Forum

Author Topic: Processing a massive underwater photoset without GPS  (Read 5257 times)

CheeseAndJamSandwich

  • Full Member
  • ***
  • Posts: 153
    • View Profile
    • Sketchfab Models
Re: Processing a massive underwater photoset without GPS
« Reply #15 on: November 06, 2021, 07:38:48 AM »
So...
Alexey...
  • For Sequential, do the filenames matter??? Or folders?
  • When is using Estimated for a 2nd run of Align Photos worth doing?
  • Are their any other tips for Aligning these UW photos with no GPS data?
Thanks!

I'd also love to know more about sequential preselection.

My guess would have been that choosing sequential additionally ensures that sequential pairs are always matched/compared, when they may have not been picked up by generic preselection, rather than excluding any non-sequential pairs.

That doesn't make complete sense though, because I would think that 99% of all photos taken by everyone with any hint of method to their madness would take one photo after another in an overlapping sequence, in which case it would make sense for Metashape to do this regardless. And as the number of sequential pairs only increases linearly with the number of photos, there's hardly any processing time penalty for doing it.

Also my guess doesn't make sense because if it was additional, then you should be able to choose this in addition to estimated and source reference preselection types, rather than only being able to choose one of three.

Anyway, my guesses are no use to anyone, including myself, so I'd love to know what it's doing really.
These are great points.

All my scans are sequential.  And even with my non-standard rig of having two cameras on a 3m pole, the time-lapse images from both do still overlap sequentially, just in a zigzag, rather than just forward...  Hence i'm seeing the alignment process zip along through the Selecting Pairs stages, etc, where it'd usually take ages.

Could Metashape ever have the ability, or an option to enable it to deal with multi-mission, multi scan projects?
Where we get the benefit of sequentially aligning the individual scans, where the photos are all overlapping sequentially, and then it does another stage without Sequential, such that it can then find matching pairs contained between different scans.  It does this already obviously, with sequential turned off, and it can align multi mission, multi scan projects but it takes ages (as i know too well).  Could it also perhaps be intelligent, and take the rough locations of the cameras in aligned sequentially, then find the first few pair using brute force, but then switching to the intelligent way of comparing pairs that are in the same rough area?


Metashape seems to be used a lot of applications where the data is collected in multiple missions, flights, dives.  Where the scans are either carefully adjacent, or crossing, but which overlap.  And each of these missions, flights, dives do have sequential cameras. So couldn't Metashape use, as James is getting at, a hybrid approach, where it will try sequentially and non sequentially to align the cameras.

UPDATE to my previous post:
The troublesome scan i was testing with Sequential, it failed to align 2/3 of the cameras, rather than a the 10% before.  I'm guessing because the sequence was 'broken' by a run of a few bad quality cameras... It just stops there.  Couldn't MS switch back to the normal 'brute force' way, then when it's finding images that have a matching pair, then see if it can switch back to sequential. 

And thinking back to a scan from yesterday, it did have the last section of the scan shooting off on a completely wrong plane...  So i selected all of those cameras and realigned them, and they then aligned great... BUT...  The scan did cross over itself a few times...  Did MS find matching pairs between the cameras in the path where it crosses overs???  Or are they independently lying on top of each other...  EEEK!  Now I'm worried...  Will have to somehow check this...


Alexey,
Please, please, please can you explain, with a lot more detail, all of these alignment options, and MS's methods, with example, and advantages, disadvantages of each.  We really need the details here.  Alignment is the step that EVERYONE has to start with! 
« Last Edit: November 06, 2021, 08:00:12 AM by CheeseAndJamSandwich »

SimonBrown

  • Jr. Member
  • **
  • Posts: 62
    • View Profile
    • AccuPixel Ltd - Dealer and Training Centre
Re: Processing a massive underwater photoset without GPS
« Reply #16 on: November 06, 2021, 02:21:09 PM »
Quote
The troublesome scan i was testing with Sequential, it failed to align 2/3 of the cameras, rather than a the 10% before.  I'm guessing because the sequence was 'broken' by a run of a few bad quality cameras..

Sequential really is aimed at automated and machine executed missions. Breaking the sequence will break consecutive alignment.

Quote
When is using Estimated for a 2nd run of Align Photos worth doing?

Every time you use sequential. Make sure Keep Tie Points is checked in preferences. Estimated will then cross-align the images.

Quote
And thinking back to a scan from yesterday, it did have the last section of the scan shooting off on a completely wrong plane...  So i selected all of those cameras and realigned them, and they then aligned great...

Check the number of tie points in the Reference pane for these cameras. If its less than 100 then consider their alignment weak and likely to see them removed during recursive optimisation - which is required if accuracy of output is a goal.

