Commit 31088f3c authored by BST 1998  Adam D. Moss's avatar BST 1998 Adam D. Moss Committed by Adam D. Moss
Browse files

The facility to specify the background colour of a transparent/animated

Tue Sep 15 21:27:10 BST 1998  Adam D. Moss <adam@gimp.org>

	* plug-ins/gif.c:
	The facility to specify the background colour of
	a transparent/animated GIF for non-transparent
	viewers now works very much more consistantly.
	The only situations in which it will fail to work
	as expected now are those where file size can be reduced
	(abeit not by much, as the plugin is sometimes more pessimistic
	than it need be) by re-using an existing unused colour
	index rather than using another bit per pixel in the
	encoded file.  That will never be an issue with an image
	which was freshly converted from RGB to INDEXED with the
	Quantize option, as that option removes any unused colours
	from the image.

	Let me know if you find the consistancy/size tradeoff more
	annoying than helpful and I can adjust it.  IMHO it is too
	arcane a feature to present to any user as a runtime option.

	* app/ink.c: #include <string.h> for a memset
parent f0b060c2
Tue Sep 15 21:27:10 BST 1998 Adam D. Moss <adam@gimp.org>
* plug-ins/gif.c:
The facility to specify the background colour of
a transparent/animated GIF for non-transparent
viewers now works very much more consistantly.
The only situations in which it will fail to work
as expected now are those where file size can be reduced
(abeit not by much, as the plugin is sometimes more pessimistic
than it need be) by re-using an existing unused colour
index rather than using another bit per pixel in the
encoded file. That will never be an issue with an image
which was freshly converted from RGB to INDEXED with the
Quantize option, as that option removes any unused colours
from the image.
Let me know if you find the consistancy/size tradeoff more
annoying than helpful and I can adjust it. IMHO it is too
arcane a feature to present to any user as a runtime option.
* app/ink.c: #include <string.h> for a memset
Mon Sep 14 18:51:54 1998 Jay Cox (jaycox@earthlink.net)
* app/gimpbrushgenerated.c app/gimpbrushgenerated.h
......
......@@ -31,6 +31,7 @@
#include <math.h>
#include <stdlib.h>
#include <string.h>
#define SUBSAMPLE 8
......
......@@ -31,6 +31,7 @@
#include <math.h>
#include <stdlib.h>
#include <string.h>
#define SUBSAMPLE 8
......
......@@ -31,6 +31,7 @@
#include <math.h>
#include <stdlib.h>
#include <string.h>
#define SUBSAMPLE 8
......
......@@ -31,6 +31,7 @@
#include <math.h>
#include <stdlib.h>
#include <string.h>
#define SUBSAMPLE 8
......
......@@ -7,7 +7,7 @@
* Based around original GIF code by David Koblas.
*
*
* Version 2.0.3 - 98/05/18
* Version 2.0.4 - 98/09/15
* Adam D. Moss - <adam@gimp.org> <adam@foxbox.org>
*/
/*
......@@ -24,6 +24,25 @@
* REVISION HISTORY
*
*
* 98/09/15
* 2.00.04 - The facility to specify the background colour of
* a transparent/animated GIF for non-transparent
* viewers now works very much more consistantly.
*
* The only situations in which it will fail to work
* as expected now are those where file size can be reduced
* (abeit not by much, as the plugin is sometimes more pessimistic
* than it need be) by re-using an existing unused colour
* index rather than using another bit per pixel in the
* encoded file. That will never be an issue with an image
* which was freshly converted from RGB to INDEXED with the
* Quantize option, as that option removes any unused colours
* from the image.
*
* Let me know if you find the consistancy/size tradeoff more
* annoying than helpful and I can adjust it. IMHO it is too
* arcane a feature to present to any user as a runtime option.
*
* 98/05/18
* 2.00.03 - If we did manage to decode at least one frame of a
* gif, be sure to display the resulting image even if
......@@ -184,8 +203,6 @@
/*
* TODO (more *'s means more important!)
*
* - animated GIF optimization routines (seperate plugin)
*
* - Show GIF comments in a popup window?
*
* - PDB stuff for comments
......@@ -201,11 +218,8 @@
*
* - Remove unused colourmap entries for GRAYSCALE images.
*
* - All the warning messages etc. which the plugin prints to STDOUT should
* pop up in windows.
*
* - * Button to activate the animation preview plugin from the GIF-save
* dialog.
* - Button to activate the animation preview plugin from the GIF-save
* dialog.
