Split depth generation as separate task?

[WickedAndy]

would it make sense to have the cpu/gpu intensive task of generating the depth data in each image
as a separate process from the actual mesh generation? say you make a target mesh of 100000 polys
but then want to have a higher res mesh, could just re-use the saved depth data go straight to the
geometry construction phase?...   just spent 24hrs doing depth generation, wish it could be re-used..

[WickedAndy]

or even just have photoscan SAVE after generating depth and before generating the mesh... just had a photoscan running for over 86 hours that failed on mesh generation.... now it's back to square one 

[akemono]

I second this, it would be great if you were able to re-use the depth stage for recalculation of the mesh.

[Infinite]

Sounds like a good idea, 2 stage process. Save after each one and be able to re-use depth pass, should you need to.

[kosmarnik]

I concur.
Been doing some tests and it was crashing on the mesh gen after 2 hours of waiting. Gave up for now.
Makes no sense to have it all in one all-or-nothing go.
Heck with the free SFM software you have it all in discrete steps.