPUs", so I'm not sure why it's acting like there are 2. Is that normal?
This is the first version since 1.7.3 that I think I will be able to use, so thanks for the great work!
-Brian
-
Are these scripts compatible with 1.8.0?
https://github.com/agisoft-llc/metashape-scripts
-
Hello Paul,
shapes in 1.8 version may be represented by the collection of multiple elements and will appear as list of lists via API. So if there's only single shape, then to access the coordinates of the Nth vertex you should use:
shape.geometry.coordinates[0][N]
Thanks again Alexey.
But it seems you could still keep the old shape vertices object though deprecated and that would avoid a lot of script adaptation, no?
-
Another question for Alexey,
in case of shape with attached markers we use to access them through shape.vertex_ids. Now in 1.8, how o we do that?
-
Hello Paulo,
If the shape has attached markers to its vertices (shape.is_attached == True), then the list of shape.geometry.coordinates will contain the keys of the related markers (marker.key).
-
I know that it's still very early pre-release but here's a first test. Same settings with Mesh generated from depthmaps:
1.7.4
(https://i.postimg.cc/ncMpgWjc/Screenshot-2021-09-20-171415.jpg)
1.8
(https://i.postimg.cc/xj69sQd4/Screenshot-2021-09-20-171442.jpg)
1.7.4
(https://i.postimg.cc/gk5mWK8y/Screenshot-2021-09-20-171522.jpg)
1.8
(https://i.postimg.cc/brV8dxMh/Screenshot-2021-09-20-171540.jpg)
On the plus side this issue which was introduced with 1.7 seems to have been mitigated:
1.7.4
(https://i.postimg.cc/nhydpX3w/Screenshot-2021-09-20-172549.jpg)
1.8
(https://i.postimg.cc/mgzdvYyH/Screenshot-2021-09-20-172701.jpg)
Mak
Just wanted to point out that things are even worst in version 1.8.0 build 13354 :(
Mak
Version 13597 is unfortunately still not as good as 1.7.x in terms of reconstruction in certain cases..
(https://i.postimg.cc/Kv5mq1qw/Screenshot-2021-11-30-170242.jpg)
but is way better at eliminating (or simply not creating) all the messy geometry inside (and outside) the models (finally!).
1.7.x
(https://i.postimg.cc/q7xSsFk3/Screenshot-2021-11-30-171619.jpg)
1.8
(https://i.postimg.cc/FK4wJvQv/Screenshot-2021-11-30-171642.jpg)
Mak
-
Second example looks good!
I will be doing some testing as well :)
-
Second example looks good!
I will be doing some testing as well :)
Well..I took a closer look at the 1.7 & 1.8 results all over the mesh and unfortunately in some places 1.8 is worst than 1.7.x..its really random 😪
Mak
-
Are these scripts compatible with 1.8.0?
https://github.com/agisoft-llc/metashape-scripts
I changed compatible_major_version = "1.8" in bounding_box_to_coordinate_system.py script, and it works. :)
-
Are these scripts compatible with 1.8.0?
https://github.com/agisoft-llc/metashape-scripts
I adapted the footprints_to_shapes.py for 1.8 and set a pull request....
-
I have a project with 7 chunks. I ran the Batch Process - Build Mesh to process all the chunks. When it was finished it had only built part of the mesh for several of the chunks:
(https://digital-mapping.net/forums/Agisoft/version_180.jpg)
Version 1.8.0 build 13354
I ran Batch Process - Build Mesh on 3 of the 7 chunks. It did not complete 2 it of them. If I run Workflow - Build Mesh... the entire model is created (it is able to reuse depth maps). Version 1.80 build 13597.
(https://digital-mapping.net/forums/Agisoft/batch_no_finish.jpg)
-
Oops! not sure how that double post happened.
-
Hello Bill,
Can you please provide the screenshot of Build Mesh task parameters from the Batch Process dialog?
-
Hello Bill,
Can you please provide the screenshot of Build Mesh task parameters from the Batch Process dialog?
Here you go Alexey:
(https://digital-mapping.net/forums/Agisoft/build_mesh.jpg)
The first time I ran it had it set to process all.
-
Hello Bill,
Can you please provide the screenshot of Build Mesh task parameters from the Batch Process dialog?
Here you go Alexey:
(https://digital-mapping.net/forums/Agisoft/build_mesh.jpg)
The first time I ran it had it set to process all.
Try to set Reuse Depth Maps to No.
