Uzilite opened this issue on Mar 02, 2005 ยท 7 posts
Uzilite posted Wed, 02 March 2005 at 5:25 AM
I am wondering, is the falloff for Depth of Field spherical or cubic? I am guessing that it is cubic since it deals with planes but I am not certain. Can someone kindly confirm this? Thanks.
Misha883 posted Wed, 02 March 2005 at 6:59 AM
Attached Link: http://www.renderosity.com/messages.ez?Form.ShowMessage=2062929
This thread provides some good links.
Uzilite posted Wed, 02 March 2005 at 7:30 AM
The formula for DOF (provided by the links in the post) is where I actually get the assumption above but still none of the it answered my question. Thanks though.
JordyArt posted Wed, 02 March 2005 at 12:57 PM
Forgive my bluntness at you clever types, but FFS.... who cares about HOW it falls off as long as it looks good when you take the picture?!? I didn't know DOF could even fall off - surely you have the picture frame upside down or summik?
ROFL.... ;-)
(".)
Message edited on: 03/02/2005 13:00
Uzilite posted Thu, 03 March 2005 at 7:20 AM
Well some people do care to know so if you can't contribute anything to this thread then kindly post somewhere else. Thanks.
JordyArt posted Thu, 03 March 2005 at 12:58 PM
Wus, I DO post somewhere else - erm.... nearly every thread... see the ROFL? See the grin? Me not serious, me injecting humour. And as for me 'kindly posting' don't bank on it - I'm only kind to nice people. mwuuuaaaaahhhhhhh!!!! (",) btw, Misha cares coz it's technical and he's a clever bugger - the only other person that was interested enough to post was me. And apart from yourself I've now posted the most. So there. lol
Misha883 posted Fri, 04 March 2005 at 7:50 AM
[Be nice to Jordy. He serves an important purpose here.] I guess one would "care" if they were designing a 3D program to simulate more realistic camera shots. Or, if they were designing lenses or cameras. All worthwhile ambitions. I suspect the answer is in the mathematics in the above links, but 'prolly not stated as a ^2 or ^3 as this thinking is more related to things like light falloff. With DOF, the "circle of confusion" is going to get bigger as distances go back, and forward!, from the exact focus point. I'm guessing, but as a first approximation I'd bet that the radius increases linearly as the distance moves away from exact focus. For 3D modeling, one could perhaps use a Gaussian Blur, with the radius varying linearly with distance? In reality, details like the shape of the aperature matter; resulting in distant highlights taking on a funky shape. I also wonder if the new "digital" lenses have different DOF characteristics because of the colluminating rear element?