Has anyone noticed rendering errors in orthoimagery generated with Trimble Business Center’s photogrammetry module?
My survey team has been performing around forty photo and/or lidar UAS missions per year, typically for landfill volumes. We’ve experimented with Pix4D, Site Scan, LP360, DJI Terra and now TBC Enterprise to process data and produce large high quality orthoimages and C3D surfaces. We’re a Trimble shop and so, begrudgingly at first, I’ve learned to use and now like TBC. However, no matter what procedure I’ve used, if there are elevated structures such as overhead conveyors, parked trailer trucks, buildings etc., the TBC orthos warp or checkerboard said structures. Bare earth seems to be rendered correctly. The same data run through Pix4D or Site Scan comes out crisp with no warping or checkerboarding. In all cases, the orthoimage matches incredibly well with my aerial targets or painted checks. While the point cloud data continues to awe and amaze me, the clients only see the pretty picture and I haven’t been able to generate anything in TBC that matches the quality of Pix4D or Site Scan. I've tried tagging the photos and just processing with RTK or PPK.
General flight and equipment parameters:
1. DJI M350 RTK Drone
2. Zenmuse L1 Sensor
3. RTK corrections from NC-VRS (quite robust as compared to most real time networks)
4. 180ft AGL with DEM used for Terrain Follow
5. 50% side and 70% forward overlap
6. 10.6 MPH
7. Aerial targets
8. Triple Return with Repetitive Scanning
9. Lawnmower flight pattern with occasion cross hatch.
Is it me or TBC or maybe a bit of both?
I'm sorry I can't address your questions but I do have one of my own. What DEM do you use and to what degree of terrain following does it allow. Does it recognize 1 foot differences 10' differences or 50' differences?
We've used DEMs from USGS and can get them free from the great state of North Carolina. They're a geo.tiff and must be in meters for to play nice with DJI.
Trusting the altitude indicator on my controller and guessing based on visual observations, I'd say it easily recognizes 10ft. I generally fly from atop the landfill and in the right light conditions can watch it ascend then descend. I did a 600acre open pit kaolin mine with deep pit and it was simply amazing how well it handled dropping into 200ft deep hole about 20 acres in area. We mostly fly the same sites and use DEMs created in house from previous missions now.
DJI has a bunch of quirks when using a DEM and flying AGL. DEMs over 100MB won't be accepted, all portions of your flight path must be within the boundary of the DEM. Occasionally, I'll want to fly from a low point within my AOI and DJI isn't happy with that. I don't fully understand why, but it seems like having a portion of DEM with a elevation higher than 400ft above the launch pad is a not acceptable. Flying from the highest elevation negates this.
Thanks for the info. The 400' ceiling above takeoff elevation is DJIs way of keeping you from violating airspace. They will have lower limits around airports some as low as 100' ceiling. Thanks for the quirks. Its much easier to find it here than learning it the hardway in the field.
Hi murphy, I don't think it is software problem! For me, the problem is those 70%-50% overlap. In height structures, overlap is getting lower and that causes defects and errors you talking about. I suggests to increase overlapping or make some additional perpendicular tracks to collect more information. Also you can try to fly at 230ft if GSD accuracy allows.
Hi murphy,
feel free to contact our support group ([email protected]) from TBC Photogrammetry and share your project and data, so we can take a closer look and check what causes your effect.
You can send in your PDF report, which would already be a good start to evaluate it.
We have a dedicated Trimble TBC Photogrammetry community group, where you can exchange your experiences and get help with TBC Photogrammetry-related questions.
Wish you a great New Year 2024,
Thomas