Lisa Williams

A little background: I fly a DJI Mavic 3 Enterprise and I've been doing recreational flying for about two years. But lately I've been really interested in the mapping and surveying side of things. I've taken a few online courses, watched probably way too many YouTube videos about photogrammetry, and finally convinced a friend who owns a small piece of rural property to let me practice on his land. About 12 acres, mix of open field and some tree cover on the western edge. Perfect for a first attempt, I figured. The Mission Planning Part I used DJI Pilot 2 for mission planning, which integrates pretty smoothly with the . Setting up the grid mission wasn't too bad once I figured out the interface. My big mistake number one was overlap settings. I had read that 70-75% front overlap and 60-65% side overlap is a reasonable starting point for most mapping work. I looked at that and thought, yeah but more is better right? So I cranked front overlap to 85% and side overlap to 80%. What happened? My 12-acre site turned into a mission that generated over 800 images. Processing that in Agisoft Metashape on my laptop took the better part of five hours and nearly crashed the thing twice. My friend's field did not need 800 photos. Somewhere around 400 probably would have done the job fine and saved me an afternoon of staring at a progress bar. Lesson learned: more overlap means more images means way more processing time and storage. There's a balance and I was nowhere near it. Ground Control Points — I Kind Of Sorta Did These This is the part I'm most embarrassed about. I knew ground control points were important. I'd watched the videos. I understood in theory that without GCPs your map might look great but the absolute accuracy is going to be off, sometimes by a lot. I set out five targets — just printed these black and white checkerboard squares I found a template for online — and I placed them around the field. What I did not do: use an actual survey-grade GPS to record their coordinates. I used my phone's GPS. My phone GPS. Which, in open sky conditions, is maybe accurate to three or four meters on a good day. So my orthomosaic looks visually really nice. But when I checked it against known reference points? There were some offsets that were... not great. I'm not going to put a number on it because I think I also made mistakes in how I imported the GCP coordinates into Metashape, so it's a compounding error situation. The point is, if you're doing this for anything that actually matters — anything where someone needs real-world accuracy — you need real-world accurate GCP coordinates. Phone GPS is not that. I've since been looking into renting a RTK GPS unit for my next attempt. Some survey equipment rental places do short-term rentals which makes it more accessible for someone like me who isn't doing this commercially yet. The Actual Flight The flight itself was the most straightforward part, weirdly. Once the mission was loaded it was pretty much hands-off. I launched, the drone went and did its thing in a nice back-and-forth lawnmower pattern, and landed itself. Total flight time was maybe 22 minutes. Wind was light, maybe 8 mph, which was fine. I flew at 80 meters AGL and set my speed to around 8 m/s. One thing I didn't anticipate: the tree line on the western edge. The automated mission doesn't really account for obstacles in any meaningful way on this hardware — it's not doing terrain following or anything like that. The trees are only about 10 meters tall so at 80 meters I had plenty of clearance, but it did make me think about how I'd handle terrain that's more variable. Something to dig into for future missions. I also forgot to check that my camera settings were locked down before the mission started. The drone adjusted exposure a couple times mid-flight as it crossed from the shadowed area near the trees back into full sun, and you can actually see slight brightness variations in some of the image tiles in the final orthomosaic. Going forward I'm setting manual exposure or at minimum using the auto-exposure lock feature before I start the grid. Processing in Metashape I went with Agisoft Metashape Standard because the price point is way more manageable than some of the alternatives for someone just learning. The workflow is roughly: import photos, align photos, build dense cloud, build mesh, build orthomosaic. Each step builds on the last and any errors kind of cascade. My photo alignment went well — Metashape aligned 792 of my 803 images which I thought was pretty good. The dense cloud looked really solid over the open field area. The tree canopy area was messier, which I've read is pretty typical. Vegetation with a lot of movement between shots is tough for photogrammetry. The orthomosaic export I did at full resolution, which was another mistake in terms of file size. The TIFF came out at around 4GB. I don't have a GIS setup sophisticated enough to really work with that efficiently yet. I've been reading about QGIS which is free and apparently handles large rasters reasonably well with the right settings, so that's my next thing to figure out. What I'd Do Differently Looking back, here's my honest list: Drop the overlap to something more reasonable, probably 75% front and 65% side, and see if the results are still acceptable Borrow or rent an actual RTK GPS for GCPs — this is non-negotiable if accuracy matters at all Lock camera exposure before starting the mission Export the orthomosaic at a more manageable resolution until I have a proper GIS workflow set up Fly on a fully overcast day if possible — even lighting across the whole scene apparently makes a real difference I've also started reading more about coordinate systems and datums because when I was trying to bring my orthomosaic into QGIS I ran into projection issues that I genuinely did not understand. That's a whole rabbit hole I'm just starting to go down. Overall Thoughts Honestly even with all the mistakes? I'm kind of hooked. There's something really satisfying about turning two hours of field work into a map you can actually measure things on and look at from above. Even my rough, GPS-phone-GCP orthomosaic shows my friend where his fence line sits relative to the tree canopy in a way he'd never really been able to visualize before. He seemed genuinely impressed, which made me feel better about the whole thing. I've got another practice mission planned for next month on a slightly larger site, about 25 acres. I'll write that one up too. Hoping to have the RTK situation sorted by then and maybe a better handle on the QGIS side of things. We'll see. If anyone else is starting out with mapping and has figured out the GCP workflow on a budget, I would genuinely love to hear how you're handling it. Drop something in the comments.