-
So far I'm having good luck with this beta. There's still a lot of blobby gibberish being generated in most of my scans, but it's usually far enough away from the real geometry and not connected to it that it's fairly easy to remove. Still, 1.7.3 was the last version that generated totally clean meshes that didn't need any cleanup (for my application). The native Apple Silicon support is nice, and there's definitely a speed boost with this build. No crashes yet which is promising :)
-Brian
-
Well, this scan had 70 cameras and came out quite blobby on the underside. Some of the blobs are connected to the main geometry so it's not usable. This is the first failure I've had with this new beta version. I'll have to go back to 1.7.3 to get this one to work properly.
-Brian
-
I just had my first system freeze with the new beta. This time I had Metashape running a batch and during the depth map generation process I launched Blender. As soon as Blender came up and showed me it's window the system froze. In previous versions of Metashape the depth map process would cause my Mac Pro and MacBook Pro to randomly freeze up even if I was doing nothing. However, I have noticed a pattern that if I try to use Blender at the same time it seems to cause these freezes to happen far more often. Perhaps both applications are fighting for GPU resources? I'm not sure why that would cause the Mac to entirely freeze up and require a hard reboot though. I can use Blender while Metashape is working in the background about 90% of the time, but sometimes it seems to trigger these freezes. Blender, of course, never causes a system freeze on it's own - only when Metashape is doing Depth Map generation in the background.
-
I also had a crash after waking up my screen just now (during dense cloud generation), did not happen with prior versions.
-
Strange things happening with face count:
(https://digital-mapping.net/forums/Agisoft/ultra_medium.jpg)
-
I'm trying to build dem using the python API and it works fine with metashape 1.7 but not with 1.8. I couldn't find anything that relates to this in change log. I've attached logs for a small dataset that throws Empty dem error.
-
Strange things happening with face count:
(https://digital-mapping.net/forums/Agisoft/ultra_medium.jpg)
Do you see the entire mesh on the low polycount result? I've had some meshes clipped almost completely (like 10% of the geometry left)
-
Strange things happening with face count:
(https://digital-mapping.net/forums/Agisoft/ultra_medium.jpg)
Do you see the entire mesh on the low polycount result? I've had some meshes clipped almost completely (like 10% of the geometry left)
Yes, the whole mesh was visible on the Medium quality.
I re-installed version 1.7.4 and ran the same batch Process - Build Mesh, on the same data set I had issues with in version 1.8.0. With version 1.7.4 the Batch Process completed all 7 chunks without issue.
-
Version 1.8.0 (build 13597) mesh better results than 1.7.4 on this model:
(https://digital-mapping.net/forums/Agisoft/mesh_compare.jpg)
-
Version 1.8.0 (build 13597) mesh better results than 1.7.4 on this model:
(https://digital-mapping.net/forums/Agisoft/mesh_compare.jpg)
Yes it's a total crap-shoot. Some are better and others are worst. :o :'( I can't find any logic as to why it's acting like that.
Mak
-
I ran Batch Process - Build Mesh on 2 chunks. On one of the chunks it created mesh outside the bounding box.
(https://digital-mapping.net/forums/Agisoft/geom_bbox.jpg)
-
Hello Bill,
I cannot reproduce the problem on our side, can you please send this chunk as PSZ project to support@agisoft.com?
-
Hello Bill,
I cannot reproduce the problem on our side, can you please send this chunk as PSZ project to support@agisoft.com?
I just zipped up the whole project and emailed a link since it was so small. Let me know if you have problems downloading.
-
Hello Bill,
Thank you for sending the project, it helped to find the bug in Batch Process, it will be fixed in the next version update.
-
I don't know if this is the same issue as the Batch bug reported above, but I have noticed that if I am processing multiple Chunks in a single Batch, the first Chuck works fine, but the meshes on the remaining Chunks tend to be malformed. For example, attached are two screenshots of Chunk #2. The first shows the malformed mesh that was created during the Batch. However, if I simply build the Mesh again manually I get the correct results as shown in the 2nd screenshot.
In this particular instance I had 4 chunks in the Batch. The first was fine, but the other 3 were completely malformed.
-
Also, I get error messages and failures on some Chunks during a batch. In that last example, the 4th Chunk will always fail to generate a mesh when using Batch. I get "Error: Operation not applicable" and "Error: Null model" towards the end of the batch on that. But as always, just doing it by hand works perfectly fine. It only fails during a Batch.
