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.

Messages - JMR

Pages: 1 ... 8 9 [10] 11 12 ... 32
General / Re: Draw rectangle polygon
« on: June 26, 2019, 04:21:24 PM »
draw it on cad and import to your project

Alignmen problems, I second that. Scroll right the ref pane table and check pixel errors for your photos. I you find images with errors greater than 1 pix, then you should check their alignment carefully. sometimes reset alignment of those poorly aligned photos and redoing alignment of the same, fixes problems. By the way, what camera/lens combo has been used? Could lens zoom have changed between shots?

General / Re: Combining imported scan data with photogrammetry
« on: June 26, 2019, 04:08:21 PM »
create markers in laser dense cloud and place them on photos, then update. Make sure source view in ref pane shows cooedinates in the tls reference system for the created markers.

General / Re: Adjust camera coordinates
« on: June 18, 2019, 06:34:31 PM »
To amend camera altitudes, a simple trick it is to export cameras in ref pane, then copy camera lines paste file to an excel sheet, modify altitudes by adding whatever, and then import back in the same fashion. However, if this constant "error" comes truely from an offset between the gnss antenna and camera nodal point, then you can set it more properly in the camera calibration dialog.


General / Re: Move marker(s) re-do the workflow again ?
« on: June 17, 2019, 08:04:38 PM »
A marker is a projection (2D) of a ground (3D) point on the image plane (the film or the image sensor). If you drag a marker in one or several photos from a to b and then you update, the entire project will chane a bit because the new projection forces the whole bundle block to shift, rotate and re-scale to best fit the new constraints in the solution.
So moving markers to line up an orto made of old photos with something (I guess with current ortho) is a bit of a weird approach. I would rather move or choose another ground point to pin the old photos to currently existing features... i mean it looks like you are trying to warp the orthomosaic rather than to do a seven parameters geometric transform (scale, rotation, translation).

General / Re: Move marker(s) re-do the workflow again ?
« on: June 17, 2019, 11:56:34 AM »
No, PRobert, a projection is a marker instance in a photo. If you move one, and the optimize, everything in your project will change this any product of the previous status becomes obsolete and requires to be rebuilt. If you just update, then what changes is only the external orientation, thus all products also become obsolete but only in scale and position. Mesh and dense cloud can be updated to the new external orientation of the chunk, but DEM and ORTHO need to be rebuilt as 3D rotation and shift in the source data, changes it all inside those products and they cannot be just updated.
by the way, what't the use of building mesh and texture in your case? just curious

General / Re: What is the best way to align photos with GCP?
« on: May 27, 2019, 07:31:52 PM »
Although 2 is strictly right, 3 works perfecly well in the very most of cases, unless your GCPs are very badly distributed and/or they are in a coordinates system that is incompatible with that of the cameras (that is the case when GCP are in a local system with totally diferent numbers than camera centers).
Normally, enabling/disabling cameras does not make any diference in EO but after the third decimal place (sub mm). Well that is assuming that accuracy settings are left as default by 0.005m for GCP and  10m for cameras.
Best regards

PD. re-align is pointless in my opinion.

check-points (unchecked ground control points) will play some role in the optimization as they are tie points placed by user, thus it is normal that project adjustment statistics change a bit if you opimize after having added check points. If you do not trust your hability to pick a point well, simply do not re-optimize.

General / Re: ultra high settings alignement
« on: May 07, 2019, 12:34:52 PM »
RTF  ;) page 23
Since tie point positions are estimated on the
basis of feature spots found on the source images, it may be meaningful to upscale a source photo
to accurately localize a tie point. However, Highest accuracy setting is recommended only for very
sharp image data and mostly for research purposes due to the corresponding processing being quite
time consuming.


