Hello dear people!
I am planning to map a local Golf course. It’s huge and has many fields. Naturally, I won’t be able to fly it all in one flight, so it’ll be about 50 small flights and I will cover a huge area.
Is there any way for WebODM to combine two maps or to process a huge ammount of data (50k images or more)?
Hey @A_Random_Dood theoretically, yes, using split-merge https://docs.opendronemap.org/large.html
I don’t think anyone has ever tried 50,000 images, but would love to hear the results. You will need a machine with substantial amount of memory (RAM) to handle the merge step, depending on the target resolution.
I will definetely report back I’m also looking in the near future to donate a sum of money to the project for it’s further development, I’m planning to do a map of my entire city as well 360 km² (139mi2) which will probably take about half a million pictures. I currently have about 120 spare terabytes hooked up with a Dell R720 server, with dual Xeon E5-2667v2’s and 384GB of Reg ECC DDR3. I could add more if needed.
While we don’t accept donations, there’s multiple ways to support the project (even financially), see GitHub - OpenDroneMap/WebODM: User-friendly, commercial-grade software for processing aerial imagery. 🛩
Definitely keep us posted with your results / findings.
Oh this sounds awesome! I’ll definetely get the book
What do you think runs the fastest? I’ll purchase the iso to support the project, but which one do you think is the most optimized and on which platform?
(I mean, either get my own Docker running and configure manually, or run the ISO).
If you are on Linux, both are equal in speed (docker or ISO).
Windows/Mac are a bit slower regardless of the choice because of the VM overhead for running docker or the ISO.
Sure thing, thank you so much! I’ll be back in a week or so with updates. See you!
Sorry for double posting, just curious. Does ODM have any usage of GPUs yet? Or is it still purely CPU?
Also, what counts in a CPU? Is it better to go for multiple threads (a server CPU) or raw power? (less cores, faster processing).
No GPU support as of yet, but efforts to implement it might start sometimes next year.
More cores is definitely better as most parts of the pipeline are multi-threaded.