Re: [Lcms-user] lcms demo program src?

[John G. <gr...@ag...>]

Marti,

Your patch works well for reading in the botched adobe images, but it=20
also causes a problem.  When I asked the ijg lib to write out a CMYK=20
jpeg, it writes out the adobe markers, but writes out the componets=20
correctly.  Now when I read in image I created, it assumes that adobe=20
created, and inverts the componets.

See a work around for this?

Thanks,

John

Mart=ED Maria wrote:

> John,
>=20
> Most CMYK JPEG comes inverted. This is a long history about a Junior
> programmer that did a mistake in early versions of PhotoShop, and adobe
> trying to keep backwards compatibilty. All CMYK JPEG from adobe=20
> programs  are stored with CMYK inks inverted.=20
>=20
> In a conversation with Tom Lane, time ago, he told me that he cannot=20
> change this behaviour on ijg code, since this will break existing=20
> applications that are aware of this, and does the inversion by itselves=
.=20
>=20
> I am using a patched version of  ijg, that does return correct inks, on
> depending of adobe markers. I attach the code as well, (is very small).
>=20
> If you don't like to use modified version of a library, you can provide=
=20
> your own reversing. Is only a matter of detect if adobe marker is prese=
nt,=20
> and if so, reverse (255-x) the channels.
>=20
> About the demo sources, well, the demo is using a image processing
> library still in development and a lot more complex that the CMM itself=
,
> so I'm afraid it would be of little use.  The cmyk->srgb is only a matt=
er=20
> of opening cmyk profile on input and srgb on output, and then apply=20
> transform. Anyway, if you experience any problem  feel free to drop=20
> me a note.=20
>=20
> Regards,
> Marti.
>=20



--=20
John Gray                           gr...@ag...
AgoraNet, Inc.                      (302) 224-2475
102 E. Main Street, Suite 303       (302) 224-2552 (fax)
Newark, De 19711                    http://www.agora-net.com