On Jan 30, 2015, at 4:38 PM, Nathan Rusch <[email protected]> wrote:
> It seems absurd, but kind of looks like its going to come down to whether you
> would rather OIIO be technically correct (as we understand it), but annoy
> people and prompt them to submit erroneous bug reports by creating images
> that look wrong in all the applications they are viewed in, or have it be
> "wrong" for the sole purpose of keeping people happy. Tough call indeed...
Head exploding...
> Is it worth getting in touch with the maintainers of libjpeg to see if they
> would stand by the comment in their source as it relates to the JFIF spec? Or
> maybe just asking The Foundry and Tweak about performing an about-face?
I'm happy to contact all three. But if they change, there will be a versionitis
problem between old and new versions of those packages. And in any case,
PhotoShop is still backwards as well, and my intuition is that my chances of
getting them to change, or to care at all, is much less than with Nuke and rv,
where at least I know people who would humor me by listening to me make a case
for it.
Sigh. I'll do some experiments to see if there's any way around this. At the
very least, I want to restrict the wrongness to be completely contained in the
JPEG read/write, and not infect the rest of OIIO (including the app side),
where aspect > 1 should certainly mean wide pixels.
Another consideration: In 6 years, we have not had a single comment about our
JPEG I/O not supporting aspect ratio or the resolution fields until this week,
so perhaps the number of people who will be annoyed by our doing it "right" may
be extremely limited, and a better solution is to make sure those few people
know the weird set of hoops to jump through to make the images right in Nuke
and rv (e.g., if you want aspect 2.0, you should ask oiiotool for 0.5).
-- lg
--
Larry Gritz
[email protected]
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
