Messages

1

Python and Java API / Re: Script for filtering mesh via gradual selection, connected component size option

« on: August 03, 2023, 02:46:43 AM »

Yes, that's possible too

This is example of how to check all chunks' labels and then, if label equal to desired name - make it active, so that all Metashape.app.document.chunk accesses were addressed to it:

Code: [Select]

for chunk in Metashape.app.document.chunks:
    if chunk.label == "My Custom Chunk Name":
        Metashape.app.document.chunk = chunk

Probably you meant to name a chunk from GUI - but for the sake of completeness - this is how to change the currently active chunk name from Python:

Code: [Select]

Metashape.app.document.chunk.label = "My Custom Chunk Name"

P.S. script was adapted from https://www.agisoft.com/forum/index.php?topic=7815.0 (googled with "metashape python make chunk active")

Thanks for that, but still incredibly confused how to implement it or what to type to replace words.

At the moment my script looks like:

Code: [Select]

PhotoScan.app.document.chunk.importModel('Path/0.obj')
PhotoScan.app.document.chunk.model.removeComponents(100000)
And yes I am doing this in Phtoscan at the moment but I know it's possible to do the same in Metashape.

2

General / Re: Possible to Import Model to specific labelled chunk(s)?

« on: August 02, 2023, 08:52:06 PM »

At the moment my script looks like:

Code: [Select]

PhotoScan.app.document.chunk.importModel('Path/0.obj')
PhotoScan.app.document.chunk.model.removeComponents(100000)
How do I get it to select the next chunk or select chunk by name?

And yes I am doing this in Phtoscan at the moment.

3

General / Re: Possible to Import Model to specific labelled chunk(s)?

« on: August 02, 2023, 04:27:39 PM »

Hello 3D_Scan_Fan,

Do you mean that you have a folder with the model files and would like to perform the import only if model filename corresponds to the chunk label in the Workspace?

If so, than what are the formats of model files available in the import folder and whether the chunk labels include the file extension?

That is correct.

The formats of everything is pretty standard, they're all OBJs.
The chunk labels don't contain the file extension (they're labelled numerically).

If this is not possible then an alternative would be, how do you set a (specific) chunk active in the workspace throgh a script perhaps?

4

Python and Java API / Re: Script for filtering mesh via gradual selection, connected component size option

« on: August 02, 2023, 02:58:50 PM »

Once again, thanks for that.

However I was hoping if there was a way to specifically name a chunk, make that active, and then apply all the actions to it?

5

General / Re: Possible to Import Model to specific labelled chunk(s)?

« on: August 02, 2023, 02:31:32 AM »

Anyone at all may I ask?

6

Python and Java API / Re: Script for filtering mesh via gradual selection, connected component size option

« on: August 02, 2023, 02:31:06 AM »

Hi, maybe this gradual_selection_mesh.py script can help you?

Thanks for that!

Although that looks incredibly complicated, I've managed to use the

Code: [Select]

removeComponents set to a certain number that's close to my mesh polycount to do the job (although I still prefer to Level/Percentage the input).

Although I am having problem doing this for multiple chunks in the Workspace. 
How do I get this to select each chunk to be active so that the script applies the actions?

7

General / Possible to Import Model to specific labelled chunk(s)?

« on: July 30, 2023, 05:50:47 AM »

Hi everyone,

As what the title says, I was wondering if it is possible to import a model to a specific chunk in the Workspace?

This is in a scenario where you do have multiple labelled chunks and you would want to import multiple models but of the same filename as the chunk itself.

Of course this will have to be done via. Scripting.

I have skimmed over many pages on the Python API PDFs and I don't think I can see much relevance or that maybe I am missing something.

Please do let me know if this was or is possible at the very least 

8

Python and Java API / Re: Script for filtering mesh via gradual selection, connected component size option

« on: July 29, 2023, 11:53:31 PM »

Sorry for bringing this old topic up but I too want to know about this.

I've just made a script for removeComponents but I want a threshold = 99% in the script, however after researching as much as I can I just can't seem to find anything that is equivalent.

Is there a way to do this in Phython by any chance???

9

General / Re: 1.8.0 produces softer meshes. Why?!

« on: November 24, 2021, 05:41:47 PM »

The latest build.

I can't share this dataset but I've tried this with others and they're all the same, soft meshes.

10

General / Re: 1.8.0 produces softer meshes. Why?!

« on: November 24, 2021, 04:02:14 PM »

It's definitely not noise and why it's only 1.8.0 that does this.

11

General / 1.8.0 produces softer meshes. Why?!

« on: November 23, 2021, 08:04:20 PM »

Hello everyone!