-
Also, I get error messages and failures on some Chunks during a batch. In that last example, the 4th Chunk will always fail to generate a mesh when using Batch. I get "Error: Operation not applicable" and "Error: Null model" towards the end of the batch on that. But as always, just doing it by hand works perfectly fine. It only fails during a Batch.
Been getting the same, first model is good but the latter ones are partially or completely empty. This seems to do only with the Mesh Generation step as the depth maps produced are fine.
-
Hello,
Can you please check, if the issue with the Batch Process mesh generation is resolved in the latest pre-release build (13706)?
-
Hello,
Can you please check, if the issue with the Batch Process mesh generation is resolved in the latest pre-release build (13706)?
Don't know about Batch mode yet but this build did fix a crash I had with Depth Map generation step. 8)
-
Last night I did a test with 3 chunks in a batch and it seemed to work properly. I’ll try more today to confirm.
-
The Changelog hasn't been updated for the latest build ?
Mak
-
Just did another small Batch test on a 2 Chunks and it seems to have worked correctly. The only annoying issue I'm still seeing is that if I leave the room my Mac will go to sleep after 15 minutes and this cause the Batch to pause. Metashape should be calling UpdateSystemActivity(OverallAct); every few seconds to prevent the Mac from going to sleep while it's working. Otherwise, I have to keep moving the mouse to prevent the Mac from dozing off and pausing Metashape.
-
So far the batch is working for me.
More testing later. :)
-
Ok, something I've never seen happen before just happened. I was creating a new Chunk from 64 JPEG files that I shot with my iPhone 13. For some reason it decided that these photos were multilayer files (I didn't think JPEGs could be multilayer), so it bought up a dialog box asking if I wanted to add 128 cameras to one chunk or create 64 chunks with 2 cameras each. Bizarre.
When I loaded them in each JPEG showed as the normal color photo along with a black and white version of it. I have no idea what that black and white version is.
So, then I ran 1.7.3 to see if the same thing happened there and it did not. It just loaded the JPEGs normally.
Any ideas?
-Brian
-
Hi bgreenstonae,
is this a camera with depth sensor? which stores depth map along with image....? Maybe look at following tutorial https://agisoft.freshdesk.com/support/solutions/articles/31000162212-smart-cameras-with-depth-sensor-data-processing to see if you can import thee special images using Import Laser Scans .... command.
-
Hi bgreenstonae,
is this a depth map camera? which stores depth map along with image....? Maybe look at following tutorial https://agisoft.freshdesk.com/support/solutions/articles/31000162212-smart-cameras-with-depth-sensor-data-processing to see if you can import thee special images using Import Laser Scans .... command.
Ah, this may be the case. I only recently got my iPhone 13 and most of my scans were from photos using the flash. This set of JPEGs were shot without a flash, so I guess it embedded some depth info in there.
But how do I tell Metashape to ignore that data? I didn't see a way to do anything other than import all of it and then manually delete all of the black & white depth images.
-
Try following the tutorial and using Import Laser scans to import the images to the chunk... You should get the rgb images along with the depth maps which can be useful for building model, dem,....
-
Try following the tutorial and using Import Laser scans to import the images to the chunk... You should get the rgb images along with the depth maps which can be useful for building model, dem,....
I'm not seeing an Import Laser Scan option in the File menu as that web page says. Looked everywhere and I don't see how to do it. Normally, I just drag & drop the folder of JPEGs onto the Chunk.
-Brian
-
What version are u using? Pro? or Standard...If Pro you should have Import Laser Scans... from File/Import menu....
However, the depth sensor applies only to the front camera (selfies)...If you used the back camera then it could be dual and taking 2 images per click....
-
I’m using Standard. Sounds like this could be a bug with it since there’s no way to prevent it from doing this.
-
Hello Brian,
Please send an example of such image to support@agisoft.com or publish a download link for such image here.
It looks more like you have R-JPEG images. In such case you should choose the first option and then remove every second image (depth data with [2] suffix in the camera label) in the chunk prior to the data processing.
-
Alexy,
It appears that *any* photo I take with an iPhone 13 with the flash *Off* causes this. I just took several random photos of an object as a quick test and it immediately gave me the same issue. Here are the test photos:
https://www.dropbox.com/s/rb9jiws07que2fs/iphone13.zip?dl=0
It seems that the iPhone 13 automatically embeds depth information in all JPEG's as long as the flash is turned off, and Metashape 1.9 Basic doesn't have a simple way to ignore that information.
