I am trying to figure out the most cost-effective way to process images and share results with clients. I am going to be surveying ranches of about 1000 acres, so expecting 3000+ images per map. My computer only has 8GB of RAM, so I figure even doing split-merge would be futile considering the processing power.
Here is my idea, would love to hear if this is doable or not:
1.) Install WebODM on my computer and set up tunneling so clients can access it online. (Sharing maps and models)
2.) Install a processing node on an AWS EC2 server and only turn this on when processing these large image sets.
I think this solution is the most cost-effective for my business while allowing me to employ all the tools of webodm. What do people think?
EDIT: After looking around the forums some more, thinking it might be easier to set up a super cheap server with ~2GB to run the WebODM UI constantly for my clients to access and then deploy processing nodes on a more robust server to use only when I need to construct maps.
You could always buy a 256 gB ram workstation pc from newegg for around $600, install a hefty SSD with your favorite linux OS distribution and process images that way and like you said use a lightweight PC as the web facing interface where the processed can be hosted. just my two cents.
I had been considering that option as well. Although I think I will start with the cloud-hosted virtual machine while I test out the concept so I don’t sink too many costs into it.