Maybe like some of you I am trying the preview release of Metashape, version 1.8.0. and wanted to show you the mesh results here compared to 1.6.4.

Reason why 1.6.4 because everything afterwards either has problems crashing and mesh generations has strange artifacts which makes no sense (particularly 1.7.3!).

Generating meshes in 1.8.0 are pretty good, it no longer produces those weird artifacts, however the results of the meshes tend to look soft (even when doing everything at the highest quality).

1.8.0


1.6.4

The details are still there but a little soft, but not as soft processing at half the resolution.

Despite this, it does really well in the more occluded areas, particularly behind the ears:

1.8.0


1.6.4

As you can see it was resolved the ear backs really well, no holes, no mesh interpolation, thus it's a lot more accurate.

Also to note it's faster, approx 40 mins on 1.8. compared to 1 hour on 1.6.4. Yet I can't comment if the speed has changed over the previous versions.
Nor can I make a solid statement if it being faster is because it's not producing higher quality meshes.

Regardless I feel this is once again another step-down from anything beforehand. Is there any reason as to why this is happening??? 

12

General / Re: Depth Map scripts doesn't work anymore! :(

« on: October 26, 2021, 07:45:58 PM »

I can't post the whole log because it exceeds 20000 characters

This is what I get when it's stuck:

Code: [Select]

