171 images takes a looong time :-O

Hello valued community!

As a test, I ran 171 images from a Phantom 4 RTK today and selected the high-resolution orthomosaic as template. The process took six hours to complete, which is fairly long for that small amount of images.

Is there any way to improve the processing time?

webODM (latest version)
Running on local node
i9 10900
RTX 2080 Super
64 Gb RAM

image

1 Like

Hmm…

Can you post your processing log?

Sure, here it is for you to download.
It was too long to be uploaded as text.

1 Like

Sorry for the hijack, but is there a difference between an RTK and non-RTK workflow if no GCP’s are used? Are you using GCP’s?

Can you try with tightening up --gps_accuracy to something closer to what you expect your mean accuracy to be for the dataset? I’m curious if that might help reduce the number of partial reconstructions, which may help things along.

Do you think 1m is a better estimate than 10m we have by default?

Changed the value to 1m, doesn’t seem to have any huge effect so far.
The accuracy should be +/- 2cm for these images

1 Like

If you’re confident in the 2cm, maybe passing --gps-accuracy 0.2 might be worth a shot, too!

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.