Forum

Author Topic: Gradual selection without GCP's  (Read 2026 times)

JohnH1

  • Newbie
  • *
  • Posts: 6
    • View Profile
Gradual selection without GCP's
« on: November 18, 2021, 10:23:39 PM »
Hi,
I'm working on footage, typically from stereo cameras underwater, but with no GCPs.
Does anyone know if there is any value to doing a process of gradual selection and camera optimization without GCPS?

Many thanks,


CheeseAndJamSandwich

  • Full Member
  • ***
  • Posts: 162
    • View Profile
    • Sketchfab Models
Re: Gradual selection without GCP's
« Reply #1 on: November 19, 2021, 01:34:51 PM »
I've been doing lots of uw scanning...  using only MS Standard... so no markers...
My scans from gopro timelapse photos produce excellent results, with the only problems coming from understandable things, poor lighting, bad vis, dropping to close to the bottom, surge moving me too fast, so less overlap, etc. etc...
But with the results, especially as uw is a nightmare for clarity, sharpness, the sparse cloud is often quite 'fuzzy'...  And so i do use Gradual Selection, but with relaxed numbers, of 15 for both Reconstruction Uncertainty and  Projection Accuracy...
These generally only remove 1/5 of the points, but they do a good job of removing the fuzzy points that are off the rocks/reef...
I find this does speed up the meshing i think.  And i'm assuming the quality... But I've never quantified this.
I'd love to know a lot more about these GS options, which others i should use, and what values are good for what needs...  The documentation is really not great here...  And the only thing i've found to follow is th USGS workflows... But I'm assuming that the goals for their scanning is vastly different from my UW dive site mappings... 

Any input on GS is desperately needed!!!


What are you scanning, and how's your stereo rig setup?
I have a twin camera setup (i.e. they're not synched), and 3m pitch, (4-5m above the bottom) but only because it makes the scanning of large areas 2 x faster... effectively two divers swimming perfectly parallel for the whole scan...  This give me fantastic data down the middle of the scan, between the cameras! I'm considering putting 3, 4, more on longer poles and closer pitches, and swimming closer to the reef for when we want to scan large rehabilitation sites with better accuracy, resolution, detail...  Easy to assemble two lengths of water pipe underwater and have two divers swim along with it, controlling the 'altitude' of their ends of it.
« Last Edit: November 19, 2021, 01:39:34 PM by CheeseAndJamSandwich »
My 'little' scan of our dive site, 'Manta Point'.  Mantas & divers photoshopped in for scale!
https://postimg.cc/K1sXypzs
Sketchfab Models:
https://sketchfab.com/cheeseandjamsandwich/models