How to spped up process?

[andy_s]

The bigger the distance between stereo-pairs, then better the results. With a short basis between pictures the angel for triangulation is small what causes errors(In object direction). I was always using all pictures because I was sure that photoscan is using more Stereo-pairs to estimate one point...

Porly - this seems not only plausible but logical [to me].

The debate [perhaps] is fundamental to accuracy-quality and/or processing time. Looking forward to more comment.

[andy_s]

Cross-reference to Topic: 1D, 2D or 3D camera setup?

http://www.agisoft.ru/forum/index.php?topic=964.0

[RalfH]

The bigger the distance between stereo-pairs, then better the results. With a short basis between pictures the angel for triangulation is small what causes errors(In object direction). I was always using all pictures because I was sure that photoscan is using more Stereo-pairs to estimate one point...

Yes, this is generally true, but only to a certain extent. If the distance (viewing angle) between images becomes too large, new problems will appear, mainly related to increasing difficulties of matching features because (a) the changes are too large between images and (b) low incidence angles (i.e. grazing view of objects). In essence, image distance should be neither to small nor too large.

[andy_s]

[My understanding] While Figure 12 in the document www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XL-5-W1/251/2013/isprsarchives-XL-5-W1-251-2013.pdf does display multi-stereo performance as much superior to single stereo model at 15 degree intersection angle - which is itself [generally] much superior to the single 3 degree intersection model I can't see what the intersection angle for the multi-stereo model is.

I'm still confused.     

[Wishgranter]

try think how is it working our 3D vision, you need some separation between views so brain calculate Zdepht. So say 15 deg is one of mentioned working angle what work great in pscan too. If have images too close together, then it cause artifacts.....

[andy_s]

I [think I] understand Wishgranter - it's the "optimal" [mult-view] solution that I don't think the paper addresses.

[Wishgranter]

Ok try think that basic of stereo can be used in multiview. will try to make clear example of this.... 

[andy_s]

Wishgranter, I still get the feeling that - rather than an optimal model - the thesis of the paper is to throw everything [including the kitchen sink] at it -

"For multiple stereo models, small such measurements can be useful for the validation of measurements with strong intersection geometry. Even though small intersection angles lead to noisy results, models with small base lines should be acquired and used within the surface reconstruction. Since large baseline models have lower image similarity - which is challenging for the matching method, small baseline models are required additionally. Furthermore, highly overlapping imagery leads to high redundancy, which is beneficial for the precision in object space."

Maybe [with a supercomputer and deeeep pockets] that is the optimal model...