[visionegg] SphereWindow gaussian bugfix
- From: Andrew Straw <andrew.straw@xxxxxxxxxxxxxxx>
- To: visionegg@xxxxxxxxxxxxx
- Date: Mon, 8 Sep 2003 16:20:27 +0930
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hi everyone,
I uncovered an embarrassing error: SphereMap.SphereWindow was
calculating "gaussian" windows with a slightly wrong formula which
resulted in ~30% narrower windows. The correct formula is
f(x)=exp(-x**2/(2*sigma**2)), but I had left out the factor of 2 in the
denominator. This is now fixed in CVS along with a version bump to
0.9.5b4. In data files and so on, you can correct the old "gaussian"
sigma to true sigma by multiplying by sqrt(2)/2. (With the bug, a
nominal sigma of 10.0 degrees would result in a gaussian with a true
sigma of about 7.07.)
Keep in mind this is only for SphereWindow gaussians -- 2D gaussians
and gabors made with Textures.Mask2D are unaffected. In other words,
the mouse_gabor_perspective.py demo is affected, but the
mouse_gabor_2d.py demo is not.
Cheers!
Andrew
-----BEGIN PGP SIGNATURE-----
iD8DBQE/XCa61xWcCSPVbpgRAm4gAKCbcM+/CvjrkwfUDSp/La32LVXgNgCgnph1
iHYKrvWZTxlvLu+YUXQ0S8M=
=iH0w
-----END PGP SIGNATURE-----
======================================
The Vision Egg mailing list
Archives: //www.freelists.org/archives/visionegg
Website: http://www.visionegg.org/mailinglist.html
Other related posts: