Messages

31

General / Re: Agisoft Metashape 1.6.0 pre-release

« on: October 26, 2019, 08:23:18 PM »

Hello BjFaeTorphins,

Using Estimated preselection option after quick Sequential alignment run may be reasonable, for example, if the frames are acquired from the vertical zig-zag survey. Sequential preselection option would not give you tie points for the images from the different flight lines, therefore further processing may result in the artifacts (like stairs) on the surface in the areas of the overlap between the flight lines.

After alignment with Sequence when I try  a second alignment using Estimated I see the following message...
"Warning: Can't resume matching without keypoints"
What am I doing wrong?

32

General / Re: Agisoft Metashape 1.6.0 pre-release

« on: October 25, 2019, 06:48:51 PM »

I was wondering about the new options in the align photos stage : source - estimated - sequential.
I think sequential speaks for itself but what do the others ?

Sequential pre-selection is trying to match the images with a bunch of their neighbors according to the camera labels.

Estimated preselection is based on the estimated camera locations after initial alignment (for example, performed using rough accuracy settings).

One of the possible scenarios of use - when aligning the video sequence frames run Sequential preselection at first and then Estimated preselection, if there are intersecting parts of the image acquisition path.

I've given this a try. 

Sequenial Preselection certainly is much quicker. 6 minutes for 594 cameras compared with 81 minutes for "Estimated".  Both runs reset alignment.

Running the alignment a second time with Estimated after first doing a Sequential didn't seem to do anything.  (Both done at High accuracy)


Is the idea that the Estimated (second alignment) should improve the accuracy?

Can you explain the steps?

33

General / Re: Mesh from Depth Maps problems (v1.5.2)

« on: April 29, 2019, 05:12:11 PM »

Hello BjFaeTorphins,
 
You can switch depth_build_lod tweak to True status in the version 1.5.2 (depth maps should be rebuilt, if you want to see the effect).

Also can you please confirm that for the meshing process in the problematic project you have input Custom Face count and haven't used one of the available pre-sets (High/Medium/Low)?

It looks like I've set it to 1,000,000 faces rather than one of the pre-sets.  What difference will this make?

Can you tell me what depth_build_lod does rather than wait 2 days 13 hours to get another bad model?

34

General / Re: Mesh from Depth Maps problems (v1.5.2)

« on: April 22, 2019, 10:42:16 PM »

Hello BjFaeTorphins,

Depth lod should be considered properly in the version 1.5.2 for the mesh generation.

When you say about max_neighbors tweak, do you mean depth_max_neighbors or dense_cloud_max_neighbors?

Also can you send the PDF reports from the projects generated in 1.3 and 1.5.2 version to support@agisoft.com?

dense_cloud_max_neighbors

How do I consider "Depth lod properly in the version 1.5.2 for the mesh generation. "?

I'd generated the reports and email to support as requested.

35

General / Re: Mesh from Depth Maps problems (v1.5.2)

« on: April 22, 2019, 09:57:02 PM »

BjFaeTorphins,

Which quality setting did you use (High, Ultra High?).

Depth Maps method is often lacking if the photos IQ isn't pitch perfect while the older dense cloud method manages to get more out of bad quality images from my experience. I can confirm that the depth maps workflow in 1.5.2 is totally messed up compared to 1.5.1 though. I already submitted a support ticket to the team 3 weeks ago and hope that things are being worked on.

Mak

Medium.  It's 5,800 cameras and took 2 days 13 hours to do the mesh

36

General / Mesh from Depth Maps problems (v1.5.2)

« on: April 22, 2019, 08:15:42 PM »

I've been having problems getting any decent  models since upgrading to MetaShape v1.5.1 and v1.5.2.  Its all underwater models that I do and the images are very poor quality so i assumed that i'd just had a run of really bad image acquisition.

However I went back to a model i did with v1.35 to compare results and it turns out that the Mesh from Depth Maps produces very much worse results.

