WebODM versus Lightning

Couldn’t find a better place to ask this. Apologies if in wrong place.

Can Lightning produce any deliverable that I can’t also produce in WebODM on my PC?

Thx… Bob R.


Nope! But it might have more resources, so potentially better output reliability.

1 Like

Interesting reply. I’d have expected that the processing time in WebODM with resource limitations would just get long rather than producing reduced quality. If your reply is really correct, how do I know if my output quality has been reduced by resource limits?

( My dual quad-core 60GB RAM PC seems to be handling most of my jobs OK, but now I’m not sure if I’m getting the best available output quality)

Thanks Saijin… R

1 Like

And they are disabled on lightning too… . :slight_smile:

My 768GB physical machine is a bit of an outlier, but a lot can be done for even pretty big datasets for 60-200 GB of RAM. The alternative is to rent on a cloud provider and use ClusterODM to spin up and destroy the instances so you don’t burn too much money.


Smathermather-cm – Could use some context for your reply. What is disabled on Lightning too ?

OK - I think I have my answer. I’ll summarize what I think I understand, and will close the thread in another day or so.

More RAM during processing means that you have more options for custom settings. Lightning has abuncha RAM, probably more than most of our local PC’s, so Lightning gives you more flexibility in settings. But otherwise, you have access to the same deliverables.

Bob R.

1 Like

Pretty much, with the caveat that @smathermather put above that the most dangerous/fun setting, for overriding the GSD, is locked on Lightning because idiots like me would try and set it to something stupid :smiley:

1 Like

Exactly. Lightning doesn’t allow --ignore-gsd nor orthophotos that exceed 1cm. So if you need those features, you need to roll your own… :slight_smile:

1 Like

Good to know.

That’s makes me realize that the whole adjustable GSD thing confuses me. I thought that Ground Sampling Distance was a direct mathematical function of the camera details and AGL flight level. So why should there be an adjustable GSD in the processing pipeline?

OK, maybe I’ve answered my own question. If the images have camera details and flight AGL in the EXIF data, then the GSD is known. If the images don’t carry that data into the processing, then a GSD has to be calculated by the user and entered into the processing engine.

Am I close?

1 Like