Forum

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Topics - CheeseAndJamSandwich

Pages: 1 [2]
16
I'm having a nightmare wrestling with the animation feature of Metashape.
When i set the viewpoints, i thought that the animation would actually visit that view on it's path around the scan... But if Smooth Camera Track is turned on, it simply doesn't.  It wonders over towards it, but never gets there, biases looking far too high above the object and then continues over to then miss the next one....  so the animation never shows what i wanted, what i'd told it to.
Then when i try to edit the viewpoints, by double clicking on it, moving the view, then clicking update for that viewpoint, it lands up editing it is a completely weird way, not resembling what i did.    Something is wrong here, are there bugs in this editing/updating feature?

So can we have an option to make the smooth camera track pass THROUGH each viewpoint we've set, on it's path?
So every viewpoint, view, angle, is present in the final exported video?

Thanks.

17
General / Workflow for Multi-mission, overlapping, UW scans.
« on: September 08, 2021, 11:13:19 AM »
Here's my workflow I'm trialling for dealing with multi-mission scans that also overlap a lot (too much)...
This has previously caused HUGE align times (3d), especially in the 'Selecting pairs' stage, most likely due to the excessing overlap.  Aligning every photo (19,000 so far) from scratch just isn't practical any more, so I'm reducing overlap as i go, and adding the next scan's photos to the growing alignment of the previous scan's photos...

The project:
I've been scanning an underwater dive site using 2x 12MP GoPros on a 3m pole, 'flying' 3-6m above, cameras normal to the bottom/slope/rocks... The 2x GoPros just make it 2x quicker to scan... It then produces a scan with two parallel, but not synchronised, time-lapse photo camera tracks, that have good to too much overlap side-to-side, depending on flying height, and good to too much overlap going forward, depending on how fast I'm swimming, drifting and the flying height again...  Sadly you can only set the GoPro's time-lapse photo interval to 0.5, 1, 2, 5, 10, etc seconds...  I'm using 2, though 3 would be better i think!
Due to the nature of diving and the dive site, each trip there and scan means i land up scanning over parts previously scanned, and it's amazingly easy to miss patches, not know where you are, and not swim in straight lines.  Some of the big features, house sized boulders need a spiral scan, the rest mostly a 'mowing the lawn' u-pattern.
Conditions are usually sub-optimal... greenish water sometimes, not that great vis (10-15m), and sun shimmering on the shallower rocks.  So each scan can look quite different... a 'different' colour. doh..
A 'good' scan dive gives me 1,500 photos from each GoPro... So 3,000 per scan.  The reduce overlap to 8 reduces this down by about a third to 2,000...
Due to these conditions, and the fact that I'm simply not looking for mm accuracy, only a 3D model of the rock formations, sand patches, corals, etc. I'm not pushing Metashape Standard with high quality settings, which has sometimes actually given better alignments for my crappy images, and which obviously saves time too.

Workflow:
Enable Keep Key Points in Preferences, to allow adding photos to a previous alignment.

For each scan pair (from both GoPros combined):
Add Photos and create Folder with scan name.
Align Photos:  Low - GenPre - RefPre-Est - 20,000 - 2,000 - ExclStatTiePts - AdptCamModFilt
Save as 'Aligned' (just in case)
Select any Unaligned - Reset Alignment - Align Selected
Gradual Selection: - ReconUncert = 15 - ProjAccuracy = 15  Then Optimise Cameras
Check Region, Resize.
Build Mesh from Sparse.
Reduce Overlap: Surface Coverage = 8
Select disabled Cameras and Remove.
Optimise Cameras
Generate Mesh from Depth Maps to
Save as "Scan 'n' reduced"
...Repeat for all other previous or new scans...
This reduces a 3,000 photo scan with too much overlap to 2,000.