*
*/
......@@ -1378,13 +1392,16 @@ typedef long int count_int;
static int find_unused_ia_colour (guchar *pixels,
int numpixels);
int numpixels,
int num_indices,
int* colors);
void special_flatten_indexed_alpha (guchar *pixels,
int *transparent,
int *colors,
int numpixels);
static int colorstobpp (int);
static int bpptocolors (int);
static int GetPixel (int, int);
static void BumpPixel (void);
static int GIFNextPixel (ifunptr);
......@@ -1420,30 +1437,46 @@ static void flush_char (void);
static int find_unused_ia_colour (guchar *pixels,
int numpixels)
int numpixels,
int num_indices,
int* colors)
{
int i;
gboolean ix_used[256];
/*
g_message ("GIF: Image has >=256 colors - attempting to reduce...\n");
*/
g_print ("GIF: fuiac: Image claims to use %d/%d indices - finding free "
"index...\n", (int)(*colors),(int)num_indices);
for (i=0;i<256;i++)
ix_used[i] = (gboolean)FALSE;
for (i=0; i<256; i++)
{
ix_used[i] = (gboolean)FALSE;
}
for (i=0;i<numpixels;i++)
if (pixels[i*2+1]) ix_used[pixels[i*2]] = (gboolean)TRUE;
for (i=0; i<numpixels; i++)
{
if (pixels[i*2+1]) ix_used[pixels[i*2]] = (gboolean)TRUE;
}
for (i=0;i<256;i++)
if (ix_used[i] == (gboolean)FALSE)
{
g_print ("GIF: Found unused colour index %d.\n",(int)i);
return i;
}
for (i=num_indices-1; i>=0; i--)
{
if (ix_used[i] == (gboolean)FALSE)
{
g_print ("GIF: Found unused colour index %d.\n",(int)i);
return i;
}
}
/* Couldn't find an unused colour index within the number of
bits per pixel we wanted. Will have to increment the number
of colours in the image and assign a transparent pixel there. */
if ((*colors) < 256)
{
(*colors)++;
g_print ("GIF: 2nd pass - Increasing bounds and using colour index %d.\n"
, (int) (*colors)-1);
return ((*colors)-1);
}
g_message ("GIF: Couldn't simply reduce colours further. Saving as opaque.\n");
return (-1);
}
......@@ -1478,14 +1511,15 @@ special_flatten_indexed_alpha (guchar *pixels,
pixels[i] = pixels[i*2];
}
}
if ((*colors) < 256)
(*colors) += 1;
/* commented out - this is now done in find_unused.. where neces. */
/* if ((*colors) < 256)
(*colors) += 1; */
}
/* Pixel data now takes half as much space (the alpha data has been
discarded) */
/* pixels = g_realloc (pixels, numpixels);*/
/* pixels = g_realloc (pixels, numpixels);*/
}
......@@ -1620,7 +1654,7 @@ save_image (char *filename,
int Blue[MAXCOLORS];
guchar *cmap;
guint rows, cols;
int BitsPerPixel;
int BitsPerPixel, liberalBPP, useBPP;
int colors;
char *temp_buf;
int i;
......@@ -1714,13 +1748,19 @@ save_image (char *filename,
if (colors < 256)
{
BitsPerPixel = colorstobpp (colors +
((drawable_type==INDEXEDA_IMAGE) ? 1 : 0));
/* we keep track of how many bits we promised to have in liberalBPP,
so that we don't accidentally come under this when doing
clever transparency stuff where we can re-use wasted indices. */
liberalBPP = BitsPerPixel =
colorstobpp (colors + ((drawable_type==INDEXEDA_IMAGE) ? 1 : 0));
}
else
{
BitsPerPixel = colorstobpp (256);
g_print ("GIF: Too many colours?\n");
if (drawable_type==INDEXEDA_IMAGE)
{
g_print ("GIF: Too many colours?\n");
}
}
cols = gimp_image_width(image_ID);
......@@ -1780,7 +1820,9 @@ save_image (char *filename,
transparent =
find_unused_ia_colour(pixels,
drawable->width * drawable->height);
drawable->width * drawable->height,
bpptocolors(colorstobpp(colors)),
&colors);
special_flatten_indexed_alpha (pixels,
&transparent,
......@@ -1793,6 +1835,18 @@ save_image (char *filename,
BitsPerPixel = colorstobpp (colors);
if (BitsPerPixel != liberalBPP)
{
/* We were able to re-use an index within the existing bitspace,
whereas the estimate in the header was pessimistic but still
needs to be upheld... */
g_warning("Promised %d bpp, pondered writing chunk with %d bpp!\n"
"Transparent colour may be incorrect on viewers which"
" don't support transparency.\n",
liberalBPP, BitsPerPixel);
}
useBPP = (BitsPerPixel > liberalBPP) ? BitsPerPixel : liberalBPP;
if (is_gif89)
{
......@@ -1816,13 +1870,16 @@ save_image (char *filename,
GIFEncodeGraphicControlExt (outfile, Disposal, Delay89, nlayers,
cols, rows, gsvals.interlace, 0,
transparent, BitsPerPixel, Red, Green,
transparent,
useBPP,
Red, Green,
Blue, GetPixel);
}
GIFEncodeImageData (outfile, cols, rows, gsvals.interlace, 0,
transparent,
BitsPerPixel, Red, Green, Blue, GetPixel,
useBPP,
Red, Green, Blue, GetPixel,
offset_x, offset_y);
......@@ -1837,7 +1894,7 @@ save_image (char *filename,
g_free(layers);
GIFEncodeClose (outfile, cols, rows, gsvals.interlace, 0, transparent,
BitsPerPixel, Red, Green, Blue, GetPixel);
useBPP, Red, Green, Blue, GetPixel);
return TRUE;
}
......@@ -2157,7 +2214,7 @@ colorstobpp (int colors)
bpp = 8;
else
{
g_message ("GIF: too many colors: %d\n", colors);
g_warning ("GIF: colorstobpp - Eep! too many colors: %d\n", colors);
return 8;
}
......@@ -2165,6 +2222,26 @@ colorstobpp (int colors)
}
static int
bpptocolors (int bpp)
{
int colors;
if (bpp>8)
{
g_warning ("GIF: bpptocolors - Eep! bpp==%d !\n", bpp);
return 256;
}
colors = 1 << bpp;
/* printf(">> %dbpp is %d colours <<\n", bpp, colors); */
return (colors);
}
static int
GetPixel (int x,
int y)
......
