2.3.0 - No cameras with depth maps in working volume

[jenkinsm]

I am testing Metashape 2.3.0 for the first time and I am trying to re-process some fisheye images from an old project. When I try to create a mesh from depth maps, the depth maps are created successfully but then I get this error: "No cameras with depth maps in working volume. Retry processing?"

I'm only using 132 images for this test and they worked in previous versions of Metashape. I tried using these same images in a new Metashape project but got the same error. I also tried generating a point cloud but got "Zero resolution" as the error.

It seems that Metashape is silently failing to generate the depth maps at all.

However, it is not consistent behavior. I started a new project to test and further narrow down the problem and I get mixed results. I tried the same batch of 132 JPEG images in the new project, and they also failed to produce depth maps. Then in the same project I made a new chunk and loaded 133 EXR images. Those aligned, produced depth maps, and produced a mesh.

Since these are fisheye images and there is a new Fisheye setting in Camera Calibration, I duplicated the chunk with EXR images and switched the fisheye type to see how the mesh differs. But after re-aligning the images with the new fisheye setting, it would no longer create depth maps in the new chunk. But the previous chunk does still produce depth maps!

So I'm quite confused as to why depth maps are failing. Nothing is out of the ordinary. To be clear, the JPEG and EXR images both used the Equidistant Fisheye setting - EXR worked, JPEG did not. Then switching the EXR to Equisolid Fisheye caused it to fail.

[Alexey Pasumansky]

Hello jenkinsm,

Can you please send the project to support@agisoft.com saved in its state after the mentioned error appears? Additionally please share the complete processing log corresponding to the same failed operation.

[tksharpless]

I have the same experience, using a 96 image fisheye set that was processed fine by v.2.2.  With either of the new fisheye models, alignment succeeds and gives a good sparse cloud, but create dense point cloud fails with error "Zero resolution", after apparently building and saving depths maps successfully.  The log shows that it was unable to locate any usable cameras.  Build model fails for the same reason, but issues a different error message.
I did verify that there is no such problem with the 'frame' camera model.
The statistics of the sparse alignment are as good as with the old fisheye model, however there is a some structure in the residuals, indicating that the actual lens curve was not fit perfectly.  That may be because my lens has neither equisolid nor equiangular projection, but a nearly stereographic one -- it is the Samyang 8mm f/2.8 series 2, widely used by panographers.  You might want to consider adopting a more flexible fisheye model like the one used in PTGui -- the gold standard pano stitcher -- which is based on the universal lens curve that Donald Gennery developed for NASA's moon missions.  It is just a simple one-parameter trigonometric formula, but it can fit all realizable lens curves remarkably well.

[Alexey Pasumansky]

Hello tksharpless,

Can you please share the log corresponding to the failed depth maps + dense cloud generation procedure?

[Alexey Pasumansky]

Hello tksharpless and jenkinsm,

Please check, if you observe the same problem in 2.3.0 build 21427.