Combining 'reduced' scans:
Load "scan 1 reduced"
Append "scan 2 reduced"
Remove all of scan 2's parts leaving its folder of now unaligned cameras only.
Move that camera folder from chunk 2 to chunk 1, next to 1's camera folder (already aligned).
Repeat the above steps from 'Align Photos', Ensuring Reset Alignment is NOT set.
Saving after alight as "1+2 aligned"
Optimising and reducing as above
Saving after optional meshing as "scans 1+2 reduced"
...Next, repeat, appending "Scan 3 Reduced" etc... Until you have "Scans 1+2+3+4+5+6+7+8+...."

This has reduced the total number of photos needed for the area covered by about a half i think.

This then give me the mesh/model of the summation of all those overlapping scans, without the 3 days of aligning all the photos in one go.

Please feel free to rip into this workflow with any suggestions...  ;D ;)
Yes, I've only gone for low, 20k, 2k... only pushed gradual selection to 15... chosen 8 as my overlap...  Should i change any of these for this type of job???

See my Sketchfab page in my Signature below for previous versions of this model of Manta Point I'm creating...
I'll post some photos later...  Waiting for the alignment after adding scan 7 of 8...

Cheers.

18
Feature Requests / Modify Trackball size
« on: August 28, 2021, 10:10:27 AM »
I find the small size of the trackball makes navigating round in 3D really twitchy, too sensitive.
I use to use CATIA V4 for years, and it had very similar 3D manipulation, but the trackball was a lot bigger... moving the solid models around the on the screen was never an issue.

Could we have an option to change the size of the Trackball?
I guess it's set to a % of the viewport.

I seem to remember that it also only appeared when rotating the model, which was a nice feature too.
Sadly, as V4 is soooo old now, i can't find any screenshots showing it.

Cheers.

19
Could we have simple 3D semi transparent water surface that we could turn on, to aid in visualising underwater scans when in Metashape?

I'm scanning underwater dive sites that go down to about 50m deep, but mostly 10-20m...  But looking at the models, unless you can recognise features and know their size, you simply don't get any cues to the sizes, depths, scale of the features/model.  What you're looking at, be it a coral formation, rocks, bommies, etc, could be the size of a shoe, or the size of a house, it's hard to tell.  It's slightly easier with wrecks, as we know roughly how big ships and their features are, but still you don't get the depth cue, which is important in itself (us divers need to know the depths).

Obviously, the ideal solution is to take the model into Blender, or similar, to create the underwater scene, where it could be made to look absolutely amazing and realistic.  But what I'd like is just a simple water surface to aid in visualisation within MS.   Plus learning Blender looks fairly involved!

Could we have an option in settings to cater for us underwater folk?
The water's surface would simply lie on the x,y plane, at z=0.
The size of it could just be to the extents of the grid size, which is already set in settings.
With MS Pro, the -z values from the GCP would set the model depths correctly... but i only have Standard, so i use the grid trick, setting the grid to meters and move the 'object', mesh to the correct depth, and to set the scale.  Works great.
The simulated surface itself could just be a simplified 3D wavy surface that resembles water enough.
It could have tint to it, so blueish, or greenish, so that it looks like you're looking thru water.
The simplified water surface could have some basic parameters that we could set, to set the wavelength, wave height, colour tint for the surface.  This would then cater for different seas states at that location, adding more size/depth cues.
Ideally, the volume under the water could also have the volumetric feel to it, but that's probably going to far, lol.

Even the most basic of simulated waters surfaces would be so, so helpful.

Cheers.

EDIT:  Oops, posted it in General by mistake... Should have been in Feature Requests.  Sorry.

20
General / Camera Groups? for multi-camera & multi-mission?
« on: July 16, 2021, 01:17:09 PM »
Does separating cameras into groups change the resulting alignment in MS?

I have a multi-camera rig for underwater reef scanning. Two GoPros on a 3m pole.  Pointing down/normal to the reef/slope/rock, they're unsynchronised but are both taking time-lapse photos at 2s intervals.  So i get a track with two lines of cameras.  With lots of overlap going forward, and good overlap to the side.

