Orthophotos of Wall Facades and other vertical surfaces

[matt07lx]

I use Photoscan in archaeology to document standing architecture and archaeological trenches. I have a question about the feasibility of using Photoscan to create orthophotos of wall facades or other vertical surfaces. Our GCPs are measured using a total station, and these allow us to create perfect "top down" orthophotos for archaeological plans, but is it possible to do the same for walls and the sides of trenches, allowing for the creation of archaeological section drawings? I know that it is possible to choose a planar projection using markers to define the horizontal and vertical axes for the orthophoto, but if these markers are manually placed onto the 3D model from Photoscan, how accurate is the resulting orthophoto? For instance, will the horizontal orientation of the orthophoto be parallel to the x and y axis from our GCPs, or will it follow the arbitrary horizontal plane between my two markers on the wall? I have the same question for the vertical axis - will it align along the z axis from the GCPs or the line between my markers? What if the wall in question was "slanted" and not perfectly vertical? I'd appreciate any input that anybody could provide.

[chrisd]

Post this question over on the CloudCompare forum, it can perform arbitrary unwrapping or projection so you might get some help there. Daniel is the developer and he is usually quick to respond.

[Alexey Pasumansky]

Hello matt07lx,

When the projection plane is defined by three markers it means that the plane contains all three points, so it wouldn't be necessarily vertical or horizontal surface. For more customization you can use python scripting, for example using one of the axis defined by the GCPs and another by the manually placed markers for the projection plane definition.
Markers that are used to define the projection plane do not require to have coordinates input.

Also Current View projection option is supported.

[chris]

Hello matt07lx,

have a look at this paper: https://www.academia.edu/12557650/The_Use_of_Structure_from_Motion_for_the_Documentation_of_Archaeological_Profiles._The_Introduction_of_a_first_workflow_to_creating_a_profile_drawing_with_Agisoft_Photoscan_Professional_QGIS_and_GeoTiffExaminer_-_The_English_Translation .

Greetings,
Chris

[matt07lx]

Many thanks to everyone for the suggestions.

Chris - I've read the article, but I'm afraid that I do not quite follow the workflow. How do you create a point that is exactly one meter vertically higher than another in Photoscan? This seems to be the issue that I was having since I can only place markers manually using the mouse. Is there a way to create a marker in a location if I know the coordinates I want? It's also unclear to me where the "profile nails" should be placed - in the profile? On the ground? Does this method require a total station in prismless mode? (If so, this isn't a possibility)

Alexey - This method sounds very intriguing. Are there any Python scripts available that would allow me to do this? Ideally, I would define the horizontal axis using the GCPs and the vertical axis using manually placed markers on the wall surface.

[chris]

Hi matt07lx!

first, you will need your Photoscan model to be georeferenced. You wrote that you have a total station, so even if it has no reflectorless-mode, that should not be a problem. Just mark four points (two at the top and two at the bottom plane of your profile as shown on fig.1 in the linked article) and measure them with your reflector. In Photoscan, you can add markers directly onto the mesh (right click on the mesh and choose  "Create marker" from the context menu). Place a first marker near the left border of your profile (corresponds to KS-U in the article) and a second one near the right border (KS-X) at about the same altitude. I then manually edit the altitude (Z-value) of KS-X to the altitude of KS-U in the reference pane (click into the corresponding field at the "Markers" section and enter new value). This step is not really necessary if you choose "Vertical axis" while building the orthophoto. You will then need to "construct" KS-Y. Right-click into the "Markers" section and create another marker. Copy/paste the easting and northing of KS-U to the new marker KS-Y and add the altitude (altitude of KS-U plus e.g. one meter).  Now you have the three points needed to build the orthophoto.

Greetings,
Chris

[matt07lx]

Hey Chris,

Thank you! I've tried this out and I believe I now have it working. I just wanted to clarify a couple steps to make sure that I am doing this correctly.

First, my models are not georeferenced, but use a local coordinate system based upon targets shot in using a total station. This doesn't seem to matter.

