Webodm vs precisionmapper

#1

Good morning to all,

I’ve been a webodm user since Sep. 2018 and have done some drone photography surveying with a DJI Phantom 3 Advanced.
I’ve prepped a Intel® Xeon® CPU E3-1225 v6 @ 3.30GHz with 32gb of ram, 32gb partitionon a very fast sata disc and a 2tb raid sata on lubuntu 18.04 with the minimum bells and whistles for running webodm on docker.
All is fine and dandy and running except for the fact that i’ve processed the same images on precisionmapper (while it’s free) and on webodm and the results are better times on my physical pc than on precisionmapper.
The ortho and 3dcloud point, on the other hand are quite the opposite, for worse on webodm.
I’m using the latest update of webodm using the “3d model” profile, changing only the meshsize to something like 1000000 and even then, it’s very edgy on the potree visualization.
The orthophoto is completely different, for the worse because I see the 3d distortions on it whereas on the precisionmapper version, the orthophoto is as flat as the 3d model seen from the top/down.
In otherwords, webodm flattens the “trees” in different directions whereas precisionmapper flattens the “trees” on an orderly fashion.
What I would like to ask for, is for a setting that will suffice for 3d objects like mountains, castles and high treetops, and another setting for the flat orthophotography for mapping and forest maintenence and analysis.
I am willing to pay you guys for these settings and other further developments of webodm instead of giving out money to a closed software developer like precisionhawk.

Can you guys help out?

Thanks!

Carlos

#2

Out of interest, for comparison with a dataset I’m looking at, are these trees with foliage, or bare branches?

#3

Hello, i’m finishing setting up links to show the dataset (the same 113 photos) with high difference in rocky faces with roof tops;


The second dataset is at:


The latter are a little bit better using:
Options: mesh-octree-depth: 11, rerun-from: dataset, use-3dmesh: true, orthophoto-resolution: 1.27, mesh-size: 1000000, orthophoto-bigtiff: IF_NEEDED, depthmap-resolution: 1000

but the result on webodm along the path where the pickup truck is, the trees are a mess compared to precisionmappers process.

Appreciate any help on the matter!

Thanks,

Carlos