1
General / Re: Visibility-consisten mesh generation vs the "new" depth maps approach
« on: April 30, 2019, 09:26:17 AM »
Hi,
Is there any follow up to this issue,
Cheers,
Stu.
Is there any follow up to this issue,
Cheers,
Stu.
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.
Pages: [1] 2
2
Bug Reports / Re: Arbitrary Mesh from Imported Point Cloud - Heavily Decimated
« on: April 30, 2019, 06:10:15 AM »
Hi Alexey,
I just tried a custom face count of 10 million faces and it yielded exactly the same face count as I am getting from the High setting.
Custom Face Count = 10,000,000
Actual Face Count = 159,211
High Face Count = 4,773,856
Actual Face Count = 159,211
Cheers,
Stu
I just tried a custom face count of 10 million faces and it yielded exactly the same face count as I am getting from the High setting.
Custom Face Count = 10,000,000
Actual Face Count = 159,211
High Face Count = 4,773,856
Actual Face Count = 159,211
Cheers,
Stu
3
Bug Reports / Arbitrary Mesh from Imported Point Cloud - Heavily Decimated
« on: April 29, 2019, 03:39:14 AM »
Hi,
I have a very large 3D modelling project for a mine site survey. I have broken the project into a large number of sub-chunks, and generated dense clouds.
The dense clouds have been exported, where I have used a Poisson filter to perform a "smart" decimation of the point cloud. The objective is to significantly reduce the point cloud density then re import the point cloud to Metashape for meshing etc in a reduced time frame.
I have successfully exported, filtered and re imported a dense cloud. However when I generate a mesh from the Dense Cloud based on an Arbitrary surface type, with a High target face count, I am getting a very low resolution result, with 159K faces when the target was 4.7M faces. If I generate the mesh using an Height Field surface type, I get the correct face count of 4.7M faces.
I have attached screen shots of the resulting meshes & the input point cloud.
Cheers,
Stu.
I have a very large 3D modelling project for a mine site survey. I have broken the project into a large number of sub-chunks, and generated dense clouds.
The dense clouds have been exported, where I have used a Poisson filter to perform a "smart" decimation of the point cloud. The objective is to significantly reduce the point cloud density then re import the point cloud to Metashape for meshing etc in a reduced time frame.
I have successfully exported, filtered and re imported a dense cloud. However when I generate a mesh from the Dense Cloud based on an Arbitrary surface type, with a High target face count, I am getting a very low resolution result, with 159K faces when the target was 4.7M faces. If I generate the mesh using an Height Field surface type, I get the correct face count of 4.7M faces.
I have attached screen shots of the resulting meshes & the input point cloud.
Cheers,
Stu.
4
General / Re: Visibility-consisten mesh generation vs the "new" depth maps approach
« on: April 04, 2019, 11:06:22 AM »
Hi Alexey,
Sorry for the delay, production has been very busy. I too have emailed a link to my data-set.
I have included the project files/folder with raw frames. Within the project there are 2 3D Models.
One has been created from the Dense Cloud and the other using the Depth Maps.
To the east there are some overhead conveyors that have picked up in the Dense Cloud 3D model, but missing in the Depth Map 3D model.
The point cloud and depth maps were both processed at ultrahigh/mild settings. Mesh settings were both high.
Cheers,
Stu.
Sorry for the delay, production has been very busy. I too have emailed a link to my data-set.
I have included the project files/folder with raw frames. Within the project there are 2 3D Models.
One has been created from the Dense Cloud and the other using the Depth Maps.
To the east there are some overhead conveyors that have picked up in the Dense Cloud 3D model, but missing in the Depth Map 3D model.
The point cloud and depth maps were both processed at ultrahigh/mild settings. Mesh settings were both high.
Cheers,
Stu.
5
General / Re: Visibility-consisten mesh generation vs the "new" depth maps approach
« on: March 30, 2019, 05:12:57 AM »
Hi,
I concur with this, I am seeing poorer 3D modelling directly from depth maps. I am processing an airborne imagery of an industrial facility with overhead conveyor structures. Processing a point cloud in ultra-fine/mild filtering reconstructs these overheard objects, but with same settings for depthmap based 3D model, very poorly reconstructs these objects, with large gaps in these features. I am happy to provide the source data-set to Agisoft to assist development.
Cheers,
Stu.
I concur with this, I am seeing poorer 3D modelling directly from depth maps. I am processing an airborne imagery of an industrial facility with overhead conveyor structures. Processing a point cloud in ultra-fine/mild filtering reconstructs these overheard objects, but with same settings for depthmap based 3D model, very poorly reconstructs these objects, with large gaps in these features. I am happy to provide the source data-set to Agisoft to assist development.
Cheers,
Stu.
