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 - taran

Pages: [1] 2
Bug Reports / 2.0.2 Standard: Displacement maps not exporting
« on: July 12, 2023, 01:08:25 PM »
After generating a displacement map, when I export a model as OBJ, all maps *except* displacement are exported. I can verify the displacement map in the viewport as well as unwrapped in the mesh UV inspector.

However, exporting the displacement map via the UV inspector just exports the diffuse map.

Exporting via Export > Export Textures and selecting Displacement results in a black square if JPEG is selected and the default "Preserve Alpha" is ticked. Changing the export format to PNG exports the UV islands but they are completely black.

This happens to any model I have made in 2.0.2 or earlier.

Does this happen to anyone else in 2.0.2 Standard (build 16404) the latest release in July 2023?

Bug Reports / Re: Merge Chunks not working 1.70
« on: April 24, 2023, 06:27:39 PM »
I've just had the same problem after manually aligning two chunks in Standard (2.0.1). The top half of the object (a ring) is a chunk, the bottom half of the ring is another chunk. They wouldn't align very well automatically (it was close but not great) so I finished aligning them manually as meshes. The [T] symbol was next to each chunk, and viewing aligned chunks together, the complete ring mesh looked great. When I merge the two chunks, the new 'merged' chunk just has both sides of the object in their original locations. The attached screenshots hopefully show the problem, similar to the original poster. The file "misalign.jpg" shows the cameras all together in concentric circles where there should be two camera circles above and two below.

Any suggestions are welcome!

This would be so useful.

Also - in Standard edition, like in Polycam, the ability to measure between two points and set the real world distance. So easy.

Feature Requests / Re: Scale function for standard version
« on: August 23, 2021, 01:32:26 PM »
Are there any plans for this in 1.8? It's well-known these days that you can import into CloudCompare or Meshlab to set units. Just do a point-to-point measurement of an object of known length, divide the required measurement by the p2p distance, and you have a scale factor. A lot of users would be very grateful to be able to do this in Metashape Standard before export.

Feature Requests / Re: Close Holes - Within Selection
« on: August 23, 2021, 01:26:02 PM »
That would be very useful. When working on large meshes, I always use the mesh filtering to just show the section I am working on. Closing holes resets the filter, which affects an efficient workflow.

General / Re: Contour lines and false colour to show depth
« on: January 29, 2021, 02:24:37 PM »
Hi John - you can do all of this in the open source CloudCompare package. Scale your model to the correct units (e.g. metres) in Metashape Pro first. Export your mesh from Metashape (obj, ply) and open in CloudCompare (where you can also set up real world units). You can use the cross-section tool to generate contours, and the Scalar Field (SF) Tools > Projection > Export coordinates to SF to produce a false-colour plot with scale showing depth distance (if you get your axis/alignment right).

Hope this helps!

General / Re: Apple Silicon
« on: December 03, 2020, 04:55:44 PM »
Thank you Alexey. Good to hear that Agisoft is working on it.

There has been a discussion on Reddit about Metashape performance through Rosetta:

It's certainly useable, even through emulation (and 16GB RAM limit). It will be interesting to see Apple's desktop-class silicon in 2021.

General / Re: Apple Silicon
« on: November 12, 2020, 02:54:32 PM »
Now that Apple has announced new Macs (albeit low-end models) that will use the Apple M1 processor, will Agisoft investigate feasibility to create a version that runs natively on the new platform?

I'm guessing that this will be initially difficult, as the first 3 M1-based Macs do not support eGPUs (and limited to 16GB RAM). But if the CPU performance is as they advertise, it will be an interesting one to watch. Early Geekbench results suggest that the new M1 13" MacBook Pro could be as fast as the current top-end 16" MacBook Pro Intel i9. What nobody seems to know yet is the M1's GPU performance.

Interesting times for Mac users...

General / Re: Apple Silicon
« on: June 29, 2020, 05:42:42 PM »
I imagine that they will, but the real question is over the - as yet - unknown graphics architecture of future Macs. The developer Mac Minis are essentially iPad Pros which have their own integrated graphics. This does perform well for gaming, but I've not yet seen any evidence that it will also be 'good' for photogrammetry. Will Apple still provide decent dedicated GPUs from AMD? Who knows?

As I'm replacing my old iMac, I did pause for a few days to have a think, but there are too many variables yet, and I'm sure that Agisoft will be scratching their heads too. So I'm just going to get a new tried and tested Intel Mac and watch curiously from the sidelines.

Interesting times ahead for Macs, for sure.

General / Re: Best Macbook Pro for Metashape/photogrammetry?
« on: June 29, 2020, 05:37:12 PM »
I'm about to get a new MacBook Pro to replace my ageing 2014 5K iMac. I'm going for the 16" 2.3GHz 8-core 9th gen i9 with 32GB RAM and the AMD Radeon Pro 5600M with 8GB of HBM2 memory. For my ~400 26MP photo projects I rarely see Metashape use more than ~20GB RAM as I tend to create meshes from depth maps rather than point clouds. If you can, get 64GB (I would if I could afford it, because you can never have too much RAM!).

Hope this helps.

General / Re: Face count
« on: November 16, 2018, 05:45:23 PM »
You were right - thanks Alexey!

General / Face count
« on: November 09, 2018, 06:27:31 PM »
 I'm using 1.4.4 build 6848 on macOS. In previous builds of PhotoScan I used to be able to set my own custom face count when at the 'build mesh' stage for Arbitrary (3D) surface type.

The 'custom' face count option is now only available if I choose the hight field 2.5D option, which isn't appropriate for object scans.

Am I missing something?

Thank you Alexey, this solved the problem.  :)

Thank you Alexey. Will you be able to post back on this thread once it is fixed? Happy to test it.

Many thanks.

I'm using the latest build - (using MacOS, 24GB RAM, AMD Radeon R9 M295X 4096 MB GPU) and I am unable to build dense clouds. The error dialogue reads "Another GPU context is already active". I have restarted and tried again, with no success.

Has anyone else encountered this error?

Pages: [1] 2