RealityScan for GIS/3D Modeling

RealityScan Team,

I have been using RealityScan side by side WebODM for several months now. I think it’s fantastic software for GIS professionals. The placement of the data is somewhat off as many have stated on the RealityScan Community. I would like to help improve your software, because the major strength are all the tools available in the current version of the software.

I have been working with AutoCAD and ArcGIS for over 30 years. I started with ArcView 1.0 when it was just a viewer. I have been working with 3D data conversion for almost 20 years, drone data conversion for close to 10 years, and flying drones for close to 7 years. Drone data conversion has been great on my Linux box since 2019 using WebODM. I recently found your software and thought, this is it.

I use a GIS grade GNSS receiver, RoboDot Touch, (6–12-inch accuracy) to capture drone data using PPK or post processing. I use OPUS to correct my benchmark and correct the images with the image shift tool from Robota. I process the drone data with WebODM and download all assets in NAD83 2011 UTM 14 for my area. I create a GIS project file in QGIS or ArcGIS Pro and add all the data, orthophoto, DEM, DSM, DTM, and GCPs. All data lines up within 6-inches or so and that’s great for GIS data capture work.

When I process the same data in RealityScan I notice some differences. The data typically sits around 2-meters south of the actual corrected benchmark. I did a test where I can manually shift the aerial to the benchmark using ArcGIS Pro. After the manual shift of the orthophoto only, I also notice that the X- and Y-axis are somewhat stretched or warped inward. I also review all photos before processing to make sure the EXIF XYaccuracy is .01 and Zaccuracy is .02 meters. The controller is connected to the GNSS receiver to make this happen during flight.

Some initial thoughts. North should always point up. Maybe, the spinning of true north is throwing the final location off somehow. The world file for any software I have ever used is always the top left corner of a square image output. I notice RealityScan places the anchor point or world file in the top right corner of a rotated image.

When GCPs are not used, there should be a checkbox that states: use EXIF georeference data 100% for spatial placement.

Please let me know how I can help.

Regards,

Gerry

BLACK DOT - RealityScan world file placement

BLUE DOT - JPEG or TFW world file placement (ArcGIS, QGIS, AutoCAD Map 3D, other)

Hey Gerry_DVG, thank you for posting in our community.
I will ensure that your feedback is shared with the team. If you would like to discuss this further, please contact us here an select technical inqiry and we will get back to you.

Thank you

Hi,

Thank you for your response.

I plan to use a Local NTRIP to capture my next project to avoid any offset of imagery. In other words, the images will be accurately georeferenced during flight, so a direct import into RealityScan and WebODM. Then, I will test to see if the export of data is better geolocated. I will try with and without GCPs and control points.

QGIS is open-source GIS and free. I have used the software for over 15 years and still believe it’s better at exporting georeferenced orthophotos and elevation data. Yes, better than Esri. Maybe, your team can find some good insight if you download and take a look at the raster functionality and export capabilities. Just a thought.

I will keep you posted and put some info together to add to the link you provided.

Regards,

Gerry