8-bit JPEG broken in 3.9.x?

[Dwight Kelly]

Hello,

I am getting strange results with JPEG compression and libtiff 3.9.1.
Below is a test program. The output for 8-bit JPEG in gray and RGB
colorspaces are wrong. Instead of horizontal stripes I get a solid color.

---
Dwight Kelly
Apago, Inc.


#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "tiffio.h"

unsigned char *
create_test_image(uint16 samples, uint16 bitdepth, uint32 wpix, uint32 hpix)
{
        uint32 stride = ((samples * bitdepth * wpix)+7) / 8;
        unsigned char *buf = (unsigned char*)malloc(stride * hpix);
        unsigned char *bp = buf;
        int state = 1;
        uint32 h, qtrh = hpix/4;
        for (h=0;h<hpix;h++) {
                if (h % qtrh == 0)
                        state = !state;
                memset(bp, (state)?0:0xff, stride);
                bp += stride;
        }
        return buf;
}

int
main(int argc, char **argv)
{
        static uint16 photometrics[] = {
                PHOTOMETRIC_MINISWHITE, PHOTOMETRIC_MINISWHITE,
PHOTOMETRIC_MINISWHITE, PHOTOMETRIC_MINISWHITE, PHOTOMETRIC_MINISWHITE,
                PHOTOMETRIC_MINISWHITE, PHOTOMETRIC_MINISWHITE,
PHOTOMETRIC_MINISWHITE, PHOTOMETRIC_MINISWHITE,
                PHOTOMETRIC_RGB, PHOTOMETRIC_RGB, PHOTOMETRIC_RGB,
PHOTOMETRIC_RGB
        };
        static uint16 bitdepths[] = {
                1, 1, 1, 1, 1,
                8, 8, 8, 8,
                8, 8, 8, 8
        };
        static uint16 compressions[] = {
                COMPRESSION_ADOBE_DEFLATE, COMPRESSION_PACKBITS,
COMPRESSION_LZW, COMPRESSION_CCITTFAX3, COMPRESSION_CCITTFAX4,
                COMPRESSION_ADOBE_DEFLATE, COMPRESSION_PACKBITS,
COMPRESSION_LZW, COMPRESSION_JPEG,
                COMPRESSION_ADOBE_DEFLATE, COMPRESSION_PACKBITS,
COMPRESSION_LZW, COMPRESSION_JPEG
        };

        int i;
        for (i=0; i < sizeof(photometrics)/sizeof(uint16); i++) {
                char *cspace = "unknown";
                if (photometrics[i] == PHOTOMETRIC_MINISWHITE ||
photometrics[i] == PHOTOMETRIC_MINISBLACK) {
                        if (bitdepths[i] == 1)
                                cspace = "mono";
                        else
                                cspace = "gray";
                } else if (photometrics[i] == PHOTOMETRIC_RGB) {
                        cspace = "rgb";
                } else if (photometrics[i] == PHOTOMETRIC_SEPARATED) {
                        cspace = "cmyk";
                }

                char fn[256];
                sprintf(fn, "test_%s_%d_%d.tif", cspace, bitdepths[i],
compressions[i]);

                TIFF *tif = TIFFOpen(fn, "w");
                if (!tif) {
                        puts("ERROR: TIFFOpen failed\n");
                        return 1;
                }

                uint32 hpix = 512, wpix = 512;
                uint16 samples = 1;
                if (photometrics[i] == PHOTOMETRIC_RGB)
                        samples = 3;
                else if (photometrics[i] == PHOTOMETRIC_SEPARATED)
                        samples = 4;
                unsigned char *img = create_test_image(samples,
bitdepths[i], wpix, hpix);

                TIFFSetField(tif, TIFFTAG_PLANARCONFIG,
PLANARCONFIG_CONTIG);
                TIFFSetField(tif, TIFFTAG_PHOTOMETRIC, photometrics[i]);
                TIFFSetField(tif, TIFFTAG_SAMPLESPERPIXEL, samples);
                TIFFSetField(tif, TIFFTAG_BITSPERSAMPLE, bitdepths[i]);
                TIFFSetField(tif, TIFFTAG_COMPRESSION, compressions[i]);
                TIFFSetField(tif, TIFFTAG_IMAGEWIDTH, wpix);
                TIFFSetField(tif, TIFFTAG_IMAGELENGTH, hpix);

                uint32 h;
                unsigned char *ip = img;
                for (h=0; h<hpix; h++) {
                        if (TIFFWriteScanline(tif, ip, h, 0) < 0) {
                                puts("ERROR: TIFFWriteScanline failed\n");
                                return 1;
                        }
                        ip += TIFFScanlineSize(tif);
                }

                free(img);
                TIFFClose(tif);
        }
        return 0;
}

_______________________________________________
Tiff mailing list: Tiff@...
http://lists.maptools.org/mailman/listinfo/tiff
http://www.remotesensing.org/libtiff/

[Bob Friesenhahn]

On Mon, 14 Sep 2009, Dwight Kelly wrote:

> I am getting strange results with JPEG compression and libtiff 3.9.1.
> Below is a test program. The output for 8-bit JPEG in gray and RGB
> colorspaces are wrong. Instead of horizontal stripes I get a solid color.

I received your two emails.  Double-check how you are building your
test program.  If you are on a system which supports it, use 'ldd' to
verify that the correct libraries are being used.  Pay close attention
that the correct header files are used.  For my first attempt with
your test program, a system provided libtiff was being used rather
than the one I installed.

With your test program, the results all look the same when viewed
using GraphicsMagick with libtiff 3.9.1 plus the tiny patch for OJPEG
to work with IJG JPEG 7.

Bob
--
Bob Friesenhahn
bfriesen@..., http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/
_______________________________________________
Tiff mailing list: Tiff@...
http://lists.maptools.org/mailman/listinfo/tiff
http://www.remotesensing.org/libtiff/

[Frank Warmerdam-2]

Bob Friesenhahn wrote:

> On Mon, 14 Sep 2009, Dwight Kelly wrote:
>
>> I am getting strange results with JPEG compression and libtiff 3.9.1.
>> Below is a test program. The output for 8-bit JPEG in gray and RGB
>> colorspaces are wrong. Instead of horizontal stripes I get a solid color.
>
> I received your two emails.  Double-check how you are building your
> test program.  If you are on a system which supports it, use 'ldd' to
> verify that the correct libraries are being used.  Pay close attention
> that the correct header files are used.  For my first attempt with
> your test program, a system provided libtiff was being used rather
> than the one I installed.

Folks,

At my request Dwight has also filed this in bugzilla as:

   http://bugzilla.maptools.org/show_bug.cgi?id=2097

I also found that things looked fine, though Dwight seems to see a problem
with an output file that looks good to me so there may be some sort of viewer
sensitivity.

Best regards,
--
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam@...
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent

_______________________________________________
Tiff mailing list: Tiff@...
http://lists.maptools.org/mailman/listinfo/tiff
http://www.remotesensing.org/libtiff/