Skip to main content
Participant
May 4, 2024
Question

Higher Res Output of Photogrammetry in Sampler?

  • May 4, 2024
  • 1 reply
  • 497 views

Hello - 

 

I'm trying to do something that I'll freely admit is a bit abitious. I'm hoping to use sampler to create a photogrammetry model of a small building site/hillside. I have EXCELLENT drone photography. I'm able to import everything, and everything lines up beautifully. However I find myself struggling to get an output that is as dense and detailed as I like. I work with laser scanned data professionally, and have the CPU/GPU to handle very dense (5+ gig) datasets with ease. I've enabled a Mesh Post Process layer on my mesh after briniging it in from the photogrammetry portion of the app, and tried playing with some of the values, but I can't seem to get anything that's noticibly more detailed. All my exported OBJs come out at like 110 megs, tops. Am I hitting some sort of internal limit of what the software will allow? Also - would be really nice for large objects like this if the texture mapping settings could go above 4096x4096. But really, I feel like the biggest thing holding this feature back is the limitation on mesh size/density. Am I doing it wrong, or is this the current limitation? 

 

-Sky

This topic has been closed for replies.

1 reply

SchuylerGAuthor
Participant
May 4, 2024

ETA - I found the "raw" setting in the middle portion of the porcess - this improved things GREATLY. However, I'm looking for any suggestions/guidance as to how to completely max out the quality of the output. I imported the mesh in progress into Rhino, and put in the proposed building structure:

 

 

Marion3D
Community Manager
Community Manager
May 14, 2024

Hi,
Are you using a mac or a PC to process the data in Sampler? There is currently a limitation in the number of polygons exported when using a mac, but it should be better if you are using a PC already.
Do you have a target in number of polygons you are/were trying to reach?

Sampler's photogrammetry features are currently target at scans of objects, and they are not made for assets of that scale/with as many details so it is possible that what you have acheived in your render is the maximum which can be acheived with Sampler today for your use case