Always running out of memory

Not sure if i am doing something wrong or not. I just reinstalled webodm. I’ve tried running a job 3 times and it keeps running out of memory. I’m using about 80 images and in the last try i let it resize them down to 512, which I’m guessing would be too small to produce anything decent, but even at that size it still says it’s running out of memory. Is this normal? I’ve got 32gb of ram installed. I would have thought less then a hundred 512px images would require that much memory.

1 Like

Little update - got it to run to completition. I lowered the max concurrency. I was setting it to 14 - i think on the successful run i lowered it to 12. However, what it produced wasn’t useable. Not sure if that’s because i resized the images so small, or if i need to change some other settings, or if the photos themselves weren’t good enough for this task.

Is there a user manual that explains some the settings and some basic usage? Because right now I’m just basically guessing and trying things at random.

Thanks

1 Like

The OpenDroneMap: The Missing Guide book is an excellent resource!
https://odmbook.com

Forgive me if I’m miss-remembering your setup, but if it is Docker, it won’t be allocating all 32GB, likely half system RAM by default, so 16GB or so.

Check the docker documentation for how to tune that up a bit. I’d recommended leaving no less than 4GB for Windows 10, provided you have swapping enabled for it.

The OpenDroneMap: The Missing Guide book is an excellent resource!
https://odmbook.com

Isn’t there a user manual for the software? The only thing in the zip file was an installation manual, but I haven’t found anything that actually explains how to use the software or an explanation of all the options/features.

Thanks

Actually, it looks like it was the max concurrency that was the real issue. It’s set to a default of 28, I dropped it to 12 or 10 and then it ran ok.

1 Like

There is documentation on the website (docs.webodm.org), GitHub, and the command help option (–help on command line).

Max concurrency should be set to the number of threads on your machine. And yeah, it will scale RAM usage up a bit during different phases as things are processing in parallel. You will lose some speed by dropping back, unfortunately.

Thank you, I will check out that link tomorrow and see if i can figure out what I am doing wrong.

My CPU has 14 physical cores, 28 logical processors - i think I remember seeing somewhere an explanation that the ram usage is dictated by max concurrency and the size of the images being processed. In this case the images are about 500kb/2400x1348px.

1 Like