Well, imagine shooting 10 cm/px GSD, and your GCP center to 10 cm radius differs by 3-4 cm. You have 1 px for this, so there must be some kind of averaging happening.
I guess we could generalize from the, and say the smaller the GSD value, the more critical it is to have a even/flat target. So, if doing volume measurements, this would be a good idea! If doing i.e. a glacierarea of multiple square km at 10 cm/px GSD, then it’s doesn’t matter much.
Usually we do UAV mapping from 70 m altitude in polar regions. With Phantom 4 Pro v 2.0 GSD on such altitude is really small. Here is a screenshot from PhotoscanPro with the GCP:
1.91 px/cm in that case, and 50x50 cm targets?
If the Photoscan can se enough texture on the targets, I guess it shouldn’t be a problem.
I usually do my volume measurements flights at 25-30 meters, which gives me a GSD at around 0.8
I sometimes use plasterboard (gypsum) cut into 50x50 squares when I don’t want to retrieve them after I’ve flown the drone.
Nice! I like the handle!
Does that impact auto-detection btw?
How do you manage your targets?
0 voters
Bucket lids are great as @michaelL mentioned. Here are mine I made in a pinch for large sites, I leave these there. If soil conditions are favorable these are pretty solid with washer and nail/spike.
In most other cases I use my trusty pack of targets like these on smaller sites, but I don’t leave them. Hole in the center for survey rod is nice.
Color/finish will affect precise placement of GCP in processing software. You want as flat/matte as possible. At peak solar hours the bucket targets began to “degrade” as too much light was reflected off the target itself. Next time, flat/matte white bucket lid with flat/matte back paint would be better. I would also clear the vegetation in a radius around target to bare dirt.
In Pix4d, notice how much more white the sensor picked up than actually on target.
Very nice. I also have some blue ones from Lowes that I paint white stripes on for light conditions. Where do you get the fabric ones from? Very portable and I don’t have much room left in my truck with the drones and survey gear. 
I know most people are going to disagree with me, but hitting the exact center in my testing has proven unnecessary. I have tested the same map multiple times - hitting the exact center twice and quickly clicking through them twice and there is as much deviation between the pairs as there is across the methods. A 1.5-2% difference on a 50ac site makes absolutely no difference in my processes. With the way that GCP’s are used in processing and how RMSE’s are calculated +/- 1.5 inches is nothing.
Hence the interest in the poll and the fact that my GCP’s remain unless there are areas with absolutely no safe haven in the construction LOC. Then I just paint an X. You might also think about putting your targets on a small concrete base for longer duration projects.
Indeed, very portable nice to fold up or wad up and get on with it. Lots of options on Amazon but the ones I showed are here: https://www.amazon.com/Control-Targets-Mapping-Surveying-Eyelets/dp/B07GGMM8LW
Can you explain this a little more? As in it doesn’t really matter if you hit it exact or not as most of your measurements off the dataset just need to be relatively accurate?
Thanks for the link. Ordering them now!
If I have a 1ft bucket lid with no defined center point I can hit within an inch or so just with the crosshair and post-click circle that just-so-happens to be the same size as the bucket lid at a certain zoom level. I feel that if your targets are too big then it is best to zoom way in. Next, if I am hitting that surveyed point within an inch and the point was located with a repeatable error of about 0.05’ the error that missing the mark by a couple of inches introduces is not even measurable on individual objects, the ortho or on-screen. Similar to my statements previously about localization to survey control, the photogrammetry processing is localizing and distributing the error across all your GCP’s so the image corrections are not just related to the GCP closest to them. An example is that is I have a 100,000cy stockpile and I miss by 2% then 2,000cy doesn’t mean a whole lot at the end of the day. Is usually misses the truck or gets spread out anyway.
Nice idea, searching a bit more, I also found these: https://www.ebay.com/itm/UAV-Ground-Control-Points-GCPs-Aerial-Targets-for-Aerial-Mapping-Surveying/333066443019
Nice as well! In construction I have a problem with black/white because of shadows and materials. One of the reasons I use ACDSee to edit my photos because I consistently have issue with exposure flyiing over trees and then something like a concrete pad. When I set my GCP’s I am frequently limited to areas in the shadows so orange and a combination of orange and blue work best for our construction sites.
These are the others I use. A little more expensive, but you can break them down and they are durable. Use the individually or make a quad with another color. I have had a dozer run over one and it survived.
I don’t think so but I don’t know for sure. I use Pix4d that doesn’t do autodetect. From what I’ve seen from Dronedeploy it seems it’s been trained to pick up any GCP
Yes, DroneDeploy is hard at work with their AI. They now have a suggested target configuration and it will grow as time goes on.
https://support.dronedeploy.com/docs/automated-gcp-tagging
I am on the Customer Advocacy Council so please submit your suggestions!
Bumping the poll…
There’s a lot of good inspiration in this thread! I love seeing everyone’s innovative solutions - particularly the lightweight and cost-effective solutions. 
I map in coastal areas, so I use a mix of water-based spray paint for hard features, and printed 1.5 x 1.5 FT vinyl banners with tent stakes. I run a tether line from one of the stakes to the banner so I can buy some extra time if the tide starts coming back in. That was a development after losing a couple of earlier banners.
We started numbering spray-painted points and vinyl banners to make it easier to match GPS reference points to photos in post. That’s been a huge timesaver.
This is fantastic! Hopefully GCP location AI continues to improve and the initial search will be right at it. One of the things I enjoy most about these forums is when people collaborate and get to see use-cases way out of their norm. I’m in construction and have learned a ton of tips from agriculture, marine and environmental users!
I do many flights in urban areas so I have the luxury of being able to place any white mark on bitumen and have it very identifiable from most altitudes I fly. I use chalk paint so as to not permanently deface the landscape. I’ve also been pushing the contrast in order to figure out the lightest surfaces I can still have a visible white mark on. This has resulted in me marking light coloured concrete sidewalks, and dirt. Circle and dot target very easy to find quickly when browsing the aerial photos. If I don’t have hard surfaces I use a roughly 1’x1’ black painted craft wood board with approx 30-40mm white dot in the middle.
Nice! I primarily work in fabrication areas where the paved areas are covered soil or mud, and where machinery works constantly (volume measurements), so I can’t make permanent markings (put down 40 cm screws instead. Then I lay my GCP’s centered on top of these for each job, so I don’t have to remeasure each time.
The client paid for the installation-time and hardware, and we then went from 4-5 hours onsite to 1 hour onsite. Win/win for both.
Do you find you have lack of precision not having a cross or small do when placing your rover rod?
I don’t think I can identify the mark in photoscan any better than I can centre the GPS pole, probably a bit worse. So there’s nothing to gain for me in squeezing out a few more mm; I’m not wanting for more horizontal accuracy.
