Agisoft Metashape
Agisoft Metashape => General => Topic started by: LFSantosgeo on November 28, 2017, 07:41:18 PM
-
Hello!
I just made some tests with Photoscan when generating Sparse Cloud:
Workflow 1:
1. Add chunk
2. Add photos (155 photos)
3. Convert from WGS84 to my desirable one (mine is SIRGAS2000 UTM)
4. Align with highest accuracy and standard tie and key points limit (using 80.000/8.000) = gives 153 aligned cameras.
5. Import markers (*xml Agisoft Markers) [previously done with the same photos getting the lowest error possible]
6. Uncheck cameras and check the proper markers
7. Update & Optimize
8. Read the reference pane
This gives me control points error as 0.006623 and check points error as 0.031124.
Workflow 2:
1. Add chunk
2. Add photos (the same 155 photos)
3. Convert coordinates reference
4. Add markers (import)
5. Uncheck all photos in the reference pane
6. Align with the same previously parameters (highest/80,000/8,000) = gives 151 aligned cameras.
7. Update & Optimize
8. Read the results in the reference pane
This workflow 2 gives: control points error as 0.006235 and check points error as 0.030192.
So my questions are (despite quite small differences):
A) Alignment: does it gives different alignment every time I run Align photos...?
B) Referencing: does it matters the order to put or am I doing it wrong?
C) Shouldn´t give the same numbers of aligned cameras and error even though different ordered steps workflows?
The non-aligned cameras are those acquired with high vegetation (on both workflows) that I was already expecting difficult in the alignment.
Thanks in advance!
PS. My workflow was updated on post bellow!
-
Further more from the order of alignment and markers importation to the chunks presented in the last post I have another (newbie) query about the referencing with GCP in Agisoft.
What about the difference among referencing only with GCPs and referecing GCP+photos checked? That gives a really small error difference (for control and check points) after updating and optimizing but still present in the total error. As expected the small error comes with the only GCP referenced cloud.
Could someone comment on it?
-
Hi lfsantosgeo,
One of my previous posts which explains an 'optimised processing workflow" might be quite useful for you:
http://www.agisoft.com/forum/index.php?topic=7851.msg37494#msg37494
Regarding photos checked/uncheck in reference pane. If you have chosen appropriate accuracy settings for your camera, then you can leave them checked. PhotoScan's default value for camera accuracy is 10m. If you used a UAV with RTK/PPK capabilities that can measure camera locations at cm accuracy, then you would adjust these settings.
If you have a series of highly accurate ground control points (e.g., measured by RTK GPS, cm-accuracy), then you can probably uncheck all photos in the reference pane because they won't add much to the general accuracy because their locations are not very well constrained due to the relatively inaccurate GPS from your drone (e.g., from a DJI Inspire 1/2 or Phantom 4). Also note that while X/Y coordinates of DJI drones are quite accurate (around 1-4m), the Z accuracy can be quite far off due to the way it is recorded by the drone (altitude above point of take off, not absolute altitude above MSL).
To sum it up. All things checked in the reference pane (markers, photos, scale bars) are used by Photoscan to optimize the geolocation/position and scale of the sparse point cloud (tie points) and all associated files 'downstream' such as the dense point cloud, mesh, orthomosaic and DEM.
I hope that helps. All the best.
Regards,
SAV
-
One of my previous posts which explains an 'optimised processing workflow" might be quite useful for you:
http://www.agisoft.com/forum/index.php?topic=7851.msg37494#msg37494
Thank you for your reply SAV! Really appreciated!
So let me update my workflow with your post recommendations (seen above):
1. Add chunk;
2. Add photography (outliers already removed -- too blurred, from take off and landing);
3. Convert photos coordinates (WGS84) to the GCP coordinate system (my case SIRGAS 2000;
3.1 Some ways to improve the GPS from photos (http://www.agisoft.com/forum/index.php?topic=4986.msg40010#msg40010)
4. Estimate image quality and disable all photos bellow 0.70;
5. When using python you can compensate for rolling shutter before alignment (http://www.agisoft.com/forum/index.php?topic=8124.msg39976#msg39976)
6. Photos alignment (highest quality, adaptative camera model fitting, generic + reference pair preselection);
7. Import markers (*xml Agisoft markers previously done with the photos) - that gives the markers photo's positioning;
7.1 Can be done locating the markers in the photos (filter by markers);
8. Update, transform;
9. Uncheck the photos in the reference pane (in case of good GCP coverage) and uncheck some markers for check point error along with control points error;
10. Sparse point cloud filtering (Gradual selection > Reprojection errors [bellow 1.0] - Reconstruction Uncertainty);
11. Adjust bounding box to the sparse cloud;
12. Compensate rolling shutter + Optimize camera alignment;
13. Build dense cloud (ultra high quality + depth filtering disabled);
Regarding photos checked/uncheck in reference pane. If you have chosen appropriate accuracy settings for your camera, then you can leave them checked. PhotoScan's default value for camera accuracy is 10m. If you used a UAV with RTK/PPK capabilities that can measure camera locations at cm accuracy, then you would adjust these settings.
My photography is based on the DJI Phantom 4. So I guess the accuracy of 10m is ok. Also considering what you said about about the DJI's vertical accuracy. RTK/PPK not enabled just PPK GCP for me.
If you have a series of highly accurate ground control points (e.g., measured by RTK GPS, cm-accuracy), then you can probably uncheck all photos in the reference pane because they won't add much to the general accuracy because their locations are not very well constrained due to the relatively inaccurate GPS from your drone (e.g., from a DJI Inspire 1/2 or Phantom 4). Also note that while X/Y coordinates of DJI drones are quite accurate (around 1-4m), the Z accuracy can be quite far off due to the way it is recorded by the drone (altitude above point of take off, not absolute altitude above MSL).
Good to know! Another point on unchecking all cameras and leaving just the GCPs.
To sum it up. All things checked in the reference pane (markers, photos, scale bars) are used by Photoscan to optimize the geolocation/position and scale of the sparse point cloud (tie points) and all associated files 'downstream' such as the dense point cloud, mesh, orthomosaic and DEM.
So it's not a good thing if you have good quality GCP (accuracy and geographic distribution) to leave the photos checked as I guessed.
Thank you SAV!
-
Hi lfsantosgeo,
All DJI Phantom 4 series drones have just a standard (single frequency) GPS. It is neither RTK nor PPK.
The issue with UAV imagery from DJI drones is that the EXIF file records the relative altitude above point of take off, not altitude above mean sea level, as you would use as a height datum for your GCPs. Therefore, X/Y coordinates recorded in the EXIF information for your images are quite accurate, but Z might vary a lot. Just make sure that you add proper accuracy values for X/Y/Z or that you set the camera accuracy to a quite large value.
If you enter a very large camera accuracy value (e.g., 30m instead of the default 10m), then you can probably leave all images checked because their location won't have any big influence anymore. It will also allow you to still use PAIR PRESELECTION REFERENCE in the ALIGN PHOTO step. Remember that Photoscan is simply trying to minimize the total error based on all the reference information and the associated accuracy values. For example, if GCPs have 3cm accuracy and your photos have an accuracy of 30m, then the GCPs will be weighted 1000 times higher than your camera locations.
Regards,
SAV
-
I just have updated my workflow on my last post. Open to comments...