[GPU 1] Camera 97 tile #3/6: level #6/6 (x1 downscale: 1760x1760) done in 2.94 s = 35% propagation + 57% refinement + 2% filtering + 0% smoothing
[GPU 2] Camera 110 samples after final filtering: 0% = 100% - 67% (not matched) - 5% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 26% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 110 tile #1/6: level #6/6 (x1 downscale: 1760x1760) done in 1.126 s = 34% propagation + 38% refinement + 6% filtering + 0% smoothing
[GPU 1] Camera 97 samples after final filtering: 0% = 100% - 66% (not matched) - 3% (bad matched) - 1% (no neighbors) - 0% (no cost neighbors) - 30% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 97 tile #4/6: level #6/6 (x1 downscale: 1760x1760) done in 1.909 s = 31% propagation + 57% refinement + 2% filtering + 0% smoothing
[GPU 2] Camera 110 samples after final filtering: 0% = 100% - 41% (not matched) - 4% (bad matched) - 1% (no neighbors) - 0% (no cost neighbors) - 54% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 110 tile #2/6: level #6/6 (x1 downscale: 1792x1760) done in 3.15 s = 33% propagation + 57% refinement + 2% filtering + 0% smoothing
[GPU 1] Camera 97 samples after final filtering: 0% = 100% - 42% (not matched) - 6% (bad matched) - 1% (no neighbors) - 0% (no cost neighbors) - 51% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 97 tile #5/6: level #6/6 (x1 downscale: 1792x1760) done in 2.432 s = 35% propagation + 54% refinement + 2% filtering + 0% smoothing
[GPU 2] Camera 110 samples after final filtering: 0% = 100% - 53% (not matched) - 5% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 41% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 110 tile #3/6: level #6/6 (x1 downscale: 1760x1760) done in 1.867 s = 31% propagation + 51% refinement + 4% filtering + 0% smoothing
[GPU 1] Camera 97 samples after final filtering: 0% = 100% - 64% (not matched) - 2% (bad matched) - 0% (no neighbors) - 0% (no cost neighbors) - 34% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 97 tile #6/6: level #6/6 (x1 downscale: 1760x1760) done in 2.13 s = 30% propagation + 57% refinement + 2% filtering + 0% smoothing
[GPU 1] group 1/1: estimating depth map for 45/47 camera 111 (26 neighbs)...
[GPU 2] Camera 110 samples after final filtering: 0% = 100% - 58% (not matched) - 7% (bad matched) - 2% (no neighbors) - 1% (no cost neighbors) - 33% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 110 tile #4/6: level #6/6 (x1 downscale: 1760x1760) done in 1.654 s = 35% propagation + 47% refinement + 4% filtering + 0% smoothing
[GPU 1] Camera 111 samples after final filtering: 0% = 100% - 43% (not matched) - 11% (bad matched) - 3% (no neighbors) - 1% (no cost neighbors) - 41% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 111 tile #1/6: level #6/6 (x1 downscale: 1760x1760) done in 1.539 s = 36% propagation + 46% refinement + 3% filtering + 0% smoothing
[GPU 2] Camera 110 samples after final filtering: 0% = 100% - 29% (not matched) - 10% (bad matched) - 2% (no neighbors) - 1% (no cost neighbors) - 59% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 110 tile #5/6: level #6/6 (x1 downscale: 1792x1760) done in 3.153 s = 35% propagation + 53% refinement + 2% filtering + 0% smoothing
[GPU 1] Camera 111 samples after final filtering: 0% = 100% - 35% (not matched) - 7% (bad matched) - 2% (no neighbors) - 1% (no cost neighbors) - 56% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 111 tile #2/6: level #6/6 (x1 downscale: 1792x1760) done in 2.311 s = 33% propagation + 55% refinement + 3% filtering + 0% smoothing
[GPU 1] Camera 111 samples after final filtering: 0% = 100% - 47% (not matched) - 7% (bad matched) - 2% (no neighbors) - 1% (no cost neighbors) - 43% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 111 tile #3/6: level #6/6 (x1 downscale: 1760x1760) done in 1.577 s = 34% propagation + 49% refinement + 3% filtering + 0% smoothing
[GPU 2] Camera 110 samples after final filtering: 0% = 100% - 44% (not matched) - 12% (bad matched) - 3% (no neighbors) - 1% (no cost neighbors) - 40% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 110 tile #6/6: level #6/6 (x1 downscale: 1760x1760) done in 1.867 s = 32% propagation + 50% refinement + 3% filtering + 0% smoothing
[GPU 2] group 1/1: estimating depth map for 46/47 camera 114 (27 neighbs)...
[GPU 1] Camera 111 samples after final filtering: 0% = 100% - 70% (not matched) - 4% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 25% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 111 tile #4/6: level #6/6 (x1 downscale: 1760x1760) done in 1.456 s = 32% propagation + 52% refinement + 3% filtering + 0% smoothing
[GPU 2] Camera 114 samples after final filtering: 0% = 100% - 69% (not matched) - 4% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 25% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 114 tile #1/6: level #6/6 (x1 downscale: 1760x1760) done in 1.528 s = 31% propagation + 48% refinement + 4% filtering + 0% smoothing
[GPU 1] Camera 111 samples after final filtering: 0% = 100% - 28% (not matched) - 10% (bad matched) - 2% (no neighbors) - 2% (no cost neighbors) - 59% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 111 tile #5/6: level #6/6 (x1 downscale: 1792x1760) done in 2.385 s = 36% propagation + 53% refinement + 2% filtering + 0% smoothing
[GPU 2] Camera 114 samples after final filtering: 0% = 100% - 29% (not matched) - 5% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 64% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 114 tile #2/6: level #6/6 (x1 downscale: 1792x1760) done in 2.525 s = 32% propagation + 55% refinement + 2% filtering + 0% smoothing
[GPU 1] Camera 111 samples after final filtering: 0% = 100% - 37% (not matched) - 15% (bad matched) - 3% (no neighbors) - 2% (no cost neighbors) - 43% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 111 tile #6/6: level #6/6 (x1 downscale: 1760x1760) done in 1.712 s = 38% propagation + 46% refinement + 3% filtering + 0% smoothing
[GPU 1] group 1/1: estimating depth map for 47/47 camera 117 (38 neighbs)...
[GPU 2] Camera 114 samples after final filtering: 0% = 100% - 55% (not matched) - 10% (bad matched) - 2% (no neighbors) - 1% (no cost neighbors) - 32% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 114 tile #3/6: level #6/6 (x1 downscale: 1760x1760) done in 1.817 s = 33% propagation + 47% refinement + 4% filtering + 0% smoothing
[GPU 1] Camera 117 samples after final filtering: 0% = 100% - 71% (not matched) - 3% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 24% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 117 tile #1/6: level #6/6 (x1 downscale: 1760x1760) done in 1.684 s = 29% propagation + 51% refinement + 3% filtering + 0% smoothing
[GPU 2] Camera 114 samples after final filtering: 0% = 100% - 51% (not matched) - 11% (bad matched) - 3% (no neighbors) - 2% (no cost neighbors) - 34% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 114 tile #4/6: level #6/6 (x1 downscale: 1760x1760) done in 1.586 s = 35% propagation + 45% refinement + 4% filtering + 0% smoothing
[GPU 1] Camera 117 samples after final filtering: 0% = 100% - 37% (not matched) - 4% (bad matched) - 1% (no neighbors) - 0% (no cost neighbors) - 59% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 117 tile #2/6: level #6/6 (x1 downscale: 1792x1760) done in 3.08 s = 32% propagation + 59% refinement + 2% filtering + 0% smoothing
[GPU 2] Camera 114 samples after final filtering: 0% = 100% - 15% (not matched) - 14% (bad matched) - 3% (no neighbors) - 2% (no cost neighbors) - 66% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 114 tile #5/6: level #6/6 (x1 downscale: 1792x1760) done in 2.806 s = 37% propagation + 52% refinement + 2% filtering + 0% smoothing
[GPU 1] Camera 117 samples after final filtering: 0% = 100% - 90% (not matched) - 1% (bad matched) - 0% (no neighbors) - 0% (no cost neighbors) - 8% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 117 tile #3/6: level #6/6 (x1 downscale: 1760x1760) done in 1.634 s = 29% propagation + 57% refinement + 3% filtering + 0% smoothing
[GPU 2] Camera 114 samples after final filtering: 0% = 100% - 70% (not matched) - 3% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 25% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 2] Camera 114 tile #6/6: level #6/6 (x1 downscale: 1760x1760) done in 1.848 s = 29% propagation + 52% refinement + 4% filtering + 0% smoothing
[GPU 1] Camera 117 samples after final filtering: 0% = 100% - 61% (not matched) - 5% (bad matched) - 1% (no neighbors) - 1% (no cost neighbors) - 32% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 117 tile #4/6: level #6/6 (x1 downscale: 1760x1760) done in 1.709 s = 32% propagation + 53% refinement + 3% filtering + 0% smoothing
[GPU 1] Camera 117 samples after final filtering: 0% = 100% - 13% (not matched) - 12% (bad matched) - 2% (no neighbors) - 1% (no cost neighbors) - 72% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 117 tile #5/6: level #6/6 (x1 downscale: 1792x1760) done in 3.411 s = 37% propagation + 55% refinement + 1% filtering + 0% smoothing
[GPU 1] Camera 117 samples after final filtering: 0% = 100% - 77% (not matched) - 2% (bad matched) - 1% (no neighbors) - 0% (no cost neighbors) - 20% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 0% (speckles filtering)
[GPU 1] Camera 117 tile #6/6: level #6/6 (x1 downscale: 1760x1760) done in 1.913 s = 28% propagation + 58% refinement + 2% filtering + 0% smoothing
[CPU 3] Camera 15 samples after final filtering: 61% = 100% - 31% (not matched) - 1% (bad matched) - 0% (no neighbors) - 0% (no cost neighbors) - 3% (inconsistent normal) - 0% (estimated bad angle) - 0% (found bad angle) - 3% (speckles filtering)
[CPU 3] Camera 15 tile #1/6: level #6/6 (x1 downscale: 1760x1760) done in 676.591 s = 36% propagation + 63% refinement + 0% filtering + 0% smoothing
Using two RX 580 8GB GPUs and still using driver version 18.6.1 because none of the new ones work with Photoscan 1.3.5..

