Just curious for some professional feedback on my computer choice. I’m wanting to test and demonstrate ODM in the field (offline) and recently picked up a Dell Inspiron 16 Plus w/ 16GB Ram, the NVIDIA 3060 video card.
I was hoping I would experience some lightning-speed processing vs. my MacBook Air… but not so. I’ve gone through a number of articles and have adjusted my WSL2 to use about 14GB, but it still doesn’t seem to want to process anything with any appreciable resolution.
Any tips on things I can tweak to get this more usable?
14GB is pretty slim for full-size imagery, and will likely have to lean heavily on swap… You can define the swapfile/location in your .wslconfig. If you can put that on a fast volume, that will help, but honestly, the most helpful thing would be more RAM (if possible) so you can avoid swapping entirely.
The cloud option is a good one, I’m going to explore it more this week with some new datasets. I do need to be able to process usable maps offline though, which I seem to be able to do… but slowly. 315 images at 4000 pixels gave me a GSD of just over 1cm (which is overkill for my use) took about 75 minutes. I’ll keep working with it, and will plan to upgrade to 64GB of ram when I can afford it.
Is there a means to pre-load baseman’s (either google or map box) into WebODM to take with me offline? Being able to overlay the data over an existing google map (or better yet, Avenza) would be helpful. Any tips on how I could do that? I’ve got 1TB HD, and all this laptop is for is mapping.
Have you looked at adding a Lightning node to your setup? Then you can still have all your data locally, but use remote nodes to do the heavy lifting. Or you can setup NodeODM servers in the cloud and add them to your WebODM session.
CKen - I have considered the lightning option and intend to use it when I have a data connection. There is a relatively good chance that some of the mapping I will be doing is going to be done ‘in the bush’ with no connectivity available. I am looking into creating my own node locally (a box to take with me and set up a generator, etc. to do the heavy lifting the laptop can’t) but that is dependant on use case and ultimately cost.
Saijin_Naib - Initial feedback from my testing thus far with WebODM, Drone Deploy and Pix4D… the ability to upload basemaps (from anywhere really, I’m not picky) to use locally for data collection without a datalink, and the ability to then overlay that data onto an existing basemap while offline; might be the single most useful thing to first responders. Additionally, perhaps only second to speed of processing offline (my laptop only has 16GB RAM at the moment, I will be upgrading to at least 64GB).
At this point, I still don’t quite know what I don’t know… I don’t consider myself a techy person, but here I am learning the processes. If there’s anyone on here that can direct me to some good ‘starting point’ literature on what one needs to understand to really get WebODM working well, I am all ears.
Agreed. Also found this incredibly important in small/local govt. work for field-ops (sewer teams, especially. Not getting much signal in a pit).
I think the OpenDroneMap: The Missing Guide book by Piero might be a good resource, though it is mostly targeted towards ODM releases around the 1.x.x range and is missing a number of updates hence. [I work for UAV4Geo, but this recommendation stands regardless of my affiliation]
My intention is to make this a repository of everything one would need to understand about OpenDroneMap as well as best-practices for data collection, processing, post-processing, etc, but it is not there yet.
No: it’s an apples to oranges comparison as far as capacity and optimization, unfortunately. There has been discussion of creating an equivalent tool in the ecosystem, but that hasn’t yet happened.