Quote
I'm having to live on less than £10 a day at the moment, as I'm proper skint now.  So i'm stuck with the cameras and hardware i already had

Dont change your kit. Change your capture method. Do you really need all those images? Or could a lesser amount do? Process and align what is needed - this will save time.
Agisoft endorsed online Metashape training - see: https://accupixel.co.uk/metashape-training-courses

CheeseAndJamSandwich

  • Full Member
  • ***
  • Posts: 153
    • View Profile
    • Sketchfab Models
Re: Processing a massive underwater photoset without GPS
« Reply #17 on: November 06, 2021, 03:00:19 PM »
Quote
The troublesome scan i was testing with Sequential, it failed to align 2/3 of the cameras, rather than a the 10% before.  I'm guessing because the sequence was 'broken' by a run of a few bad quality cameras..

Sequential really is aimed at automated and machine executed missions. Breaking the sequence will break consecutive alignment.
I think a lot of my scans will be compatible with this... I can mimic an AUV if i need to!  I've got a lot of scanning that is just straight flying over the flatish reef...  It's only when i have house sized boulders to spiral around and under the swimthrus, that it then gets messy.  But running the good scans with Sequential is definitely gonna save me hours of processing, for sure.  So many thanks for getting me to test that!

Quote
Quote
When is using Estimated for a 2nd run of Align Photos worth doing?

Every time you use sequential. Make sure Keep Tie Points is checked in preferences. Estimated will then cross-align the images.
I have it set in Prefs, and have tried with Estimated, but it didn't align any more...  I've tried it with and without selecting all the unaligned and resetting them.  Just can't get it to work.  I'll keep experimenting when i can.  Otherwise though, these problematic scans can just be aligned well enough without Sequential i guess.

But, wouldn't it be nice if MS could deal with multiple sequential sets of cameras in one alignment...

Quote
Quote
And thinking back to a scan from yesterday, it did have the last section of the scan shooting off on a completely wrong plane...  So i selected all of those cameras and realigned them, and they then aligned great...

Check the number of tie points in the Reference pane for these cameras. If its less than 100 then consider their alignment weak and likely to see them removed during recursive optimisation - which is required if accuracy of output is a goal.
This lot actually aligned pretty sweet, and were not thinned out any more than the rest of the cloud when gradual selecting...

Quote
Quote
I'm having to live on less than £10 a day at the moment, as I'm proper skint now.  So I'm stuck with the cameras and hardware i already had

Dont change your kit. Change your capture method. Do you really need all those images? Or could a lesser amount do? Process and align what is needed - this will save time.
Swimming faster is my only hope!
The GoPro Time-Lapse interval choices limit me...  IF i could set it to 3 seconds, instead of 2, that would be fantastic.
I'm processing each scan/dive/mission, which is quick, and then Reducing Overlap set to 9, this removes 1/4 to 1/3 of the photos.  This 'optimised' set of photos then gets merged it the master model, which contains folders for each dive, and their optimised sets of photos...  So at least the humongous alignment job is only dealing with the excessive overlap between each dive, rather than the excessive overlap of every scan too.  It's been working well, compared to not doing it, that is...
I've done some of the more recent scans using the traditional lawn mowing approach, as the reef in that area allows for it, and it sure does allow me to add acres and acres really easily and fast!  I'm fortunately being donated tanks such that i can get out to do the scanning for these community dive site maps... So i'm kinda having to just go and scan fresh areas, or fill holes, rather than rescanning areas properly... If i was starting the whole project from scratch, I'd definitely be able to scan most of the site this way...  But the rest is just a maze of boulders, swim-thrus, ledges, canyons and overhangs, which are always just gonna be a nightmare.  They make for an excellent dive site though!

Thanks again for your help.
« Last Edit: November 06, 2021, 03:03:56 PM by CheeseAndJamSandwich »

CheeseAndJamSandwich

  • Full Member
  • ***
  • Posts: 153
    • View Profile
    • Sketchfab Models
Re: Processing a massive underwater photoset without GPS
« Reply #18 on: November 06, 2021, 03:07:53 PM »
Hi,

I am processing a massive underwater photoset (so far 50 000 photos) and I keep taking 10 000 photos every day. There is no GPS data and I am looking for a solution to speed up the matching part of the process. Therefore, I need to understand how each of the following work.

1. Reference preselection set to sequential + generic preselection. What is it going to do? Will the program do the sequential preselection first and only the photos that have no pair will have generic preselection applied? Or will it do the generic preselection for all photos, finds all matches and then also add sequential preselection on top to find even more matches and actually extend the length of the process?

2. Reference preselection set to seqential while every second photo disabled, so only half of the photos are being aligned. Provided that they will align somehow... Then I enable the other half of the photos and what are my options then? Sequential again? Or perhaps Estimated? With generic preselection on or off?

Any other ideas how to match only the photos that should be matched?
Mariusz, how is your alignment going?