Should these photos be put into one Camera Group, or two?


I also have done many missions/dive/scans, over several days.

Should these missions be in separate camera groups too?


Does MS change how it aligns the photos if they're in Camera Groups?

21
General / When to run "Reduce Overlap", and what to follow it with?
« on: July 03, 2021, 12:28:05 PM »
I have a big project with many thousands of images, but the scanning unavoidably has too much overlap.   If i run 'Reduce Overlap' set to 8, it halves the number of cameras and gives much better processing times, and great quality results!!!  :D 8)

But what is the correct time to run it in the workflow???
Before, after running 'Gradual Selection' tasks???
And do i need to do anything after it?
Optimise Cameras?

Do i also need to delete the disabled cameras?  Are their tie-points removed?  EDIT: They're not disabled with the cameras, so only deleting them actually gets rid of extra tie-points.
And how would i delete them if i only have Standard, and can't run scripts?

Cheers.

22
Could Build Depth Maps be it's own Job in the workflow and Batch Process dialog?
Or could the batch process at least save the project after the Depth Maps are created?

I've had Metashape and my machine crash for various reason when Building Mesh, and with my quite large projects, it's rather upsetting to know that the Build Depth Maps had been done successfully, but now it has to build them all over again, which is many hours of wasted time.  It'd be nice to resume Build Meshing and choose to 'reuse depth maps'!

Plus i guess Depth Maps are used as the source for a few things now...

Cheers.

23
Feature Requests / UI: True 'Full Screen' mode
« on: July 02, 2021, 07:31:25 AM »
Would it be possible to optionally have a true full screen mode, where none of Metashape's UI elements are show, no menu, no toolbar...  So it could show just the model viewport?
The menus/toolbar could perhaps show/autohide when the mouse is moved to the top edge of the screen.

E.g. the behaviour of Chrome.

Sometimes i've used OBS Studio to record the screen, as it's way quicker that doing an animation, fly-thru, and i then have to crop the video later, leaving a non-standard aspect ratio.
Also just for showing people, it's nice to just show the model only.

Cheers.

24
Feature Requests / UI: Show 'enabled' for Cameras: "69/123 enabled"
« on: July 02, 2021, 07:21:46 AM »
In the workspace, Cameras already show how many cameras are aligned, out of the total.
But after doing, say, a Reduce Overlap, you get no indication of how many have been disabled, just red dots.

So could we also have the 'enabled' numbers showing, like the 'aligned' already are?
Code: [Select]
Workspace (1 chunk, 123 cameras)
    Chunk 1 (123 cameras, 54,321 points)
        Cameras (111/123 aligned) (69/123 enabled)

The showing of this could be conditional.  If all cameras are enabled, nothing is show, if some are disabled, then it's shown.

Cheers.

25
General / Align taking exponentially longer w/ more tracks added...
« on: July 01, 2021, 12:51:02 PM »
Is it normal that increasing the number of photos, exponentially increases the Align Photos time?
13,000 took 1 day
15,000 took 2.5 days!

What alignment settings may cause this?  I'm sure it's running with defaults.

I've started doing an underwater scanning project, mapping a rocky dive site.
It's a big site, so each dive I'm coming back with about 3,000 time-lapse photos from 2 GoPros (7+8) on a 3m pole (gives good data between them, and 2x quicker).
The tracks I'm swimming on each dive are obviously not nice and neat, mowing the lawn style tracks like a drone can give you... Instead there's a lot of overlapping, crossing, holes... And to add, the water visibility changes each dive, the colour of the water changes and the sun angle/clouds do too... There's also certainly a little too much overlaps going forwards i think... GoPros can't do 3s timelapses, only .5, 1, 2, 5, 10s, etc...  I use 2s...   So a nightmare for the Metashape i guess. Doh.
I have 7 pairs of camera tracks, which are in their own folders.  so 14 folders.

