resume a corrupt model

[Andre]

Dear All,

we just suffered a black out in the middle of dense reconstruction - stereo reconstruction step.

It turns out that I can open the project (previously saved) and import the sparse model. Then I open the reconstruction panel, open the model and refresh, and I see that the undistortion is OK for all photos, and the stereo and depth just for some of them.

If I try to run again the stereo step, Colmap starts with "reading workspace", then "reading configuration", and when it starts processing the first image it crashes suddenly.

I guess there is something corrupted in the model that causes this problem, but I don't know how to fix it.

Please help us, otherwise we'll have to restart from the very beginning (and it took three days to get to this point).

Thanks very much!

Kind regards,

andrea

[Shaun Berryman]

Andrea,

I've had something similar happen in the past and all I had to do was go to the dense workspace folder, then under the stereo sub folder you'll have 'depth_maps', 'normal_maps' folders. Go into each of those sub folders (technically the additional sub-folder of 'images') and look for files which have 0 bytes. Delete the files that are zero bytes and start the stereo step again.

That has worked for me in the past, hopefully it is something equally as easy for you to continue the reconstruction.

-Shaun

[Andre]

Hello, thanks fro the quick answer! Unfortunately, I tried but this does not work.

I also tried to delete all images generated by undistortion and stereo, in order to "save" the sparse model, but also this fails.

Now I've tried to export the sparse model in VSFM format .nvm, but then I am no more able to import this format in Colmap. This seems a bit odd, I supposed it was possible.

Any other suggestion? Thanks!

andrea

[Andre]

Update: I also tried to export as PLY and then import again. In this case Colmap starts opening the file and then crashes.

Thanks in advance!

andrea

[Johannes Schönberger]

As the other person in this thread already pointed out, this is likely caused by a corrupt depth or normal map, as the process probably crashed while writing the data to disk. If your images are all the same size, then you could check for incomplete files and delete them. Otherwise, you could write a little script that reads the images from disk and checks that the corresponding depth and normal maps have the appropriate file contents. There is an example for how to read a depth/normal map in the scripts/python and scripts/matlab folder. 

--
You received this message because you are subscribed to the Google Groups "COLMAP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to colmap+un...@googlegroups.com.
To post to this group, send email to col...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/colmap/936a720d-3cbe-4947-9463-56afaeaab067%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Andrea Bistacchi]

Thanks!