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.

Messages - CheeseAndJamSandwich

Pages: [1] 2 3 ... 9
We could consider all of Metashape's jobs, not just depth maps...

If you look at the console, MS is going through cycles and cycles, stopping to write out zips, over and over again, then moving on to other stages, and more and more cycles...
Many, or all could be considered checkpoints and could trigger an 'autosave', where it could write out the progress done so far and committing the data to disk...  Some jobs would allow multiple, frequent autosaves, with no performance hit... Some might want to run for longer as it's all in RAM...  Whatever...

Then, if there was a system crash, or whatever, whereby we'd consider the data/settings to be good, we could then opt to tell it to resume from the last good checkpoint.
IF the dataset was bad, or the settings chosen wrong, or a stubborn bug, then we might choose to just tell MS to run the job(s) from scratch (as we have to now).

Metashape is only gonna be employed for bigger and bigger projects in the future... and something like this would allow it to scale up to those tasks.

Having MS crash after almost getting to the end of a 4-day alignment is not fun, i can tell you... And for a lot of people, that would be costing money.
Being able to resume from where it got to, maybe loosing an hour of processing, would easily acceptable.

Bug Reports / Reduce Overlap = Error: empty geometry
« on: January 05, 2022, 08:02:48 PM »
Metashape Standard 1.8.0 build 13794

Aligned, done some mild Gradual Selection, ran Optimise Cameras, then created a mesh from Tie Points (i guess this is 'Sparse Point Cloud' renamed??)... Mesh looks normal.
Then, running Reduce Overlap set to 9 gives the Error: empty geometry.

Code: [Select]
2022-01-06 00:50:33 Reducing overlap...
2022-01-06 00:50:33 Building params...
2022-01-06 00:50:33 ROI: [-123.287, -95.6844, -12.6808] - [138.38, 146.938, 5.44214]
2022-01-06 00:50:33 n voxels g: 292 -> 250
2022-01-06 00:50:33 n voxels n: 292 -> 292
2022-01-06 00:50:33 voxels in meter g: 0.955414
2022-01-06 00:50:33 voxels in meter n: 1.11592
2022-01-06 00:50:33 restricted distance in voxels g: 15
2022-01-06 00:50:33 capture distance in voxels g: 1
2022-01-06 00:50:33 margin g: 3
2022-01-06 00:50:33 restricted distance in voxels n: 17
2022-01-06 00:50:33 capture distance in voxels n: 1
2022-01-06 00:50:33 margin n: 3
2022-01-06 00:50:33 filling geometry...
2022-01-06 00:50:33 fillGeometryInfo : fill
2022-01-06 00:50:33 fillGeometryInfo : no voxels found
2022-01-06 00:50:33 filling geometry...
2022-01-06 00:50:33 fillGeometryInfo : fill
2022-01-06 00:50:33 fillGeometryInfo : no voxels found
2022-01-06 00:50:33 Finished processing in 0.019 sec (exit code 0)
2022-01-06 00:50:33 Error: empty geometry

This workflow i've done dozens of times before on 1.7.n


Feature Requests / Re: Modify Trackball size
« on: January 05, 2022, 05:13:57 PM »
Spotted this in the release notes...

Code: [Select]
Version 1.8.0 build 13597 (30 November 2021, preview release)
Standard and Professional editions

• Added trackball size setting to Preferences dialog.

And it's working exactly as hoped.
I'm using 80% at the moment, and it's much, much easier to use, way less twitchy, and very easy to make tiny adjustments to get the viewing angle just right.

Many thanks!!!  :-*

Suggestion; Run sequential. Then re run alignment with sequential off (generic) and reset alignment ticked off. This should at least align parallel track images.

Just tried this, and the adjacent track, that had good overlap, created a hilariously bad '2nd skin'...  a few inches above the other one where they overlapped.

Will try again with 'Estimated' set...

Feature Requests / Total Elapsed time in Processing in Progress dialog
« on: December 19, 2021, 09:18:06 PM »
When processing a batch/task, the Processing in Progress dialog doesn't tell us how long it's taken in total...  Only the elapsed time for the current task.
Can we have it display this total time please?


We need something that also works for us lowly 'Standard' users too...
We can't run scripts.

Also, Sequential fails to align U-pattern, 'mowing the lawn' scans, even when there's good side overlap.  Which is surprising, as this type of scanning is so, so common.
They're aligned sequentially, but when you look at the side overlaps, you see there's two surfaces, that are not joined...doh.
How come metashape can't sequentially align them, and then knowing that the two paths are overlapping along their sides, then align them...?

