As far as gcp’s, what about the trimble catalyst, has anyone used this yet? Seems like an affordable start to mapping?
Learning about mapping
- Thread starter SWA Pilot
- Start date
PrudentialUAS
Member
Folks have already mentioned the whole "surveying" terminology matter, no need for my input there. BUT, I do think I can provide some advice to a new operator looking to get into the game, especially when it comes to generating deliverables that I can be confident in:
- Overlap: I prefer at least 75%
- Perimeter Path: On top of the double overpass in a crisscross pattern, I have found if you include perimeter images you get even better topographic data.
- Avoid Blurry Images: Have your drone stop at each point where it is taking an image. This actually also makes it so that I can have the drone flight speed between shots be faster than if I were having it take the images while under way.
- Software Trials: Oh, I'm going to get in trouble here...I'm a CHEAP son'***** and since I'm doing this still as a "side hustle" as the kids say, I was struggling to capture data AND process it all under one trial time period. Here's a hint (and its not just for UAV stuff): gmail will ignore any "." that is before the "@" of an email address. This helps when registering for trial software, because you can reuse the same email address multiple times. Example - if your email address is "[email protected]", you can register trial software (4) different times, using:
R.Perry
Well-Known Member
I have used Drone Deploy for the past three years, at least once a week. If I had the drone stop at each picture I would never get the mapping done. I have been flying the Inspire 2 averaging 28 to 32 mph, hatch pattern, 250 altitude and a 75/65 overlap and have never had a blurred image.
I normally didn't include perimeter images because the client wasn't interested. Most mapping was 160 acres.
I normally didn't include perimeter images because the client wasn't interested. Most mapping was 160 acres.
PrudentialUAS
Member
It depends on the camera on the Inspire 2:
The X5S has had issues with rolling shutter effects, but (I've heard/read) the X4S and X7 with their mechanical shutter provide better results in that regard.
I was erring on the side of caution.
I perform perimeter just because with any photogrammetry application, the more the better. I find the perimeter run to result in better ODM output, especially with regards to the DSM files for topographic deliverables.
Exactly
There's something quite wrong if you can't get sharp images at 20+ mph from 200-400 feet up.
PrudentialUAS
Member
Meh, not for me. It works and I don't have to worry about my outputs.
Usd it necessary? Not really, but I only want to fly once.
Are you only flying 5 acre sites?
There would be no way to fly stop, start and map average size sites.
In daylight with ISO 100, you should be looking at exposures of around 1/250th or faster at f5.6.
That's not going to result in any blurred images at normal flying speeds and heights.
I've flown many mapping missions and only once had to go back.
The issue wasn't blurred images due to speed, but the problems mapping with thick tree cover.
PrudentialUAS
Member
The amount of time it takes for the unit to stop and take an image is negligible.
I'm not really sure what an "average size" site is, is that an industry standard?
I'm not really here to argue over whose processes are better though.
I guess you must be flying tiny sites then.
One of my regular sites is 33 acres and involves about 3.7 miles of flying to shoot 300 images.
That's just not practical with stopping for each image, and it's not needed anyway.
There's a reason DD's default speed suggestion is >20 mph.
R.Perry
Well-Known Member
The 160 acre site averaged 1300 photos. The Inspire has the X5 camera with 15 mm lens, worked perfectly.
I've also done some maps with my Mavic with close to same numbers, all came out good.
I've also done some maps with my Mavic with close to same numbers, all came out good.
Are you guys using gcp’s? Do your clients use the tools in DroneDeploy , measure, volume if no gcp’s were used to get accurate measurements?
R.Perry
Well-Known Member
At UC Merced project we used gcp's supplied by the surveyors. Yes they did volume reports as well as downloaded to the engineers CAD programs. One of the engineers I worked with there checked the depth of an excavation for sewer line and then checked DD data with on site measurements and DD was within a couple of inches. That gave their engineering department a lot of faith in the data they were getting from DD.
You got off easy. I'm working a 100 acres site right now and I average about 5500 images if I shoot the whole thing from 330 ft AGL with an Sony RX1RII. Processing time is about 34-38 hours.
I‘m going to show my ignorance here. I was looking at doing a job with over 150 acres a day (didnt get it BTW)...construction progress. I had a rough estimate of 110 images per 5 acres. Roughly 3300 JPEG images. They were expecting to run images from the field on a hotspot. Are you shooting 3D? Additionally, can I realistically run that much on hot spot? Sounds extremely large.....and I’m not sleeping in a tent waiting. ?
R.Perry
Well-Known Member
My goodness what kind of overlap did you have. I never came close to that even when doing hatch pattern and ortho. What lens were you using? I always used the 75/65 overlap shot from 250 feet, X5 camera with 15mm lens. This gave my client a very nice map with the ability to zoom in on any part of the project. Each photo was averaging nine megabytes.
What has amazed me I come up with the same results with my Mavic Pro, and the resolution is fantastic, at least to me it is.
R.Perry
Well-Known Member
The number of images are going to be based on your overlaps, flight altitude, and camera and lens. Obviously if you are doing cross hatch, and Ortho then you are going to shoot a lot of photos. To be honest, I couldn't see the difference between a single pass verses hatch. If Fred is hanging around here he is a professional photographer and can probably give you some very good advice.
If you are going to use Drone Deploy, I found their support to be tremendous and their training aids excellent.
Also my clients haven't been interested in 3D presentation.
If you are mapping, the number of images is going to be driven by 1) the required GSD and 2) the camera you are using. If you (or the client) can live with a higher GSD and less resolution, you can decrease the total image count (and a hell of of lot of flight time). So to try and answer your question, it takes me about 22-25 minutes to transfer the 5500 14-18m images from the memory stick to the SSD for processing. Never bothered to figure out the total gigs of data. That is really not something I would want to try and run off a hotspot from the field. Is it possible? Sure. Don't have a clue how long it would take to transfer. That would depend on the reliability of the net connection and the throughput.
85/85 @ 330' AGL constant with a crosshatch pattern for 3D work. The ortho tiff output approach a terabyte; not something you normally use for every day. We just do this once every two months on the job site to track changes over time in grade to wow the suits.
R.Perry
Well-Known Member
Here is a link to my uTube site, I've been allowed to post a couple videos to show the UC Mereced site and Stocton site.
I'll see if I can post a mapping link.
I'll see if I can post a mapping link.
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 397
- Reaction score
- 232
- Age
- 49
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
How's the mapping going? Are we off the ground yet?
I just wanted to throw my two cents in there that I'm seeing a lot of recommendations for some really high overlaps. About the most overlap you need is 80/65 and 70/60 will suffice in times of need or when you just need a 2D map. I work allot with photogrammetry engineers and anything over 80/65 is overkill. The very thin slices that get created when running too much overlap actually start removing the geometric advantage and cause undo processing which can actually lead to poor results and very high processing times. Especially if you are not running PPK or RTK. Also, there is no need for cross hatch if you are just mapping terrain. Crosshatch is best suited for oblique images for mapping structures. I would highly recommend running a deer flight in addition to this.
I just wanted to throw my two cents in there that I'm seeing a lot of recommendations for some really high overlaps. About the most overlap you need is 80/65 and 70/60 will suffice in times of need or when you just need a 2D map. I work allot with photogrammetry engineers and anything over 80/65 is overkill. The very thin slices that get created when running too much overlap actually start removing the geometric advantage and cause undo processing which can actually lead to poor results and very high processing times. Especially if you are not running PPK or RTK. Also, there is no need for cross hatch if you are just mapping terrain. Crosshatch is best suited for oblique images for mapping structures. I would highly recommend running a deer flight in addition to this.
