[Bf-committers] DOF camera parameters.

[Jonathan Merritt]

Alexander Ewering wrote:

>The biggest problem the circle-of-confusion method has, in my opinion,
>is that it is awfully slow.
>  
>

LOL! :-)  Yes.  That's also my biggest concern.  DOF rendering in Aqsis 
is also very slow at the moment, but I suspect Aqsis would win out in 
the speed-quality race over the suggested circle-of-confusion method for 
Blender.  It was just that when I saw the question, I though: "oh... I 
could implement that!".


Alejandro Conty Estevez wrote:

>For that reason I have proposed to move it to blender, as some kind of
>filter somewhere. Does that make sense? I'm just commenting it in case
>it is relevant for this discussion.
>  
>

For me, the relevance is that you'd need to store the depth parameters 
somewhere, and I'd personally like those to be stored with the camera.

Another point I think is important is that whatever parameters are 
stored (with the camera or elsewhere) should be fully described 
mathematically, so that they can easily be used by other renderers.  So, 
for example, parameters like focal length, focal distance and f-stop are 
OK, since their meaning is well understood in physical optics.  However, 
parameters that specify factors in some custom equation for depth 
blurring really need good documentation so that they can be understood.

Jonathan Merritt.