Thanks,
-Brian
-
Hello Brian,
Thank you for sharing sample data, we'll check what can be done on this matter. It seems that this additional page in the image is related to a kind of gain map (meta data contains HDRGainMap tag).
-
Another new anomaly has started happening a few days ago. Suddenly, almost 100% of my scans have incorrect sparse clouds and meshes. What happens is I get an "echo" of the scan in the resulting data. Difficult to explain, and I have not been keeping the bad ones to show you, but next time it happens I'll upload a photo and samples.
After a few hours of experimenting I believe I have found the problem to be with Generic Preselection because as long as I have that OFF it works fine, but with it ON I'm getting these garbled scans almost every time. The strange thing is that this only suddenly started happening. I've always used Generic Preselection and it was working fine until maybe 4 days ago. Nothing else changed here - no change in other settings, and nothing particularly different about the photos I'm scanning.
Anyway, just a head's up on that. I'll send samples later.
-Brian
-
Here's an example of what I'm now getting almost 100% of the time when using Generic Preselection. I have uploaded the test files and the batch.xml here: https://www.dropbox.com/sh/vq9871ija9we1r4/AADT5ajSgePUN1A-e5X9xpNJa?dl=0
In the screenshots you can see what looks like a flap going diagonally thru the main part of the scan. Sometimes it's diagonal like that, while other times it's almost parallel. If I run this exact same test through 1.7.3 with Generic Preselection it works fine, but on 1.8 it does this. Turn off Generic Preselection and it comes out fine.
Like I said before, this just started doing it a few days ago. I never had this happen with 1.9 using this same Batch previously.
-
Hi Brian,
I processed your 55 image data set with latest 1.8 pre relaease using Generic preselection and did not get any problem withresulting 3d model see screen copy
I used following processing parameters:
Processing Parameters
General
Cameras 55
Aligned cameras 55
Markers 2
Scale bars 1
Coordinate system WGS 84 (EPSG::4326)
Rotation angles Yaw, Pitch, Roll
Point Cloud
Points 38,324 of 105,044
RMS reprojection error 0.137578 (0.446816 pix)
Max reprojection error 0.436883 (2.48347 pix)
Mean key point size 3.14795 pix
Point colors 3 bands, uint8
Key points No
Average tie point multiplicity 2.58216
Alignment parameters
Accuracy High
Generic preselection Yes
Reference preselection No
Key point limit 30,000
Key point limit per Mpx 1,500
Tie point limit 7,000
Exclude stationary tie points No
Guided image matching No
Adaptive camera model fitting Yes
Matching time 7 minutes 8 seconds
Matching memory usage 382.07 MB
Alignment time 2 minutes 7 seconds
Alignment memory usage 140.56 MB
Optimization parameters
Parameters f, cx, cy, k1, k2, p1, p2
Adaptive camera model fitting No
Optimization time 8 seconds
Date created 2021:12:26 21:21:23
Software version 1.8.0.13736
File size 7.39 MB
Depth Maps
Count 55
Depth maps generation parameters
Quality Medium
Filtering mode Mild
Max neighbors 16
Processing time 23 minutes 2 seconds
File size 51.64 MB
Model
Faces 53,745
Vertices 27,403
Vertex colors 3 bands, uint8
Texture 8,196 x 8,196, 4 bands, uint8
Depth maps generation parameters
Quality Medium
Filtering mode Mild
Max neighbors 16
Processing time 23 minutes 2 seconds
Reconstruction parameters
Surface type Arbitrary
Source data Depth maps
Interpolation Enabled
Strict volumetric masks No
Processing time 2 minutes 14 seconds
Memory usage 742.79 MB
Texturing parameters
Mapping mode Generic
Blending mode Mosaic
Texture size 8,196
Enable hole filling Yes
Enable ghosting filter No
UV mapping time 1 minutes 18 seconds
UV mapping memory usage 150.70 MB
Blending time 5 minutes 12 seconds
Blending memory usage 4.06 GB
Date created 2021:12:26 22:25:30
Software version 1.8.0.13736
File size 77.86 MB
System
Software name Agisoft Metashape Professional
Software version 1.8.0 build 13736
OS Windows 64 bit
RAM 15.95 GB
CPU Intel(R) Core(TM) i7-3630QM CPU @ 2.40GHz
GPU(s) GeForce GTX 670MX
Note that you used a digital zoom ratio of 2. For best photogrammetric processing, it is advised to not use or turn off digital zoom during image capture...