General / Re: GCPs with no elevation value?
« on: May 06, 2019, 04:02:06 PM »
You can add GCP with XYZ using precise XY and guessed Z taking it from google earth, for example, then set accuracy accordingly, for example an accuracy of 0.005/10 would mean 5mm in horizontal positioning and 10 meters in vertical.
Also, if you are using a drone geo-tagged images, you can use 10/10/0.2 for camera accuacy. So the drone altitude measuered by the baro altimeter (which is often quite accurate in relative) will make your model to be properly leveled while GCP will put it into right XY


General / Re: MetaShape for determining spatial data
« on: April 04, 2019, 01:17:41 PM »
Hi there,
I read the Photoscan instructions in the manual, according to that I have the following question:
Could you please explain to me how to determine markers' coordinates (x, y, z) (20 markers) in Metashape? / is it possible to do it in Agisoft Metashape?

What I have done so far:

• loading photos into Agisoft Metashape
• removing unnecessary images;
• aligning photos;
after alignment
  • Import your ground control coordinates by ref pane import button, choose the right coordinate system according your survey data
  • Open a photo where a known point is to be found and zoom in to that point, then right click on the very point and use Place point command, choose from the list the one that is supposed to correspond
  • place the same point at least in three photos and do the same with all of your control points
  • update project by the update button in the ref pane toolbar. keep adding points instances and finally optimize

General / Re: Alignment problem - 1.5.2 build 7838
« on: April 04, 2019, 12:53:04 PM »
Q1. Do your flights have similar altitude records in goeotags? drop all photos in one project, hit keys 5 (ortho) and then 3 (side view) and see if the blue balls are more or less well placed in terms of altitude, i they are not, you can amend altitudes by export reference inputs (geotags), use excel to fix altitudes by using ground home elevation + flight altitude and import them again into ref pane... then all flights should align without problems with reference checked
Q2. Rather than using two psx projects you should have used separate chunks in same psx. However, you can still merge project by append command under file menu

good luck

General / Re: Camera Calibration and Distortion Plot
« on: March 08, 2019, 11:45:26 AM »
I guess the second graph looks weird because the vertical axis is unsigned, so try to imagine the second hump upside down and you will see it is not-so-odd. Also note that the vertical scale is rather smaller, decentering abs values are smaller in the second plot, so actual decentering distortion might be overestimated when using less parameters, so adding P3, P4 could actually make sense.

General / Re: Camera positions?
« on: March 06, 2019, 02:14:24 PM »
Two points can't define a reference system, you need at least 3. your model is not referenced but just scaled (note the [S] after chunk's name.
What you can do is to add a third point with given coordinates, that you can extract making use of the already well-established scale. calculate x,y,z coordinates of this third control point by asking distances to the two existing(*). and then update. You should get [R] after chunk's name.
(*) Add empty scale bars between the new point and the two existing and change reference pane view to "estimated values" mode.
Regards, Geobit

General / Re: DJI P4 RTK Workflow
« on: March 06, 2019, 01:09:07 PM »
3. Import gcp and changing the settings for coordinate system. For me in sweden its: Project system, SWEREF99. Camera system, wgs84. gcp system SWEREF99. I had only 1 gcp for testing the RTK drone.
(the gcp does not show up in right place, maybe 4m beside but the photos has accuracy 10-40mm)
4. Place the gcp. The accuracy for the gcp will be like 1m
5. Optimize the pictures with both controlpoints and photos checked.
6. Import the exif data (the accuracy for gcp are under 10mm after that)
After 3, you have the model driven by on-board coordinates. If the just imported check point seems to be out of its place... then something is wrong in the orientation.
If by "place" you mean just pinning the check point reprojections to right pixel coordinates... then the error of 1m proves the alignment is not sufficiently well constrained by camera locations alone.
After 5, (opimize with cameras and ground point checked) the new orientation is driven by camera geotags AND your only ground control point (thus it is not any more a check point)
6. import exifs again should change nothing, but will probably revert your coordinates system to WGS84 and require you to change this setting again.
So it is plainly normal that the GCP looks fine after all.
In my opinion, your workflow proves that adding 1 control point might help to constrain the external/internal orientation of the project... but I do not see in this method a proper checking procedure, and you should wonder if the 10mm ground accuracy is consistent all over the model.
Best regards

Pages: 1 ... 8 9 [10] 11 12 ... 32