See the attached images.

v1.3.5 was produced via a DPC.  I was pretty happy with this result as the visibility was less than 4m with lots of matter in the water and out popped this excellent model.

v1.5.1 was produced using the Depth Maps method using v1.5.1

v1.5.2 also was produced with the Depth Maps method but using v1.5.2.  When the "black veil" is manually removed (which is a very time consuming task and never completely successful) the model looks very like image v1.5.1.

Tweaks..
Max Neighbors 60
depth_build_lod false
optimization true

37

Bug Reports / Re: Metashape 1.5 error when Remote Desktop

« on: January 08, 2019, 12:10:48 AM »

Email sent.  sorry for the delay.

38

Bug Reports / Metashape 1.5 error when Remote Desktop

« on: January 04, 2019, 06:05:12 PM »

When I Remote Desktop to my workstation running Metashape I get an "Invalid Operation" error written to the Metashape log.  The stack trace is very long so I won't post it here unless you are interested in having it.

I have seen this in previous versions which, if i recall correctly, actually caused Photoscan to crash.  Metashape does not crash.

Not such a big issue really but thought best to mention it.

39

General / Re: Agisoft PhotoScan 1.4.0 pre-release

« on: January 03, 2018, 01:09:02 AM »

I believe I'm experiencing some kind of stability problem with v1.4 Build 5532 Standard Edition.

The symptoms are that a very large number of Handles (16M and 31M on two separate runs) are created over time and my machine pretty much comes to a halt. By coming to a halt I mean unresponsive to mouse or keyboard.

Only one of the 36 available cores (core 2) is doing anything, it is at 100%.  all other cores are essentially idle.

It's an Intel i9 with 128Gb memory and twin EVGA GTX 1080 Ti SC Gaming 11 GPUs.

This happens many hours (10+) into the photo alignment stage of processing a 7200+ camera model.

On one occasion while running the above mentioned big model I got impatient and started up a 2nd copy of Photoscan to build a mesh using the new algorithm on a model that I had originally aligned using version 1.3.4.  It was a  smaller model (3200 cameras). I noted during this run that again only core 2 was at 100%.  It got to a stage with 16,000,000+ handles that neither copy of Photoscan seem to be making any progress.  I tried pausing one then the other but neither paused or made any progress (no output to the console).  When the copy of Photoscan that had been running for 12+ hours was killed the handles reduced to 15,000 or so proving it was the long running copy of Photoscan that appears to be leaking Handles.

PS: Memory usage is fairly low (<25% of 128GB).

HTH

Hi Alexey

In the above quoted post I mentioned a problem with a large number of handles.  I've repeated my test and found the source of the Handles leak to NOT be Photoscan.  I hope that has not caused any confusion or unnecessary work on your part.

Turns out there is a scenario where Photoscan appears to have a Handles leak after all.  See attached screen grab showing Photoscan holding 14,520,358 Handles while 65% (11:07 hours) of the way through Align Photos stage. I've upgraded today to build 5650.

Progress has stopped. See attached screen shot for Out of Memory error.  Handles had reached 16,711,680 and only 2GB of memory is in use.

40

General / Re: Agisoft PhotoScan 1.4.0 pre-release

« on: January 02, 2018, 10:45:33 PM »

I believe I'm experiencing some kind of stability problem with v1.4 Build 5532 Standard Edition.

The symptoms are that a very large number of Handles (16M and 31M on two separate runs) are created over time and my machine pretty much comes to a halt. By coming to a halt I mean unresponsive to mouse or keyboard.

Only one of the 36 available cores (core 2) is doing anything, it is at 100%.  all other cores are essentially idle.

It's an Intel i9 with 128Gb memory and twin EVGA GTX 1080 Ti SC Gaming 11 GPUs.

This happens many hours (10+) into the photo alignment stage of processing a 7200+ camera model.

On one occasion while running the above mentioned big model I got impatient and started up a 2nd copy of Photoscan to build a mesh using the new algorithm on a model that I had originally aligned using version 1.3.4.  It was a  smaller model (3200 cameras). I noted during this run that again only core 2 was at 100%.  It got to a stage with 16,000,000+ handles that neither copy of Photoscan seem to be making any progress.  I tried pausing one then the other but neither paused or made any progress (no output to the console).  When the copy of Photoscan that had been running for 12+ hours was killed the handles reduced to 15,000 or so proving it was the long running copy of Photoscan that appears to be leaking Handles.

