packages icon
This directory contains Release 1.2 of the Fuzzy Pixmap Manipulation
(FBM) library of programs.  This collection provides a complete package
for manipulating and converting color and black and whites images,
including 24bit RGB, 8bit mapped color, 8bit grayscale, and 1bit
bitmapped images.


	Unpack all files in one directory.  Edit Makefile and put in
	appropriate values for defaults.

	If you wish to have the tiff2fbm and fbm2tiff programs, first
        obtain, unpack, and compile libtiff.a, and then modify FBM 's
	Makefile to include this library.  Then type "make tiff".  Note
	that some versions of libtiff are not careful to put file
	tiffcompat.h into the intstall directory.  The tifflib is
	available from UCBVAX.BERKELEY.EDU, file pub/tiff/v3.0.tar.Z.

	If you have the Utah Raster Toolkit, the 'fbcat' program can be
	compiled with the -DRLE flag to include support for RLE files.
	Credit for RLE support to Paul Milazzo <>.
	Get the Toolkit from CS.UTAH.EDU, file pub/toolkit-2.0.tar.Z.

	Get Jef Poskanzer's PBMPLUS source from EXPO.LCS.MIT.EDU, file

File Formats

	FBM   (-F)	FBM file with 256 byte header (with title & credits)
	Sun   (-S)	Sun rasterfile with 32 byte header
	Tiff  (-T)	Aldus TIFF
	PBM   (-P)	Poskanzer format for 1bit files
	GIF   (-G)	CompuServe GIF format
	JPG   (-J)	JPEG JFIF format (cf The Independent JPEG Group)
	RLE   (-R)	Utah RLE format
	IFF   (-I)	Amiga IFF files (except HAM mode files)
	PCX   (-Z)	PC PaintBrush format for IBM
	Face  (-B)	Bennet Yee's 1bit files, used at CMU

	All programs (except fbinfo) can read any of the formats, and use
	the magic number to decide the file type.  If the input is not a
	pipe and is compressed using the net standard 'compress' program,
	the file will be uncompressed as it is read.  Note however that 
	the operation to be performed must make sense; that is you cannot
	halftone a color image, quantize a grayscale image, or resize a
	1bit file.

Files and Programs:

	README		This file
	README.lib	Lists the subroutines in libfbm.a
	Makefile	Make utility configuration file
	fbm.1		General manual entry
	*.1		Individual manual entries
	fbm.h		Defines FBM data structures

	clr2gray  Convert color to grayscale
	fbcat	  Copy image (used for format conversion)
	fbclean	  Flip isolated pixels (clean image)
	fbconv	  Arbitrary convolutions
	fbedge	  Compute derivative image (edge detection)
	fbext	  Extract region, resize, change aspect ratio
	fbgamma	  Gamma correct grayscale or color image for display
	fbhalf	  Halftone grayscale image (Blue noise, Floyd-Steinberg, etc)
	fbham	  Convert FBM 24bit RGB to Amiga HAM mode
	fbhist	  Compute histogram
	fbinfo	  Dump image header
	fbm2pod	  Convert grayscale image to Diablo graphics (!)
	fbm2tga	  Convert image to Targa format
	fbm2tiff  Convert FBM files to 1, 2, 4, 8, or 24 bit TIFF
	fbmask	  Set region to gray value
	fbmedian  Median filter 3x3 regios
	fbnorm	  Normalize image intensity / increase contrast
	fbpalet	  Replace an imager's colormap with another, remove duplicates
	fbps	  Convert to PostScript
	fbquant	  Color quantization (24 bit to 8..256 colors) Mod. Heckbert
	fbrot	  Rotate 90, 180, or 270 degrees
	fbthin	  Thin 1bit images
	fbsample  Sample a 1bit file to produce an 8bit file
	fbsharp	  Sharpen (edge enhancement) by digital Laplacian
	gray2clr  Add a "gray" colormap to a grayscale image
	idiff	  (and udiff) convert raw byte stream into byte difference
	pbm2ps	  Convert PBM file to PostScript
	pbmtitle  Add a title to a PBM file
	raw2fbm	  Convert raw file to FBM format (eg: Amiga Digiview files)
	tga2fbm   Convert Targa file to FBM format
	tiff2fbm  Convert tiff file to FBM format
	uunet2fbm Convert UseNet Facesaver file to FBM format


	Several file extension conventions are used, but none are enforced

	.Z		Compressed file, appended to other extensions

	.fbm		Standard extension for FBM format files
	.cbm		Sometimes used to mean a 24bit color FBM file

	.ip		8bit grayscale Amiga file (with 1.2 aspect ratio)
	.cp		24bit color Amiga file (with 1.2 aspect ratio)

	.1bit		Sun 1bit rastefile
	.clr		Sometimes used to mean a 8bit mapped color Sun file

	.bm		Bennet Yee's "Face" format (1bit)
	.pbm		Poskanzer 1bit format
	.gif		Compuserve GIF format

	.jpg or .jpeg	JPEG JFIF format
	.pcx		ZSoft's PCPaintbrush format
	.tif or .tiff	Aldus's TIFF format (used on NeXT)

Example Unix pipelines to view files:

	Starting with an 8 bit raw Amiga grayscale file, foo.ip, first
	create an FBM format file with a title and credit, then process
	that file for viewing on a Sun screen by normalizing, rotating,
	resizing, sharpening, and halftoning.