I've just aligned with the latest scans, starting with an blank model, adding the 7 pairs of tracks, totallying 15,000 photos... aligning on high.  Other settings default.....
On my Thinkpad P51 laptop, it took 2.5 days!!!  During the processing, i'm getting times like: "64302505 matches found in 1006.06 sec" during the "Selecting pairs..." cycles!
The previous time i ran it, with 13,000 photos, it was 1 day, and previously with fewer photos, much shorter.

After doing some gradual selection stuff, reduce overlap, it only takes a few hours to mesh and again to texture.  No need to create chunks.

I've attached some pics.
If you squint, you can see some manta rays i photoshopped into the capture from ms.  The mantas are 3-4m across.
And here's a decimated model in Sketchfab.
https://skfb.ly/oo7nq

I thought that the alignment was supposed to be the 'quick' part!
Help!
 

26
General / 100% CPU for minutes after loading heavy project
« on: June 26, 2021, 02:47:18 AM »
I have a big project with 13,000 cameras and a mesh of 50,000,000 faces...

When i load it, my laptop shows it pretty quick, bit it then continues with 100% CPU usage for many minutes, leaving any interaction with the model really sluggish.  After a while it subsides, and then works just fine.

I have noticed that if i switch to a decimated mesh of 3,000,000 faces, it subsides quicker, but then switching back, it starts running at 100% again until it later subsides...

What is Metashape doing during this time?
All functionality is otherwise working, just sluggish.

Using Metashape Standard Version 1.7.3 build 12473 (64 bit) with 64GB of ram.  Of which, only 12GB is being used, total.

27
General / Using dGPU AND iGPU is faster?
« on: June 26, 2021, 02:36:09 AM »
Reading the forums, articles, manual, etc, it says to only enable the discrete GPUs if you have them.
But i've done a little bit of testing, and it seems that Aligning and Building Mesh (from Depth Maps) is actually faster with both the discrete and integrated GPUs enabled for aligning and meshing.

Align Photos:  25:33  vs.  23:42
Build Mesh:  19:41  vs.  16:00
Build Texture:  Coming soon...

Hardware:
Thinkpad P51 laptop
CPU: Intel Xeon E3-1505M v6 (Undervolted 100mv)
iGPU: Intel HD Graphics P630
dGPU: Quadro M2200
Using external display, so it's driven by the dGPU rather than the iGPU.
Metashape Standard v1.7.3 build 12473 (64 bit)


I've also previously tested a different model, repeating the test back and forth, and it saw a similar result each time.  Unfortunately i didn't keep the times.
The result I'd gotten for the Build Texture did show that it was slower for that task with the iGPU also enabled.

Is it common that this may be the case?
Or perhaps just with my combination of hardware?
That my laptop is driving the display with the dGPU, leaving the iGPU idle?
Differences with using Depth Maps?
Or changes in the engines?

Does enabling 'Use GPU while performing GPU accelerated processing' always slow it down too?  ???

28
Bug Reports / Metashape doesn't like long paths (260+) on Windows.
« on: June 24, 2021, 08:51:53 PM »
I have been using descriptive filenames, which tend to get quite long, and it seems to have caused Metashape to fail.
When ms had completed creating depth maps, and moved onto creating the mesh from them, it came up with some errors saying that files weren't accessable, couldn't read or write to them... can't remember exactly, sorry.
As a hunch, i thought about my long filenames/paths, so shortened it and reran build mesh and it completed just fine.

So, by default,  Windows 10 doesn't like paths that exceed 260 characters.  But since 1607 you can enable support for them.  I did (do) have them enabled on my system.
Registry key, set to 1:
Code: [Select]
Computer\HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\FileSystem\LongPathsEnabled
Could ms be made to work with long paths on windows, if they're enabled?  Or at least be aware of them so that the error messages point to them causing the errors in processing?

Background info:
https://docs.microsoft.com/en-us/windows/win32/fileio/maximum-file-path-limitation?tabs=cmd

Pages: 1 [2]