Update: the correct model was produced without batch processing. With batch process, I do get a completely distorted model as u did, See 2nd attachement. So definitely there is a problem with batch process in 1.8....it seems to be at the texture generation phase as the solid model seems Ok.
3rd attachment shows model with distances from Camera center to image center projection to model (red lines). It does seem that a digital zoom of 2 was used...
-
Thanks, Paul. Actually there's no digital Zoom. That 2x is a bug that I reported a while back. It seems if you use any of the Zoom lenses on an iPhone it gets reported as 2x (even though my iPhone 13 is 3x).
It's interesting that it worked for you because like I said before, everything was working fine until just a few days ago. It's like something broke that's causing this, but I don't know what. I suppose I could delete everything and reinstall it, however, I'm not entirely sure where Metashape stores all of its settings and such.
-Brian
-
Yes Brian,
it seems to be a bug at the Build Texture phase of batch process as shown in my previous post 2nd attachement....as my first model (correct) was produced without using batch....
By the way, what was about the average distance to your object while taking photos?
-
I think most shots were maybe 2-3’ away. Why?
-
I am trying to check if there was really a zoom factor looking at the distances calculated after alignment and scaling with cm scale I get average distance of about 19 cm ....
-
I think those test photos were from an iPhone X. Not sure if he used the 2x or normal lens. If it’s the normal wide angle lens then 19cm is about right.
-
Hello ,
I have found that in 1.8 pre release, when I place shapes with labels in the chunk and save project. Then when I reopen the project, the shapes appear but without their labels.
In first scrren, I show 2 linestrings placed with their label equal to the line length.
When I save and reopen project, same 2 linestrings appear without their labels (2nd screen copy).
The shape label is however well present inside the doc.xml from the shapes.zip compressed folder from the <project_name>.files folder structure as seen in:
</shape>
<shape id="1" group_id="0" type="polyline" label="0.486 m" has_z="true">
<vertices>-1.1511491242687065e+02 3.5940266001437223e+01 8.1961011157548512e+02 -1.1511491430060157e+02 3.5940268773995314e+01 8.1927352960266114e+02</vertices>
<meta>
<property name="Dist_CC_PP" value="0.48635831257152956"/>
<property name="Photo" value="20211220_214523866_iOS.jpeg"/>
</meta>
</shape>
-
Hello Paul,
Which pre-release build you are using and if you are assigning labels via Python API or manual input via GUI?
-
Hi Alexey,
I am using latest release 1.8.0 build 13736 and shapes were placed with Python API.... if I open same project in 1.7, labels are visible....see following
-
The new filtering functions have certainly helped with 98% of the giant blobs of extraneous geometry that earlier versions put underneath meshes, but the new version still generates a lot of this stuff in every scan. It's almost always tiny, disconnected blobs so they're easily removed in Blender, but still, they shouldn't be there, and in 1.7.3 and earlier you never get these.
In this screenshot from blender you can see all of the small blobs of geometry in orange.
-
I just ran 2 batches and both had pretty bad blobs on the underside. One in particular had them *really* bad - it's unusable geometry, so I'll have to process both of these in 1.7.3 for now.
-
chunk.model.cropSelection() executed from python deletes entire Undo history in the latest 1.8.0 build with the new Undo list.
-
I just ran 2 batches and both had pretty bad blobs on the underside. One in particular had them *really* bad - it's unusable geometry, so I'll have to process both of these in 1.7.3 for now.
Hello Brian,
We need a project data to reproduce the problem, ideally camera alignment results and source images, so that we could run the depth maps based mesh reconstruction procedure on our side.
As for the "tiny blobs", you can remove them using Main Menu -> Gradual Selection -> Connected Components criterion, but a sample project would be also helpful for additional investigation.
-
Unfortunately, there’s no way to do Gradual Selection in a batch :(
-
Sorry for always being the bearer of bad news, but it seems every day I encounter a new issue with 1.8 that works perfectly in 1.7.3.
Here is a new issue that came up yesterday on several objects that I tried processing from one customer. First some explanation what we are doing: We make 3D printed display bases. The customer makes an impression mold of the object they want a base made for. Then they photograph the mold and upload the photos to us where we process them in Metashape and build their display base in Blender. That's why in all of my samples you see a box of sand and a ruler. The ruler is for scale, and the box of sand is the impression mold the customer makes.
