Hi there,
I’m trying to get a High Quality orthomap with WebODM (and Lighting) with these options: resize-to: -1, depthmap-resolution: 1000, mesh-octree-depth: 11, dsm: true, dem-resolution: 1.0, orthophoto-resolution: 1.0
How can I get better quality? If you compare with Dronedeploy there are a big gap:
WebODM:
Dronedeploy:
Thanks.
Lightning has a few restrictions on configuration parameters that relate to fine detail, in an effort to keep best effort/fair service level for everyone.
You may need to process locally to really push the limits.
Thanks for your answer. I have short limits locally for that reason I was trying lighting but It is not really useful because I can`t reach best quality with Lighting neither for the restrictions of fine detail that you said. Are there options? Do you know if there are rental servers or something like that?
You could try to set it up yourself on a cloud services platform like Azure, Google Cloud Platform, or Amazon.
Here’s an article that may get you headed in that direction:
The author does provide the service of configuring a hosted WebODM for you, as well.
Also, we have another pretty major infrastructure upgrade going in like, right now.
We’re changing part of our processing pipeline from MVE to OpenMVS, and from what I’ve seen & have been able to understand, this should represent a significant improvement in our point cloud reconstruction, and thereby, the quality of all further derivatives (3D models, orthophotos, DEMs, etc.)
Check out this comparison from four years ago to get an idea of where the two tools stood back then:
Perhaps wait and see when that update goes live on WebODM Lightning and try re-running your dataset against the new pipeline with OpenMVS to see if it matches your expectations more?
If you are able to share this dataset (privately, if need be), I will run it through the latest ODM and we can see what improvements are coming and how they compare.
Share it, share it, share it!
Piero is doing some really cool stuff, and I believe it will help you directly!
Improved Point Cloud Filtering:
Improved Reconstruction:
Improved Feature Extraction & Matching:
Massive stuff.
Excellent stuff
Sorry guys I was outside taking photos with my drones.
@smathermather-com and @Saijin_Naib All these sounds really nice!! Here is my dataset to try (Middle Age Castle close to my House):
Do you know when will you have ready this new pipeline to Lighting? It works with no restrictions for sure it will be my solution. I really like the idea of WebODM with external Node but I need to reach High Quality 2d Maps and 3d models.
Thanks for your comments.
Hi,
Do you have to invoke some flag in order to use OpenMVS or is it automatically in the pipeline? I ask since when I run odm 2.1.0 I still get an mve folder in the working directory. Is it running OpenMVS or MVE?
Tnx
Should be automatic. If you run via the console with --debug --verbose you should see OpenMVS spit out a bunch of info when it begins its stage.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.