13

General / Depth Map scripts doesn't work anymore! :(

« on: October 24, 2021, 02:34:08 AM »

Hi everyone,

I don't know if anyone has the same problem as me but I've been using version 1.6.3. for a majority of work for a long time, I recently tried out version 1.7.3. & 1.7.4. and loaded my buildDepthMap script to (of course) generate Depth Maps for my chunk. However to my surprise it kind of stayed stuck at around 30%. I would leave it on for almost an hour and it just wouldn't budge! I tried cancelling the process but it wasn't responsive afterwards, it just hanged on the progress window forever, thus I had to "kill" the software through task manager.
I was even looking at the log for it and the last bits of lines doesn't seem to indicate any error.

Code: [Select]

Metashape.app.document.chunk.buildDepthMaps(downscale=1, filter_mode=Metashape.FilterMode.MildFiltering, reuse_depth=False, max_neighbors=10, subdivide_task=True, workitem_size_cameras=60, max_workgroup_size=120)
I don't know which version started for this to happen but this is incredibly annoying, a newer version of Metashape meaning to be better but yet for something so simple and that's been around for a very long time all of a sudden has problems. If anything, I've noticed performance getting laggier ever since Photoscan version 1.3.5. (still my preferred), and still can't generate 16K textures through GPU, but that's another topic for another time.

Has anyone been experiencing this? and what might be the problem perhaps?

I would really appreciate the help...I can't figure it out! 

14

Python and Java API / Re: python script differences between 1.7.3 and 1.7.4 releases

« on: October 03, 2021, 01:23:54 AM »

Are there script changes between 1.6.3 and 1.7.3?

Everytime I do a BuildDepthMap it takes WAY too long and the depth maps come out incomplete

15

General / Re: Agisoft Metashape 1.7.0 pre-release

« on: January 11, 2021, 03:40:38 AM »

I'm still getting the Error: ciErrNum: CL_BUILD_PROGRAM_FAILURE (-11) at line 1087 when trying to process Depth Maps

Might this be fixed by any chance??