Hi Quality TLS model - procedure

[Darko]

Hello,
I've been working with large data sets of aerial images with the 80% overlap. I would like to get the best 3D model out of it and, so far, it seems that ContextCaptue is doing it better. What is the procedure to get the BEST POSSIBLE 3D TLS model. So far, I've been using photos only, but from now on I am planning to use DJI L3 lidar. Most of my work is done using 5cm GSD, do you have any suggestion for the best workflow?

Best, Darko

[Alexey Pasumansky]

Hello Darko,

Usually higher quality 3D model output can be achieved using depth maps based model generation based on High quality depth maps.

If you are targeting on tiled model as an export result, you may consider using Block Model concept (split in blocks option in Build Model dialog) instead of building single solid model.

[BenW22]

Hi Darko,

I could write a very long respose to this question. The Context Capture engine does seem like a better output but that can be alot to do with the subject matter. Also CC is 10x the price over 10years!

Using aerial images I get very good results out of Metshape using DJI M300 / Zenmuse P1 combo this for ortho, model, dem, point cloud and tiled model. I dont use it yet for my lidar reductions

But my short answer is, when using photos (not lidar) clean the sparse cloud very well through gradual selection before creating your dense cloud, this is a must for getting good results. There is alot of info out there but this but this guy has very good videos on it - https://www.youtube.com/@lightninggis4758/videos

Also, as Alexey says,  creating the depth maps gives me better results.

Kind Regards
Ben