General / Re: How to straighten a 'banana' scan in MS Std?
« on: December 03, 2021, 05:45:01 PM »
As already mentioned, oblique and rotated images add a lot to a more solid bundle adjustment, so certainly give that a try.
I'll integrate it into the 'flight plan' for future scans.  It'll actually allow us to add detail thru the coral frame areas.

After adding a few more scans to add width to the site, it came out much better, leaving a small bow... which you can see in the Sketchfab model here:

But it may also really help to calibrate your Gopros in a more controlled "lab-setup" (this doesn't necessaraly mean photos of a calibration pattern).
Gonna have to learn all that... never needed to do it with my other scans, because they're not as long and 'flat'.  I'll follow the tutorials when the next batch of processing is finished...

I've recently used a few GoPros and their lens distortions are very "challenging", but the cams have their use, just like in your project.
Yep, they're just so convenient for UW, easy to get hold of and fairly capable.  And awesome for filming manta rays on my day off scanning.

The advantage for pre-calibrating Gopros could be that they have a fixed focus, so this parameter will not change.
So you could just try importing the calibration (individually per camera, not just generic per cam model). Of course their will still be some variations in the field (or below water ;) due to temperature differences, general calibration tolerences etc...
With a more complex topography, that naturally involves a lot of oblique, rotated photos anyway, that comes out looking accurate enough, is there any advantages to doing the pre-calibration?  Does it speed anything up? (remembering that my alignment of 20-30k photos takes 3-4 days)

Attached a downsampled pic of the UV-distortion map for a Gopro Hero 10: pretty crazy, especially the corners.
Do you have any experience with the 7-8?  How does it compare with the 10?  There's a big jump in mp and a new chip... So i'm assuming it's a lot better...  If only GoPro would sponsor me and give me 6!  :-*

General / Re: How to straighten a 'banana' scan in MS Std?
« on: December 02, 2021, 06:16:14 PM »
It's a thing of little calibration errors which are adding itself and results to a bended banana model. In standard version the only thing you can do is improving the calibration by improving your shooting pattern for a better on the fly calibration by adding oblique and rolled photos, like you do for lens calibration in Agisoft. Or as Simon Brown suggested you can import a working calibration from another project.
As i'm adding more scans, making it wider, it's a lot, lot less bowed up...
And the shape of the reef i'm scanning is a flat section on top, then a shallow angled slope, then a steeper angled slope that continues down deeper... If i widen the scans more to include more of the flat and deeper steeper sections, perhaps it'll come good?

And adding a couple of scan passes at an oblique angle would help straighten it up?
I could add an oblique scan track or two thru the middle of the site where the coral frames are placed, which will actually add more detail to them.  My cameras are normal to the reef usually.

Feature Requests / Re: Manuals
« on: December 02, 2021, 07:39:05 AM »

Also, where Alexey, etc explains a feature in more depth in forum replies, perhaps some of these explanations could be added to the manual?  That knowledge, information needs to be captured.

The better the manuals, the less forum posts, the quicker new users get up to speed, and optimise their workflow and get better results... This will make MS more valuable, increasing sales.

Good documentation is very important.

General / Re: How to straighten a 'banana' scan in MS Std?
« on: December 02, 2021, 07:25:48 AM »
I don't have pro.

How can i straighten a bowed model with Metashape Standard?

General / Re: How to straighten a 'banana' scan in MS Std?
« on: December 01, 2021, 11:46:36 AM »
Here's some images of a single scan pass (with 3x gopros on the 4m pole) along the reef.
When i combine it with the other scans at different depths (making it wider) it becomes much less bowed, but is still bowed.
I have a full site scan processing, but that might take another 12 hours  :o to finish...  It has an annoying slight bow in it.

General / How to straighten a 'banana' scan in MS Std?
« on: December 01, 2021, 11:21:29 AM »
Until now, the models of my underwater scans using gopro timelapse photos has been almost perfect, geometry wise...
I'm now scanning a long site, with long scans at varying depths, and also having 3 x gopros on a 4m pole...  to make it 3x quicker to scan... (works great!)
But now the long scan shows a banana bend in it, bowing up.

Is there any way of correcting this bow in Metashape Standard?  (i.e. i can't use markers, GCPs :'( )


General / Re: Seabed mapping -> alignment of 100.000 pics
« on: November 21, 2021, 10:44:50 AM »

No as you can see in the screenshot i have a single chunck with some "subchuncks".
All pics allign but these subchuncks do not.


You know what it should look like...  I'm not sure what it should look like, or what your picture shows us...
Should it be just one long scan with a few parallel tracks???  The arrows are confusing...

Pages: [1] 2 3 ... 9