2010.12.14 17:55 "[Tiff] patch to tif_jpeg.c", by Dwight Kelly

2011.01.04 19:20 "Re: [Tiff] Regression in libtiff 4.0 CVS when creating a JPEG RGB contig", by Even Rouault

Lee,

Thanks, latest CVS solves the issues I've had.

I didn't realize that the initial change also made its way to the 3.9 branch. That seems a bit dangerous to me for a stable branch, no?

Best regards,

Even

Le mardi 04 janvier 2011 03:52:49, Lee Howard a écrit:

Again, please forgive the top-post.

I've made the change as suggested to both the 3.9 and HEAD (4.0) CVS branches.

Dwight,

If this does not work for you please communicate with Even to come up with a solution.

Hum, and there's also another case that doesn't work.

If I have a 2 band image as source:

$ ./tools/tiffcp 2band.tif out.tif -c jpeg -p contig JPEGLib: Bogus input colorspace. out.tif: Error, can't write strip 0.

So, I'd also suggest changing:

>       if  == PHOTOMETRIC_MINISWHITE || td->td_photometric (td->td_photometric

==

> PHOTOMETRIC_MINISBLACK)

> sp->cinfo.c.in_color_space = JCS_GRAYSCALE;

into:

if ((td->td_photometric == PHOTOMETRIC_MINISWHITE || td->td_photometric

> == PHOTOMETRIC_MINISBLACK) && td->td_samplesperpixel == 1)

> sp->cinfo.c.in_color_space = JCS_GRAYSCALE;