I’ve noticed that in general, the photo spacing for mapping missions on the Mavic 2 Pro is much more consistent than the Phantom 4 Pro. Also noticed the P4P V2 is more consistent than the original P4P. Seems DJI may be using a better GPS receiver in the M2P and P4P V2?
My question is, I’ve read studies (including DroneDeploy’s own article on it) that found volumetric calcs without ground control within 1-2.5% accurate (meaning 1-2.5% error) depending on which article you read.
My question is, for volumetric calcs of very small volumes, such as a square pit that is 10 feet deep and 20 X 20 foot footprint, you’d need to fly pretty low in my opinion to get enough overlapping photos for an accurate DTM which the calcs are performed from. So the photos are going to be taken just feet apart at very low speed.
I would guess the P4P would still pull it off but if the M2P and P4P V2 are getting better GPS accuracy (evident by the much more consistent photo spacing) that would indicate to me that the geotagging of the photos is going to be relatively more accurate hence wouldn’t the photogrammetry results be more accurate than the P4P, especially for such a small size area?
The Engineer gut in me says that these claims of 1-2% error in volume calcs are based on measuring much larger volumes (like a pile 100 feet in diameter and 15 ft tall, or a 10 ft deep excavation that is more like 100 X 100 ft rather than 10 x 10.
Also the question of rolling vs global shutter may come into play? I would assume DD compensates for rolling shutter like Pix4D does? But for this type of application is the P4P with global shutter still much preferred over the M2P rolling even if we assume the geotagging is better on the M2P. When I say better I’m really talking about the error in the RELATIVE distance between the coordinates of the photos being more accurate/consistent than the P4P.
I’ll bid that I can reconstruct a 20x20x10ft hole with 9 images. Anyone else think it can be done with less?
I think 3% is a more realistic number from what we have experienced and unless it was really large area covered GCP’s made no relevant improvement in the individual stockpile volumes. The Engineer in you should also know that you need to include shrink/swell into calculations if you want to even utter the word accurate. The materials industry works with tonnage.
From all of the modules that I have seen there is very little to no difference in the actual module. There are several other factors that feed into GNSS accuracy like firmware, multipath interference and antenna configuration that are more likely to explain any major differences in accuracy of an uncorrected GNSS system.
I think DD will only process with at least 30 images if I recall. Although we can use other software of course. The flight time between 9 and 30 images is insignificant though. Just have to fly lower on the smaller areas.
I asked the question to DD Support also. They seem to think the global shutter of the P4P is more important than the GPS accuracy of the M2P and P4P V2 (it must be the firmware in the V2? I’ll have to check if our P4P firmware needs to be updated – I see other other differences in the V2 vs non-V2 that would explain the more consistent photo spacing similar to the M2P.)
From DD Support: “the rolling shutter will have a tendency to create lower accuracy maps relative to global. We do apply some correcting for rolling shutter, similar to what you’ve linked from Pix, but here at DroneDeploy, we have still had more success with global shutter maps historically.”
There are a few things different about the v2.0 is h.265 encoding at 100Mbps (which noone uses) and an improved lens which contributes to the improved image quality. One or both of these factors may help the image spacing. Another thing, and maybe more important to the image spacing is the Occusync radio signal.
Not going to go buy a P4P V2 when the V1 I have is adequate for my use. Also going to check when the last firmware update for the P4P was. I don’t think I’ve updated it in a long time. From what I could find there was an update May 19 2020. I probably never got around to applying it. Doubt it makes the GPS any better.