Re: Endless gamma in eXIf 2017-0119

John Bowler <[email protected]> Wed, 8 Feb 2017 09:00:59 -0800
Newsgroups gmane.comp.graphics.png.general
Message-ID <CAP7U39-=2ZE+euNdLf+7o8OoULfcBOC-nxbYFXX3JRpFCu1tww@mail.gmail.com>
On Tue, Feb 7, 2017 at 8:43 PM, Glenn Randers-Pehrson <[email protected]> wrote:
rors (at the cost of discarding the Exif data)
>
> We still need some complex explanation of what an editor
> that recognizes eXIF/zXIF is supposed to do when it makes
> changes to the image that would render it copy-unsafe, to
> make it permissible to write a new Exif profile.

I think that is the same as the description in CIPA 2016 of what must
be done when writing JPEG, except that none of the JPEG specific stuff
is written and there is *no* PNG specific stuff.

I don't think there are any PNG specific rules imposed by including
EXIF beyond that.  Of course it is possible to make it complicated,
but why?  For example common sense should tell someone that if they
write an image where the samples are not aligned to a rectangular grid
it won't get displayed right and the PNG spec already defines how an
image is interpreted, more so than ISO-JPEG does; the only way to
change that is to use a critical chunk.

If you are referring to case where the input PNG already had EXIF data
then the decoder rules apply to that; only tags which are known to be
unrelated to the image data or which directly relate to it and which
are still valid will be written to the new EXIF data.  This is what
the rules for editors in the ISO-PNG spec says; unrecognized tags have
to be removed!

The only missing thing in what I said before (apart from concise
wording) is the definition of when a processing is *transcoding*
rather than *editing* the stream.  Like I said, everything could be
regarded as editing, but we know EXIF tags do not depend on the
precise PNG format so that is overly restrictive (recompressing IDAT
will never invalidate an EXIF chunk.)

It's probably easier for me to write it than explain it; most people
seem to regard an explanation as a route from where their thought
process is to where the other person's thought process is and that
rarely works.  A simple definition from scratch is better.

-- 
John Bowler <[email protected]>
+1 (541) 450-9885
PO BOX 3151
KERBY OR 97531-3151
USA

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot