What I've Been Up To

Well heck, you guys sure know how to make somebody feel welcome!

So far, my first task is done, which was to survey a hill quite heavily covered in trees. here is an orthophoto (excuse the mess):

My goal here was to get a terrain model sufficient for permitting a home build. I struggled mightily (if not efficiently). In the end, I took the filtered output (cloud file) into cloudcompare for a little manual caressing (and the use of the CSF filter). It also lets me create different ply output files at different resolution (useful for us idiots who didn’t learn about GCPs until after the fact…). So from cloudcompare I get essentially two separate clouds, the ground point in RGB:

and the non-ground points:

Now, in cloudcompare I mesh the various entities (things like the cement pad and the tent/road get separate meshes to maintain resolution. Examples…

I found that with my limited laptop capacity, and the need to use particular graphics cards for the next step of my adventure, it was best to mesh in cloudcompare to reduce the complexity…before moving on to the next package…Meshlab.

So, into Meshlab, go the PLY meshes (so i keep georeferencing), and this allows me to tweak the vertical positions of the “ground” to account for the changes due to smoothing, before flattening into a single mesh of various mesh densities…

and, now…since Sketchup is unable, or unwilling, to consider any outputs from meshlab (dae/PLY) that would be useful. Output to STL it is… and then into Sketchup. And now, all the hassle about the tent and the quonset pad mesh precision is apparent…thanks to the recent update in google earth images, since I now have after the fact GCP (to a degree):

Now, i can use a corners of the tent and the quonset pad as geolocation points for the sketchup model, which has taken the stl, and converted it, finally, into a 3D surface for the site layout and architectural work:

The entire goal of this endeavor was to get a quick survey of the site before clearing, so as to be a bit specific on the tree removal, and also to try to integrate the house into the topography. In July, I was quoted a few thousand dollars and a couple of months. So I started down this journey, Quite a steep learning curve on many levels… And I am sure that there is an easier way to do this, or at least reduce 2 or 3 of the software packages…But try as I might, crash, after crash, after crash (I now own an old r610 server for this specifically) this is what I came up with. And, it was a blast (well, most of the time).

I know this sounds like an infomercial for every open source package out there…

My next goal is the survey of the creek and meadow, since our intent it to install some fake beaver dams and try to flood and resuscitate the riparian zone (and keep the neighbors cows out…). In an attempt to get better resolution in some areas, I am running that at the “ultra” level, and it has been churning in my garage for about 5 days now… I wish there was a countdown clock… I told my wife that me, walking out the garage to pop in a quick DF or TOP to check to see if it was still alive is my form of video slot machines.

Thanks for putting up with the ramble, and I hope this at least provided an amusing chuckle of sympathy (“Poor fool!”). I learned a heck of a lot from reading this forum and the fantastic (POSITIVE) conversation. I don’t think I saw a flame in the entire forum!. For all of you who I have learned from, know that this was instrumental in our moving forward in our restoration effort. And hopefully, if the beaver come back, they will thank you as well.

Anyone out there who has experience in restoration? I would love to find a few folks to bounce ideas off of, before putting together an application. This is all self-funded, so we are looking to do this as low cost/low tech as possible. So trees and dirt for the dams. But identifying the “best” locations for these and other obstructions is also something that has its own steep learning curve…

here is an overall view of creek in its current state, after about 70 years of cattle…at the west end, the cut is 8’ deep…but there is water there at depth in the summer, even during this drought, so I have hope, if we can just slow the water down a bit…

best to all


I had very similar challenge this year to yours in many aspects of it. Even obtaining an old server to process things more efficiently, but ended up hiring the computer power on the cloud. Lucky there is a service where you can hire a mighty machine with up to 48 core CPU and 192Gb RAM for about 0.7 EUR an hour… if such might needed… I usually was fine with less powerful options.
You can create fully configured server for your needs first using cheapest 4 EUR a month cloud server, then save it in a snap that cheap to store… once you need a powerful machine you can deploy that snap it in a server with computation power up to you needs then when job is finished move the result in some cloud storage to download and kill that server until you need it again, paying only for hours you used it…


Hey Jeff_d,

I work in forestry and we’re working on trying to get some information from UASs about dense stands of forest over varying terrain.

That was a fun read, looks like you got to explore some cool software!

It looks like your forest is relatively dense, and your DSM looks great, all of the trees came out really well. Can you tell me about your mission parameters and your camera setup?


cherrmax - That is fantastic advice, and definitely makes sense. I didn’t think to go that route, because I basically didn’t know what I was doing…and needed something to learn on. At that point in time, this stuff was voodoo to me. Still is, but less threatening…

Mr7743 - I ran a grid pattern, with 90% overlap. this was my third flight. The first one was me, white knuckled confirming altitude. Then, after diving into WEDOBM with the first flight, I returned and flew two additional grid patterns on the same day, but at a 45-degree offset from each other. I chose the “best”, based on how things looked after bringing into cloud compare. Camera setup - i plead ignorance here, other than using a simple UV filter and letting the default settings work (Phantom 3 Pro). If it is of interest/use, I am more than happy to somehow make accessible the image files, as well as my interim files (cloudcompare/meshlab).



Mr7743 - one more thing, that is likely important. Features and point cloud both set to ultra. simple statistical outlier filter applied in cloudcompare. I can also provide the webodm output, if desired…


I didn’t know what I was doing too… It would be safe to say that I learned ODM and other stuff on this project…
I was just deployed in the jungle like a commando to map 5 square kilometres of dense tropical rain forest, located on hard to move through terrain with elevation variation from 0 to 300m above sea level… given slow internet and 6 hours a day electric power to process things… the conditions that boost creativity a lot… :rofl:
You can check this one
Might me useful sometimes… especially when you need to process something with ultra feature quality once in a while, but don’t have your own hardware powerful enough…


I feel much better now! I had a Motel 6 and a hot shower (and that only because I was too lazy to drive 3 hrs home…and I though I was suffering ;-). Thanks for putting it into context!.

Seriously, that sounds pretty dang cool, by comparison. I tried to search through the forum to see some of this (if you posted any). Would love to see some, if it is something you can share or point me towards.



There were couple of posts here where I was fighting with international date line. Yeah… There was another challenge… Half of my dataset was technically “today” and another half “yesterday”… or when you load it all in QGIS… half of your work appears on western edge of the world map and another half on eastern… not mentioning the risk of drone flying away if it crosses the date line accidentally on automatic mission. I did it manually once… it was exciting… I’ve got 16k km flown distance in the log during that flight :rofl: :rofl: :rofl: virtually (in drone’s head) reaching 0 meridian and going back… can write a story about that mission…
I am still processing that stuff putting everything together, but will definitely write a blog post somewhere and will share in this thread…


Ok. I feel better now about my annual battle with daylight savings time. :slight_smile:


On this photo, if you look at the last visible house on the right, everything further right is today, everything across the invisible line which is 180 meridian to the left from it including the house, is tomorrow :slight_smile: but that actually depends on what side related to the date line you are… how crazy that is?..
My mapping area’s bottom boundary lays somewhere half way between the settlement and the mountain tops… stretching higher up and over the date line to the left and to the right…
I am still working on 3d model and elevation contours…


This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.