6
Feature Requests / Adaptive Tiling/Sub-Chunk based on RAM.
« on: March 23, 2019, 02:32:46 AM »
Hi,
It would be a great feature for large projects if Metashape could "internally" subdivide a chunk in smaller tiles for sequential processing, based on the available RAM.
This would avoid the user having to guess how many sub-chunks to break a large project down into.
Cheers,
Stu.
It would be a great feature for large projects if Metashape could "internally" subdivide a chunk in smaller tiles for sequential processing, based on the available RAM.
This would avoid the user having to guess how many sub-chunks to break a large project down into.
Cheers,
Stu.
7
Bug Reports / Merge Chunks - Mesh Merge not working
« on: March 18, 2019, 07:48:13 AM »
Hi,
Using Metashape Pro Build 7618, I have used the Split into Chucks Python script to split a chuck in 2 x 2 chunks. I have batched processed a 3D Model for each chunk, using high settings and 10,000,000 faces. The 3d Models within the chunks are fine. However when I try to merge the chunks together, ticking the Merge models checkbox, a merged chunk is produced very quickly, but no merged model is produced.
No error is reported in to console, only;
MergeChunks: merge_dense_clouds = off, merge_models = on.
Merging chunks....
I have tried to decimate the models down to 100,000 faces, but it makes no difference. I have attached a screenshot showing the 4 chucks displayed together and the empty "Merge Chunk"
Cheers,
Stu.
Using Metashape Pro Build 7618, I have used the Split into Chucks Python script to split a chuck in 2 x 2 chunks. I have batched processed a 3D Model for each chunk, using high settings and 10,000,000 faces. The 3d Models within the chunks are fine. However when I try to merge the chunks together, ticking the Merge models checkbox, a merged chunk is produced very quickly, but no merged model is produced.
No error is reported in to console, only;
MergeChunks: merge_dense_clouds = off, merge_models = on.
Merging chunks....
I have tried to decimate the models down to 100,000 faces, but it makes no difference. I have attached a screenshot showing the 4 chucks displayed together and the empty "Merge Chunk"
Cheers,
Stu.
8
Bug Reports / Re: Mesh from Depth Maps - False Surface
« on: March 14, 2019, 05:43:27 AM »
Hi Alexey,
Please find attached the link to the tiled model as requested,
http://astron-geodata-delivery.s3.amazonaws.com/Test_Model.zip
Cheers,
Stu
Please find attached the link to the tiled model as requested,
http://astron-geodata-delivery.s3.amazonaws.com/Test_Model.zip
Cheers,
Stu
9
Bug Reports / Re: Mesh from Depth Maps - False Surface
« on: March 07, 2019, 02:27:04 PM »
Hi Alexey,
Reducing the vertical extents of the bounding box resolved the issue with the mesh generation. However generation of a Tiled Model is also effected. The Tiled Model was also generated from Depth Maps. I have attached the profile views of the single file mesh and the tiled model for comparison.
Cheers,
Stu.
Reducing the vertical extents of the bounding box resolved the issue with the mesh generation. However generation of a Tiled Model is also effected. The Tiled Model was also generated from Depth Maps. I have attached the profile views of the single file mesh and the tiled model for comparison.
Cheers,
Stu.
10
Bug Reports / Re: Mesh from Depth Maps - False Surface
« on: March 06, 2019, 11:45:08 PM »
Hi Alexey,
The imagery as all Nadir, shot with a wide angle (20mm FL) lens and Sony a6000 camera. However we will be acquiring higher resolution nadir & oblique imagery in the near future. I downgraded to Metashape 1.5.0 (build 7492) and can confirm the issue is still present. I have adjusted the region bounding box height as suggested and will advise the outcome.
Cheers,
Stu.
The imagery as all Nadir, shot with a wide angle (20mm FL) lens and Sony a6000 camera. However we will be acquiring higher resolution nadir & oblique imagery in the near future. I downgraded to Metashape 1.5.0 (build 7492) and can confirm the issue is still present. I have adjusted the region bounding box height as suggested and will advise the outcome.
Cheers,
Stu.
11
Bug Reports / Mesh from Depth Maps - False Surface
« on: March 06, 2019, 11:25:55 AM »
Hi,
Using the latest build of Metashape, I am getting a false dark surface as per the attached image, when processing a mesh using depth maps as the datasource. As my project is urgent I reverted to Photoscan 1.4.4, no issues were found.
The mesh is a small region from a much larger project (~14300 photos). I extracted a low resolution dense cloud over the entire project area and no issues we detected in the area of this mesh. The project aligned well, with ppk camera locations and plenty of GCPs/Check points.
Cheers,
Stu.
Using the latest build of Metashape, I am getting a false dark surface as per the attached image, when processing a mesh using depth maps as the datasource. As my project is urgent I reverted to Photoscan 1.4.4, no issues were found.