......@@ -7,7 +7,7 @@
* Based around original GIF code by David Koblas.
*
*
* Version 2.0.3 - 98/05/18
* Version 2.0.4 - 98/09/15
* Adam D. Moss - <adam@gimp.org> <adam@foxbox.org>
*/
/*
......@@ -24,6 +24,25 @@
* REVISION HISTORY
*
*
* 98/09/15
* 2.00.04 - The facility to specify the background colour of
* a transparent/animated GIF for non-transparent
* viewers now works very much more consistantly.
*
* The only situations in which it will fail to work
* as expected now are those where file size can be reduced
* (abeit not by much, as the plugin is sometimes more pessimistic
* than it need be) by re-using an existing unused colour
* index rather than using another bit per pixel in the
* encoded file. That will never be an issue with an image
* which was freshly converted from RGB to INDEXED with the
* Quantize option, as that option removes any unused colours
* from the image.
*
* Let me know if you find the consistancy/size tradeoff more
* annoying than helpful and I can adjust it. IMHO it is too
* arcane a feature to present to any user as a runtime option.
*
* 98/05/18
* 2.00.03 - If we did manage to decode at least one frame of a
* gif, be sure to display the resulting image even if
......@@ -184,8 +203,6 @@
/*
* TODO (more *'s means more important!)
*
* - animated GIF optimization routines (seperate plugin)
*
* - Show GIF comments in a popup window?
*
* - PDB stuff for comments
......@@ -201,11 +218,8 @@
*
* - Remove unused colourmap entries for GRAYSCALE images.
*
* - All the warning messages etc. which the plugin prints to STDOUT should
* pop up in windows.
*
* - * Button to activate the animation preview plugin from the GIF-save
* dialog.
* - Button to activate the animation preview plugin from the GIF-save
* dialog.
*
*/
......@@ -1378,13 +1392,16 @@ typedef long int count_int;
static int find_unused_ia_colour (guchar *pixels,
int numpixels);
int numpixels,
int num_indices,
int* colors);
void special_flatten_indexed_alpha (guchar *pixels,
int *transparent,
int *colors,
int numpixels);
static int colorstobpp (int);
static int bpptocolors (int);
static int GetPixel (int, int);
static void BumpPixel (void);
static int GIFNextPixel (ifunptr);
......@@ -1420,30 +1437,46 @@ static void flush_char (void);
static int find_unused_ia_colour (guchar *pixels,
int numpixels)
int numpixels,
int num_indices,
int* colors)
{
int i;
gboolean ix_used[256];
/*
g_message ("GIF: Image has >=256 colors - attempting to reduce...\n");
*/
g_print ("GIF: fuiac: Image claims to use %d/%d indices - finding free "
"index...\n", (int)(*colors),(int)num_indices);
for (i=0;i<256;i++)
ix_used[i] = (gboolean)FALSE;
for (i=0; i<256; i++)
{
ix_used[i] = (gboolean)FALSE;
}
for (i=0;i<numpixels;i++)
if (pixels[i*2+1]) ix_used[pixels[i*2]] = (gboolean)TRUE;
for (i=0; i<numpixels; i++)
{
if (pixels[i*2+1]) ix_used[pixels[i*2]] = (gboolean)TRUE;
}
for (i=0;i<256;i++)
if (ix_used[i] == (gboolean)FALSE)
{
g_print ("GIF: Found unused colour index %d.\n",(int)i);
return i;
}
for (i=num_indices-1; i>=0; i--)
{
if (ix_used[i] == (gboolean)FALSE)
{
g_print ("GIF: Found unused colour index %d.\n",(int)i);
return i;
}
}
/* Couldn't find an unused colour index within the number of
bits per pixel we wanted. Will have to increment the number
of colours in the image and assign a transparent pixel there. */
if ((*colors) < 256)
{
(*colors)++;
g_print ("GIF: 2nd pass - Increasing bounds and using colour index %d.\n"
, (int) (*colors)-1);