After I create the first two markers by right clicking onto the mesh (near the left and right side of the wall to be projected in the orthomosaic), I've noticed that the only way to view their coordinates is to click on the "View Estimated" icon in the Reference panel. The "View Source" icon has no values assigned to the markers unless they are inputted manually. When I've attempted to change these values, as you suggested to put "KS-U" and "KS-X" at the same altitude (Z-value), it seems to have no effect upon the location of the markers. They remain in the same location where I clicked on the mesh. Rather, I suspect that it is adding incorrect information to the model's scale and reference system, so I've decided to simply skip this step.

However, by right-clicking within the Reference panel to create the third marker (KS-Y), I am then able to use the x, y, and z coordinates of KS-U from the "View Estimated" to move this marker around, and make it exactly one meter higher than KS-U.

When I create the orthomosaic, I use "KS-U" --> "KS-X" for the horizontal axis, and "KS-U" --> "KS-Y" for the vertical axis, leaving the vertical axis box checked.

Matt

[Yoann Courtois]

Hi everybody !

I try to produce almost the same, which are orthomosaics from buiding walls in order to be mapped.
Thus my models are georeferenced using tacheometric suveyed points. I finally got my orthomasaics but i still have a problem:
When i try to export then, PhotoScan is asking me a coordinate system. So I inputed my systems by it failed and i get "Can't tranform coordinates".
I've tried with some other systems but it's always the same.
Is there a problem due to the verticality of my orthomosaic ?
Is there any solution to fix it, of even get my orthomosaic without georeferencement, which is currectly impossible because i HAVE to input a coordinates system.

Thanks for your help

Regards

[Alexey Pasumansky]

Hello Yoann,

If you have generated the orthomosaic in the custom planar projection, it should be exported in the same way. You cannot export the orthomosaic of the vertical wall in Geographic projection.

[Yoann Courtois]

Indeed I have generated a custom planar projection orthomasaic, using 3 markers.
But as you can see below I cannot export it in another way than geographic projection, as I need to input a global coordinate system. May you can help me ?
Regards

[Alexey Pasumansky]

Hello Yoann,

In this case you need to choose the default option, the orthomosaic should be exported like it is displayed in the Ortho view.

[Yoann Courtois]

Hello Alexey,

What is exactly "the default option" ? Actually I didn't find anything like that
I would like to get the orthomosaic and keep the scale, or even more the georeferencement, even if it's verticale.

Thanks for your help

[Alexey Pasumansky]

Default is the one you see, when open the Export Orthomosaic dialog. I think that the one on your screenshot (chunk coordinate system name, but without EPSG code).

And how do you imagine vertical georeferenced orthomosaic? Would there be many pixels with the same XY values then?

[Yoann Courtois]

As I said I cannot export without coordinates system / EPSG code:
Actually the "default" mode (parameters I get when open the Export Orthomosaic dialog) is RGF93 coordinates system.
Then if I try to export (without changing anything in the dialog box), I get this error message again: "Can't transform coordinates"

And about the vertical georeferencement, I was imagining it could be possible to saved (X,Z) coordinates in order to includ it in a verticale view, which seems not te be possible.
At least I would need firstly to get my orthoimage (which is for now impossible), and then save the scale due to GCPs.

UPDATE: Guess I had a bug... I finally can export my orthoimage, without changing anything compare to what I've done before... aha ! Thanks for your help Alexey

UPDATE.2: It seems one of the problem that cause the impossibility of orthomosaic export is the Outer Boundary Line. While one polygon is define as Outer Boundary, it's then impossible to export the orthoimage (cutting bug when it's not a classic flat orthoimage ?). If none are define it's ok. But the exported image present careless borders.

[Alexey Pasumansky]

Hello Yoann,

Can you re-save a project copy in PSZ format (PhotoScan Archive) with only results of alignment (you can remove mesh, dense cloud and depth maps to reduce the file size) and send it to support@agisoft.com?

Also specify, if we can reproduce the issue, if we use Current View option for the orthomosaic generation in this project?