The mesh is a small region from a much larger project (~14300 photos). I extracted a low resolution dense cloud over the entire project area and no issues we detected in the area of this mesh. The project aligned well, with ppk camera locations and plenty of GCPs/Check points.
Cheers,
Stu.
12
General / 4DMapper Upload Broken?
« on: February 22, 2019, 07:16:09 AM »
Hi,
I have generated an number of 3D Tiled models with Metashape (1.5.0 build 7492), and have attempted to upload them to 4DMapper. The upload appear to be successful, and the relevant project and 3D Tiles layer was created and no errors were reported. However the model does not load, there is just a black globe. I have provided a link to the dataset.
https://app.4dmapper.com/user/share/35bec1539cfb5b6f933a131a1f303c2de7107f6bb347f7ce
I also tried to upload a model from Metashape that I had previously uploaded using Photoscan and I got the same failure to load. It looks like something has changed/broken with Metashapes 4DMapper upload.
Cheers,
Stu.
I have generated an number of 3D Tiled models with Metashape (1.5.0 build 7492), and have attempted to upload them to 4DMapper. The upload appear to be successful, and the relevant project and 3D Tiles layer was created and no errors were reported. However the model does not load, there is just a black globe. I have provided a link to the dataset.
https://app.4dmapper.com/user/share/35bec1539cfb5b6f933a131a1f303c2de7107f6bb347f7ce
I also tried to upload a model from Metashape that I had previously uploaded using Photoscan and I got the same failure to load. It looks like something has changed/broken with Metashapes 4DMapper upload.
Cheers,
Stu.
13
General / Re: Cesium 3D Tiles - Poor Overview Geometry & Texture Resolution
« on: November 10, 2018, 03:44:13 AM »
Hi Alexey,
That worked well, i had even better results setting maximumScreenSpaceError equal to 1. The model looks great upon load and zooming in. If you zoom out further the model geometry does start to degrade, this may become an issue for loading very large or multiple models.
Thank you for pointing me in the direction of these additional parameters. I will experiment further and provide my results next week.
Cheers,
Stu.
That worked well, i had even better results setting maximumScreenSpaceError equal to 1. The model looks great upon load and zooming in. If you zoom out further the model geometry does start to degrade, this may become an issue for loading very large or multiple models.
Thank you for pointing me in the direction of these additional parameters. I will experiment further and provide my results next week.
Cheers,
Stu.
14
General / Re: Best way to deliver 3D model locally
« on: November 01, 2018, 08:41:04 AM »
Hi dersuperpro1337,
For a user to simple view a model, I find it simplest to generate it as a tiled model, output it in photoscan tiled format and simple have the user install Photoscan and use the free viewer. Very large models can be viewed this way.
Alternatively I have been trialing Cesium, which allows you to view a tiled model via a web page, but requires a web server (apache) to be installed. With only a few extra lines of code in the html, you can display models using this interface. You can also at camera paths to the code too.
But I am not getting not getting good results from Cesium tiles as per the below thread; The look poor when zoomed out, only once you zoom in to an intermediate zoom level the coorect geometry and texture is rendered.
http://www.agisoft.com/forum/index.php?topic=9861.0
For single file meshes, Blender can be used for fly-thoughs, whilst opensource it has a learning curve.
Cheers,
SP
For a user to simple view a model, I find it simplest to generate it as a tiled model, output it in photoscan tiled format and simple have the user install Photoscan and use the free viewer. Very large models can be viewed this way.
Alternatively I have been trialing Cesium, which allows you to view a tiled model via a web page, but requires a web server (apache) to be installed. With only a few extra lines of code in the html, you can display models using this interface. You can also at camera paths to the code too.
But I am not getting not getting good results from Cesium tiles as per the below thread; The look poor when zoomed out, only once you zoom in to an intermediate zoom level the coorect geometry and texture is rendered.
http://www.agisoft.com/forum/index.php?topic=9861.0
For single file meshes, Blender can be used for fly-thoughs, whilst opensource it has a learning curve.
Cheers,
SP
15
General / Re: Cesium 3D Tiles - Poor Overview Geometry & Texture Resolution
« on: October 25, 2018, 01:41:01 PM »
Hi,
Further to this issue the Cesium 3D tiles are not as sharp at maximum zoom as the Photoscan tiled model. The attached example on the left shows the Photoscan model the right is the Cesium model. It is most noticeable toward the bottom of the image.
Cheers,
Stu.
Further to this issue the Cesium 3D tiles are not as sharp at maximum zoom as the Photoscan tiled model. The attached example on the left shows the Photoscan model the right is the Cesium model. It is most noticeable toward the bottom of the image.
Cheers,
Stu.
Pages: [1] 2
