Alternate proposal to https://github.com/OpenImageIO/oiio/pull/196
I think this is equivalent but is a bit less mysterious.
The idea here is that the filtering ellipse can be so thin that the minorlength 
is significantly subpixel sized, even at the finest MIP level, which will lead 
to a high nsamples pointlessly -- the samples will be too close together.  So 
this detects that case and clamps minorlength to be at least 1/2 texel wide, 
and readjusts aspect, which in turn will end up leading to a sane number of 
samples along the ellipse axis.


You can merge this Pull Request by running:

  git pull https://github.com/lgritz/oiio ramon2

Or you can view, comment on it, or merge it online at:

  https://github.com/OpenImageIO/oiio/pull/229

-- Commit Summary --

* More efficient subpixel filtering

-- File Changes --

M src/libtexture/texturesys.cpp (7)

-- Patch Links --

  https://github.com/OpenImageIO/oiio/pull/229.patch
  https://github.com/OpenImageIO/oiio/pull/229.diff

--- 
Reply to this email directly or view it on GitHub:
https://github.com/OpenImageIO/oiio/pull/229
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org

Reply via email to