	% raw2fbm -d1 -a1.2 -w640 -h400 \
	  -t'Random Title' -c'Random Credit' < foo.ip > foo.fbm

	% fbinfo foo.fbm
	foo.fbm        	"Random Title"
			[ Random Credit ]
			[640x400x8]    8 physbits  1.2 aspect ratio
			row length 640, plane length 256000, colormap length 0

	% fbnorm < foo.fbm | \			(normalize)
	  fbrot | \				(rotate 90 deg clockwise)
	  fbext -a1 -W900 -H768 | \		(resize & make aspect ratio 1)
	  fbhalf -s10 -S > foo.1bit		(sharpen, halftone, write Sun)

	Starting with a 24 bit color Amiga file, bar.cp, first create an
	FBM format file with title and credit, then render that file for
	viewing on a Sun by normalizing, rotating, sharpening, and

	% raw2fbm -d3 -a1.2 -w640 -h400 \
	  -t'Random Color Title' -c'Random Color Credit' < bar.cp > bar.cbm

	% fbinfo bar.cbm
	bar.cbm        	"Random Color Title"
			[ Random Color Credit ]
			[3x640x400x8]    8 physbits  1.2 aspect ratio
			row length 640, plane length 256000, colormap length 0

	% fbnorm < bar.cbm | \			(normalize)
	  fbrot | \				(rotate 90 deg clockwise)
	  fbext -a1 -W900 -H768 | \		(resize & make aspect ratio 1)
	  fbsharp 3.0 | \			(sharpen it)
	  fbquant -S > bar.clr			(quantize and write Sun)

	(Note that sharpening is an integral option to the halftoning
	 program using the -s option, but that color images must currently
	 be sharpened using a separate program.  Note also that more
	 sharpening is used on black and white images than on color images).

      Starting with a 256 color Sun raster image, baz.clr, generate a NeXT
	readable 2bit grayscale image (note fbm2tiff is not a filter, but
	writes to a named file.  Tiff2fbm reads a named file):

			-- Without Dithering --
      % clr2gray < baz.clr | \			(convert to gray)
	fbm2tiff -N baz.tiff			(chop 8bit grays to 2bit)

			-- With Dithering (note 1) --
      % clr2gray < baz.clr | \			(convert to gray)
	fbnorm | \				(normalize, optional)
	unmap | \				(convert to RGB)
	fbquant -c4 | \				(fbquant also dithers RGB)
	clr2gray | \				(back to gray: 0, 85, 190, 255)
	fbm2tiff -N baz.tiff			(convert 4 grays to NeXT 2bit)

	Note 1: Yes, this is gross...and inefficient.  But the
	images look fine.  Dithering should be an option of fbm2tiff,
	and really the halftoning algorithms should be generalized
	to produce more than just bitmaps.

Notes on the 'fbext' program

	Usage:	fbext [ -w<width> -h<height> -W<maxwdith> -H<maxheight>
		        -s<size> -a<aspect> -t'title' -c'credits' ]
		     [ x y [ width height ] ]
		< foo.fbm > bar.fbm

	fbext (bitmap extract) is a very useful anmd multipurpose program.
	Its basic operation is to copy a rectangular portion of one image
	(color or black and white) to a second image, possibly changing
	either the width or height or both.  It can also change the title
	or credits fields of the image.  Because it is so versatile, there
	are a host of options:

	-w<num>  or  -h<num>	Force width and/or height to a certain size
	-W<num>  or  -H<num>	Limit width and height to a certain size
	-a<num>			Force the output to have a given aspect ratio
	-s<num>			Scale image to have no more than a given
				number of pixels.
	<x0> <y0> <width> <height>  select a rectangular region of the image
				to extract.  The default is the entire image.

	Note that the output size limits are cumulative.  Thus if you set a
	maximum width, height, and number of pixels, the image will be the
	smallest of the three limits.

Notes on JPEG support

	All routines support JPEG output, but not all routines actually
	produce the 8 bit grayscale or 24 bit RGB color suitable for
	writing to JPEG format.  In those cases, the JPEG routines will
	Only the fbcat(1) program supports the -q<quality> switch
	which allows you to specify a JPEG quality other than the 
	default of 75.