In this case, the customer used a featureless white box to hold the sand (I don't know why his photos are all tinted pink). Metashape obviously cannot properly scan the blank white surface, so that geometry is never reproduced in any version of Metashape. No big deal since we don't care about the box anyway. However, in 1.9 the area of the box that didn't scan is causing the ruler nearby to have the texture map incorrectly generated. In these photos you can see that it came out perfectly in 1.7.3, but with 1.8 there is some incorrect texture overlapping the ruler.
I have uploaded the entire project with a Chunk done in 1.9 and a Chunk done with 1.7.3 so you can easily compare. I have also included the Batch file I used in each version. The Batches are identical.
http://gofile.me/61Yhd/Kea9kcby7
-
Sorry, but another bug to report. I normally do Orthophoto texture map generation at 8192x8192 for my scans. Today I decided to try a Generic texture instead, and it froze up 73% of the way in. The only way out is to Force-Quit Metashape. I tried a 2nd time and the same thing happened but at 74%.
Of course when I do it with 1.7.3 it works perfectly. Only freezes up on 1.8 :(
I have uploaded the test project here: http://gofile.me/61Yhd/xzM4z3TlE
-
In this case, the customer used a featureless white box to hold the sand (I don't know why his photos are all tinted pink). Metashape obviously cannot properly scan the blank white surface, so that geometry is never reproduced in any version of Metashape. No big deal since we don't care about the box anyway. However, in 1.9 the area of the box that didn't scan is causing the ruler nearby to have the texture map incorrectly generated. In these photos you can see that it came out perfectly in 1.7.3, but with 1.8 there is some incorrect texture overlapping the ruler.
Hello Brian,
Projection plane for the orthophoto texture mapping is selected differently for these models, since the distribution of the mesh faces is different. For the 1.8.0 project the orthophoto plane is slightly tilted relatively to the table's surface, therefore the average blending mode mixes the colors from the images of the ruler and the box in certain area.
As for the second issue - freezing of the Generic mapping mode, we are currently working on the fix.
-
Thanks, Alexey! I've always wondered how the orothophoto plane is determined. How does Metashape determine where the Z-axis is to project the texture along? I've done about 500 of these things, and previously it always managed to correctly project straight down the object's z-axis. This is the first time it hasn't, and I don't know what is different that would cause it. Also, why did it project correctly in 1.7.3?
-Brian
-
Hello Brian,
I have cropped a part of the model from 1.8 chunk and then rebuild the texture using Orthophoto mapping option and it produced the expected result.
The mesh in 1.7 chunk is bigger (due to bigger bounding box), so the distribution of mesh faces is different and it allows for proper selection of the orthophoto mapping projection plane. If you try to build the texture in 1.8 version using the model from 1.7, you should get similar result, as you have in 1.7.
-
Alexey, I do see that the mesh generated by 1.7 is a little bit larger (less crop) than 1.8. That is interesting that 1.8 cropped it in more than 1.7. Any reason why it would be different? I've had scans cropped pretty tight on 1.7 before and never had a problem with the projection mapping, however. Is there anything I can do on my end to make this come out correctly on 1.8?
Thanks!
-
Hi, just installed 1.8 on my second machine to process DNG taken with iphone13Promax and got pinkish texture.
Went back to 1.7.5 and all is fine.
Any idea what i did wrong and when I can update Metashape on my two computers?
-
Hold on...So 1.8 is now released (out of pre-release/Beta) even though its depth-maps based reconstruction is arguably worse than 1.7.x ? 😑🤔
Mak
-
Yep, Mesh from Depth Maps in 1.8 is still nowhere near as good as it was in 1.7.3, but the release version is definitely better than 1.7.4 thru earlier 1.8 betas. I keep trying to use 1.8, but inevitably I have to go back to 1.7.3 to get usable results.
-
Well, this is interesting. I just re-ran three chunks that failed miserably with the last 1.8 beta. I ran them with the release version of 1.8 and they all came out perfect. Same exact settings as I ran with the beta. The only difference is that when I did these with the beta I had multiple chunks running in the batch. This time I just did one chunk at a time with the same batch script. I will have to check later if the problem was caused by having multiple chunks, or if that’s just coincidence and something major got fixed in the release version of 1.8. Either way this was a very surprising result, because the difference between what I got a couple days ago and what I just got now is night and day.
-
Hello,
Mask defocus area seems to be broken in released 1.8
Edit : I have the same issue in 1.7.6
-
Hello all,
I am trying to export an orthomosaic using the new JPEG XL format and get an error that it cannot write image...
see attached log html
Any help will be welcome,