return ((*colors)-1);
}
g_message ("GIF: Couldn't simply reduce colours further. Saving as opaque.\n");
return (-1);
}
......@@ -1478,14 +1511,15 @@ special_flatten_indexed_alpha (guchar *pixels,
pixels[i] = pixels[i*2];
}
}
if ((*colors) < 256)
(*colors) += 1;
/* commented out - this is now done in find_unused.. where neces. */
/* if ((*colors) < 256)
(*colors) += 1; */
}
/* Pixel data now takes half as much space (the alpha data has been
discarded) */
/* pixels = g_realloc (pixels, numpixels);*/
/* pixels = g_realloc (pixels, numpixels);*/
}
......@@ -1620,7 +1654,7 @@ save_image (char *filename,
int Blue[MAXCOLORS];
guchar *cmap;
guint rows, cols;
int BitsPerPixel;
int BitsPerPixel, liberalBPP, useBPP;
int colors;
char *temp_buf;
int i;
......@@ -1714,13 +1748,19 @@ save_image (char *filename,
if (colors < 256)
{
BitsPerPixel = colorstobpp (colors +
((drawable_type==INDEXEDA_IMAGE) ? 1 : 0));
/* we keep track of how many bits we promised to have in liberalBPP,
so that we don't accidentally come under this when doing
clever transparency stuff where we can re-use wasted indices. */
liberalBPP = BitsPerPixel =
colorstobpp (colors + ((drawable_type==INDEXEDA_IMAGE) ? 1 : 0));
}
else
{
BitsPerPixel = colorstobpp (256);
g_print ("GIF: Too many colours?\n");
if (drawable_type==INDEXEDA_IMAGE)
{
g_print ("GIF: Too many colours?\n");
}
}
cols = gimp_image_width(image_ID);
......@@ -1780,7 +1820,9 @@ save_image (char *filename,
transparent =
find_unused_ia_colour(pixels,
drawable->width * drawable->height);
drawable->width * drawable->height,
bpptocolors(colorstobpp(colors)),
&colors);
special_flatten_indexed_alpha (pixels,
&transparent,
......@@ -1793,6 +1835,18 @@ save_image (char *filename,
BitsPerPixel = colorstobpp (colors);
if (BitsPerPixel != liberalBPP)
{
/* We were able to re-use an index within the existing bitspace,
whereas the estimate in the header was pessimistic but still
needs to be upheld... */
g_warning("Promised %d bpp, pondered writing chunk with %d bpp!\n"
"Transparent colour may be incorrect on viewers which"
" don't support transparency.\n",
liberalBPP, BitsPerPixel);
}
useBPP = (BitsPerPixel > liberalBPP) ? BitsPerPixel : liberalBPP;
if (is_gif89)
{
......@@ -1816,13 +1870,16 @@ save_image (char *filename,
GIFEncodeGraphicControlExt (outfile, Disposal, Delay89, nlayers,
cols, rows, gsvals.interlace, 0,
transparent, BitsPerPixel, Red, Green,
transparent,
useBPP,
Red, Green,
Blue, GetPixel);
}
GIFEncodeImageData (outfile, cols, rows, gsvals.interlace, 0,
transparent,
BitsPerPixel, Red, Green, Blue, GetPixel,
useBPP,
Red, Green, Blue, GetPixel,
offset_x, offset_y);
......@@ -1837,7 +1894,7 @@ save_image (char *filename,
g_free(layers);
GIFEncodeClose (outfile, cols, rows, gsvals.interlace, 0, transparent,
BitsPerPixel, Red, Green, Blue, GetPixel);
useBPP, Red, Green, Blue, GetPixel);
return TRUE;
}
......@@ -2157,7 +2214,7 @@ colorstobpp (int colors)
bpp = 8;
else
{
g_message ("GIF: too many colors: %d\n", colors);
g_warning ("GIF: colorstobpp - Eep! too many colors: %d\n", colors);
return 8;
}
......@@ -2165,6 +2222,26 @@ colorstobpp (int colors)
}
static int
bpptocolors (int bpp)
{
int colors;
if (bpp>8)
{
g_warning ("GIF: bpptocolors - Eep! bpp==%d !\n", bpp);
return 256;
}
colors = 1 << bpp;
/* printf(">> %dbpp is %d colours <<\n", bpp, colors); */
return (colors);
}
static int
GetPixel (int x,
int y)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment