Happy new year!
Indeed, I've too noticed something bizarre about the fisheye option. The
typical fisheye spherical distortion doesn't seem quite right. But when
projected on a dome it looks great; which is most likely due to there is no
in-shot context to judge otherwise by.
I believe what might be happening is this. I've learned to change the FOV
from 1 to .8 otherwise there is some strange edge bubbling (which can only
be experience with an frame sequence). Perhaps the edge bubbling is due to
some projection imbalance with the rays of the fisheye at the image
corners. It may have something to do with the implementation of the
fisheye. So to combat this, I change the FOV to .8 and therefore limit the
field of view and *I think* affectively crop the fisheye... so the fisheye
distortion is slightly incorrect. Just a theory... I may be way off!
Next time I am in Mandelbulber, I will do a comparison between fisheye
and equirectangular projection. Then convert the equirectangular projection
to fisheye and see what looks different.
So to have a completely 180 degrees field of view, the equirectangular
projection would need to be used.
Charles Hayden Planetarium
On Thu, Dec 22, 2011 at 9:39 AM, Pete Carss <dmtquarter@...> wrote:
I've just downloaded and compiled 1.08 - quite excited by the prospect
of a proper fisheye option:
Seems to me the fisheye is still wrong, although I'm not completely
sure. I'm just about to try a diff between the sources.
The issues are more apparent with a moving sequence - I've posted a
test online - this was with the FOV set to 0.8. I masked it for rear
truncation, so the edges didn't confuse the issue:
On 8 December 2011 15:23, Michael Narlock <mnarlock@...> wrote:
Is the equirectangular output available in 1.08?
The equirectangular projection is still there.