Using EMLID GPS GCP with Zephyr

[dtmcnamara]

I am waiting for my trial code to show up in my email but until then I wanted to see if anyone else is using GPS GCP in their photogrammetry work and if so if they have a guide to follow. We have been using Pix4D for some time now and they just recently changed their cloud processing model so we are now trying to find an alternate software package to do what we need to do. Our workflow is as follows:

Create a point cloud with anywhere from 50-1500 photos
Import GPS GCP and align them to the marked locations on the point cloud
Create top down ortho photo (being able to clean this ortho photo is also nice but I think only Pix4D does this)
Load laser scan data from Faro scanner and align all data

I know we can easily create the point cloud, and load the laser data in, but I haven't found much information about loading GCP in and this is a must for our work flow. ANy help would be appreciated, likewise I will post up any findings in here once I get the key and start to pay around with the software

[Andrea Alessi]

Hi dtmcnamara,

you can add control points from images (preferred solution) or directly in 3D.

Once you have defined your control points, you can define the 3D constraints either by text input or by text file import.

You can find a tutorial for the control points https://www.3dflow.net/technology/do...es-3df-zephyr/

Or you can find our videotutorial (part 1 & 2) here https://www.youtube.com/watch?v=nQozgfnKBM8 and https://www.youtube.com/watch?v=HZtcJOZOstU

Let me know if you have any specific questions

[dtmcnamara]

Thanks Andrea, im sure I will have a handful of questions once I dive into the software. One question someone over here brought up is when we import the point cloud data into the work area with the photogrammetry point cloud, how are each clouds being used? Is it going to just try and merge the two together, or is there a process where it pulls the data from one cloud and color from anther and merges them together?

[Andrea Alessi]

External points clouds (regardless of their origin) are initially unstructured (meaning, they have been generated with no information from the current cameras).

It's up to you to do the registration of those clouds (a rough alignment followed by the ICP fine alignment). Once you've done so, the clouds are aligned but they are still unstructured. You can then right click a cloud and make it structured (which will generate visibility information etc. from all the cameras). Once the cloud has been made structured, it will be treated normally by all in-zephyr tools - you can merge it, edit, extract mesh, etc regardless of their origin as long as they are structured.

You can find a videotutorial on this topic here https://www.youtube.com/watch?v=oT0CK2Vcu_E

Happy to help feel free to ask any questions anytime!

[dtmcnamara]

Hey Andrea Alessi in the ICP Parameters when doing fine registration there is a field for MAX DISTANCE POINT REMOVAL. Is this the maximum distance that point cloud will move when trying to fine register it to the reference cloud?

[Andrea Alessi]

Hi dtmcnamara,

no, the max distance point removal is a research parameter - this means that points will look for their correspondence/neighboring only within that distance (per each iteration). Points above that threshold will not be considered outliers for that iteration.

There isn't a "maximum distance that the point cloud will move" value.

[dtmcnamara]

Thanks.

So im working on a project we have done in the past to get a feel for the work flow and I have done the following.

Pulled in 10 Faro scans already aligned in Scene and merged them together in a project
Pulled in 450 photos into a project and created a Dense point clound
Picked 6 GCP on both projects and saved them
Merged the photo project with the laser project and preformed the ICP registration

Now im trying to import our GPS data into the scene. In the project we have spray painted marks on the ground that line up with the GPS data. I converted our CSV file to a TXT file that is similar to one of the tutorial files you guys have posted, but when I try to import it it looks like this. Im guessing our project coordinate system is off. Where can I check what is being used? I honestly assumed it would auto pull from the Faro file, but I guess thats what I get for not checking it.

[Andrea Alessi]

hard to tell without having a proper look, but fromt he screenshot it simply look that X/Y are inverted. When importing from file, right click and select "invert X/Y" coordinate if this is the case

[dtmcnamara]

I think I know what the issue is, Im using a demo version of Pro and not Aerial. Should I be seeing different options in the tool bar with Aerial that I cannot see with the Pro version?

[Andrea Alessi]

kind of - aerial also has the feature to select and convert to different reference systems. If you have however your control points and constraint in a certain reference, you can use them but it will not be able to convert it e.g. from WGS84 to another reference system. I can happily send you a trial for zephyr aerial if you want to double check, but it shouldn't be the issue.

[dtmcnamara]

I just got a message from Giacomo and get an Aerial trial license. Ill let you know if if things work

[dtmcnamara]

Awesome, so it looks like we are getting somewhere. Now my only issue now is trying to find a quick way to match all of the GPS data to the aerial. Ive got the points imported into the aerial point cloud but it looks like I am needing to manually edit each individual GCP. I know in Pix4D we were able to match one on either end of the project and a few in the middle and we could re-optimize the project. This would then bring the project up to the GPS data and make aligning the other data so much quicker since everything should almost be in the exact location. Is there any way to do this in 3DF? Right now I am doing the following

Double clicking on the GCP that I want to work on
Click on Edit Control Point
Change visibility to All Photos, since the GCP and aerial are off
Find the GCP in the photos and add it then close out of the editor.

This should the new location for that point is now on the map.Not sure if I need to re-run a set once this is done or what.

[dtmcnamara]

Alright, getting even further now. Just like Pix4D you can select a few points then re-optimize things. 3DF just calls it Align Model with 3D constraints. Reading over the documentation they suggest at least 3 points to do this so I grabbed 5 points, 2 on either side and one in the middle and got the model below.

[dtmcnamara]

With that done, I was able to click on all the other GPS GCP and easily make them on the aerial point cloud and the re-run the Align Model with 3D constraints process.

Once question I do have, when selecting points for GCP, once you select at least 2 points there is a button that says Automatic matching, what does that do? Will it auto find other pictures with the point in it? Doe sit auto select the points? I pressed it a few times and didnt really notice any change.

Additionally when running and Aerial, GPS GCP and Laser date what order do you recommend processing everything? Currently I have

Create aerial project
Add GPS GCP to aerial project
Create a a few manual GCP to align projects together
Create Laser project
Create a a few manual GCP to align projects together
Merge aerial project with GPS to laser using the laser as the reference

Not sure if this is the best way to do everything.