PS: Memory usage is fairly low (<25% of 128GB).

HTH

Hi Alexey

In the above quoted post I mentioned a problem with a large number of handles.  I've repeated my test and found the source of the Handles leak to NOT be Photoscan.  I hope that has not caused any confusion or unnecessary work on your part.

Turns out there is a scenario where Photoscan appears to have a Handles leak after all.  See attached screen grab showing Photoscan holding 14,520,358 Handles while 65% (11:07 hours) of the way through Align Photos stage. I've upgraded today to build 5650.

41

General / Re: Agisoft PhotoScan 1.4.0 pre-release

« on: December 31, 2017, 04:12:30 PM »

Hello bjb,

Some steps of the Build Mesh stage are using only one core, so it is not the bug, however, the long time may be related to the large number of photos in the set and model complexity.

Thanks Alexey
I'll wait a while longer then.

While not a bug it would be good if there was some indication that progress is being made otherwise it just looks like it has hung.

That's been 42+ hours since any progress has been noted.  I just tried to "Pause" and as you can see in the attached screen capture Photoscan has not paused.  It looks pretty hung up to me.

42

General / Re: Agisoft PhotoScan 1.4.0 pre-release

« on: December 31, 2017, 03:47:33 PM »

I believe I'm experiencing some kind of stability problem with v1.4 Build 5532 Standard Edition.

The symptoms are that a very large number of Handles (16M and 31M on two separate runs) are created over time and my machine pretty much comes to a halt. By coming to a halt I mean unresponsive to mouse or keyboard.

Only one of the 36 available cores (core 2) is doing anything, it is at 100%.  all other cores are essentially idle.

It's an Intel i9 with 128Gb memory and twin EVGA GTX 1080 Ti SC Gaming 11 GPUs.

This happens many hours (10+) into the photo alignment stage of processing a 7200+ camera model.

On one occasion while running the above mentioned big model I got impatient and started up a 2nd copy of Photoscan to build a mesh using the new algorithm on a model that I had originally aligned using version 1.3.4.  It was a  smaller model (3200 cameras). I noted during this run that again only core 2 was at 100%.  It got to a stage with 16,000,000+ handles that neither copy of Photoscan seem to be making any progress.  I tried pausing one then the other but neither paused or made any progress (no output to the console).  When the copy of Photoscan that had been running for 12+ hours was killed the handles reduced to 15,000 or so proving it was the long running copy of Photoscan that appears to be leaking Handles.

PS: Memory usage is fairly low (<25% of 128GB).

HTH

Hi Alexey

In the above quoted post I mentioned a problem with a large number of handles.  I've repeated my test and found the source of the Handles leak to NOT be Photoscan.  I hope that has not caused any confusion or unnecessary work on your part.

43

General / Re: Agisoft PhotoScan 1.4.0 pre-release

« on: December 30, 2017, 04:06:38 PM »

log file is too big.

Maybe you can provide its final part that is related to the problematic operation only?

Without documentation of the console log how can i know what is the problematic operation?

Log file from the start of the process is 17Mb. so it's hard to see how to cut it down to 512kb

I've cut out the repitative bits.  Hopefully I've left in the bits that can help.

44

General / Re: Agisoft PhotoScan 1.4.0 pre-release

« on: December 30, 2017, 03:37:07 PM »

Hello bjb,

Some steps of the Build Mesh stage are using only one core, so it is not the bug, however, the long time may be related to the large number of photos in the set and model complexity.

Thanks Alexey
I'll wait a while longer then.

While not a bug it would be good if there was some indication that progress is being made otherwise it just looks like it has hung.

45

General / Re: Agisoft PhotoScan 1.4.0 pre-release

« on: December 30, 2017, 12:47:29 PM »

Please see attached screen shot.

The Building Model stage is only using one core.

15 hours later.

Is it hung?  One core is 100%