A few questions about focal blur which are not answered in the docs.


1. What is the shape of depth of field (DOF)? Plane? I'd like to suggest 
(in future revisions) a possibility to alter the shape as in real 
lenses. However, I don't know what is the correct mathematical 
representation (paraboloid maybe?).

2. How is this DOF plane aligned? Is there allways a normal in DOF 
plane, that goes through look_at point and location or can it be aligned 
differently so that the normal goes from focal_point to camera location 
and the look_at point is not on this line?

Severi Salminen

Post a reply to this message

Severi Salminen wrote:
>A few questions about focal blur which are not answered in the docs.
>
>
>1. What is the shape of depth of field (DOF)? Plane? I'd like to suggest
>(in future revisions) a possibility to alter the shape as in real
>lenses. However, I don't know what is the correct mathematical
>representation (paraboloid maybe?).
>
>2. How is this DOF plane aligned? Is there allways a normal in DOF
>plane, that goes through look_at point and location or can it be aligned
>differently so that the normal goes from focal_point to camera location
>and the look_at point is not on this line?
>
>Severi Salminen

You may care to study the content of this article from a SIGGRAPH
conference.

'A realistic camera model for computer graphics'

by Messrs. Kolb,Mitchell and Hanrahan

 http://portal.acm.org/citation.cfm?id=218380.218463&dl=GUIDE&dl=ACM&type=series&idx=SERIES382&part=Proceedings&WantType=Proceedings&title=International%20Conference%20on%20Computer%20Graphics%20and%20Interactive%20Techniques

It's a capability that would be a 'useful' upgrade ( for me ,anyway!) to
POV3.5 - but quite intensive to implement....and probably of not wide use.

I use the 'photon-mapping' capability of POVRAY3.5 to analyse lenses of
similar complexity to those described in that article.  I have developed
macros which take the optical prescription used to build the lens and allow
analysis of aberrations ( more easily done during the normal design process
) but more importantly the stray light from reflections at lens surfaces
and barrels.  This latter is normally extremely difficult but is relatively
straightforward in POVRAY3.5

A more general improvement to POVRAY of interest to many users would be the
capability of modeling the rotationally symettric surfaces of the form

       z = z(sphere) + Ad* r^4 +Ae*r^6+....etc

           where r^2 = x^2 + y^2

I use isosurfaces to do this....but they seem very slow to raytrace compared
with the routines used in traditional optical design programs,  typically
Welford's classic text 'Aberrations of symettrical optical systems'
>

Post a reply to this message