PFM Format Description(5) File Formats Manual PFM Format Description(5)
NAME
PFM - PFM graphic image file format
DESCRIPTION
This document describes the PFM graphic image file format as understood by the Netpbm converters pamtopfm(1)
and pfmtopam(1)
There are multiple similar formats known as PFM in the world, none of them authoritatively documented. The format described here is one
that Bryan Henderson deduced from a program he found somewhere that dealt with a 'PFM' format.
The PFM format is inspired by the Netpbm formats, and you will see lots of similarity. It is not, however, an official Netpbm format. Its
goal is not consistent with those of Netpbm formats.
The format
A PFM image is a stream of bytes. The stream consists of a header followed immediately by a raster. These two components are described
below. There are no delimeters before or after the sections as described.
PFM header
The PFM header is 3 consecutive 'lines' of ASCII text. After each line is a white space character. That character is typically a newline
character, hence the term 'line,' but doesn't have to be.
pamtopfm uses a newline in the PFM it generates.
Identifier Line
The identifier line contains the characters 'PF' or 'Pf'. PF means it's a color PFM. Pf means it's a grayscale PFM.
Dimensions Line
The dimensions line contains two positive decimal integers, separated by a blank. The first is the width of the image; the second is the
height. Both are in pixels.
Scale Factor / Endianness
The Scale Factor / Endianness line is a queer line that jams endianness information into an otherwise sane description of a scale. The
line consists of a nonzero decimal number, not necessarily an integer. If the number is negative, that means the PFM raster is little
endian. Otherwise, it is big endian. The absolute value of the number is the scale factor for the image.
The scale factor tells the units of the samples in the raster. You use somehow it along with some separately understood unit information
to turn a sample value into something meaningful, such as watts per square meter.
PFM raster
The raster is a sequence of pixels, packed one after another, with no delimiters of any kind. They are in standard Western reading order:
left to right and top to bottom within the image.
Each pixel consists of 1 or 3 samples, packed one after another, with no delimiters of any kind. 1 sample for a grayscale PFM and 3 for a
color PFM (see the Identifier Line of the PFM header).
Each sample consists of 4 consecutive bytes. The bytes represent a 32 bit string, in either big endian or little endian format, as deter-
mined by the Scale Factor / Endianness line of the PFM header. That string is an IEEE 32 bit floating point number code. Since that's the
same format that most CPUs and compiler use, you can usually just make a program use the bytes directly as a floating point number, after
taking care of the endianness variation.
netpbm documentation PFM Format Description(5)
Check Out this Related Man Page
pam(5) File Formats Manual pam(5)
NAME
pam - portable arbitrary map file format
DESCRIPTION
The PAM image format is a lowest common denominator 2 dimensional map format.
It is designed to be used for any of myriad kinds of graphics, but can theoretically be used for any kind of data that is arranged as a two
dimensional rectangular array. Actually, from another perspective it can be seen as a format for data arranged as a three dimensional
array.
This format does not define the meaning of the data at any particular point in the array. It could be red, green, and blue light intensi-
ties such that the array represents a visual image, or it could be the same red, green, and blue components plus a transparency component,
or it could contain annual rainfalls for places on the surface of the Earth. Any process that uses the PAM format must further define the
format to specify the meanings of the data.
A PAM image describes a two dimensional grid of tuples. The tuples are arranged in rows and columns. The width of the image is the number
of columns. The height of the image is the number of rows. All rows are the same width and all columns are the same height. The tuples
may have any degree, but all tuples have the same degree. The degree of the tuples is called the depth of the image. Each member of a
tuple is called a sample. A sample is an unsigned integer which represents a locus along a scale which starts at zero and ends at a cer-
tain maximum value greater than zero called the maxval. The maxval is the same for every sample in the image. The two dimensional array
of all the Nth samples of each tuple is called the Nth plane or Nth channel of the image.
Though the format does not assign any meaning to the tuple values, it does include an optional string that describes that meaning. The
contents of this string, called the tuple type, are arbitrary from the point of view of the PAM format, but users of the format may assign
meaning to it by convention so they can identify their particular implementations of the PAM format.
The Layout
A PAM file consists of a sequence of one or more PAM images. There are no data, delimiters, or padding before, after, or between images.
Each PAM image consists of a header followed immediately by a raster.
Here is an example header:
P7
WIDTH 227
HEIGHT 149
DEPTH 3
MAXVAL 255
TUPLETYPE RGB
ENDHDR
The header begins with the ASCII characters "P7" followed by newline. This is the magic number.
The header continues with an arbitrary number of lines of ASCII text. Each line ends with and is delimited by a newline character.
Each header line consists of zero or more whitespace-delimited tokens or begins with "#". If it begins with "#" it is a comment and the
rest of this specification does not apply to it.
A header line which has zero tokens is valid but has no meaning.
The type of header line is identified by its first token, which is 8 characters or less:
ENDHDR This is the last line in the header. The header must contain exactly one of these header lines.
HEIGHT The second token is a decimal number representing the height of the image (number of rows). The header must contain exactly one of
these header lines.
WIDTH The second token is a decimal number representing the width of the image (number of columns). The header must contain exactly one
of these header lines.
DEPTH The second token is a decimal number representing the depth of the image (number of planes or channels). The header must contain
exactly one of these header lines.
MAXVAL The second token is a decimal number representing the maxval of the image. The header must contain exactly one of these header
lines.
TUPLTYPE
The header may contain any number of these header lines, including zero. The rest of the line is part of the tuple type. The rest
of the line is not tokenized, but the tuple type does not include any white space immediately following TUPLTYPE or at the very end
of the line. It does not include a newline. If there are multiple TUPLTYPE header lines, the tuple type is the concatenation of
the values from each of them, separated by a single blank, in the order in which they appear in the header. If there are no TUPLE-
TYPE header lines the tuple type is the null string.
The raster consists of each row of the image, in order from top to bottom, consecutive with no delimiter of any kind between, before, or
after, rows.
Each row consists of every tuple in the row, in order from left to right, consecutive with no delimiter of any kind between, before, or
after, tuples.
Each tuple consists of every sample in the tuple, in order, consecutive with no delimiter of any kind between, before, or after, samples.
Each sample consists of an unsigned integer in pure binary format, with the most significant byte first. The number of bytes is the mini-
mum number of bytes required to represent the maxval of the image.
PAM Used For PNM (PBM, PGM, or PPM) Images
A common use of PAM images is to represent the older and more concrete PBM, PGM, and PPM images.
A PBM image is conventionally represented as a PAM image of depth 1 with maxval 1 where the one sample in each tuple is 0 to represent a
black pixel and 1 to represent a white one. The height, width, and raster bear the obvious relationship to those of the PBM image. The
tuple type for PBM images represented as PAM images is conventionally "BLACKANDWHITE".
A PGM image is conventionally represented as a PAM image of depth 1. The maxval, height, width, and raster bear the obvious relationship
to those of the PGM image. The tuple type for PGM images represented as PAM images is conventionally "GRAYSCALE".
A PPM image is conventionally represented as a PAM image of depth 3. The maxval, height, width, and raster bear the obvious relationship
to those of the PPM image. The first plane represents red, the second green, and the third blue. The tuple type for PPM images repre-
sented as PAM images is conventionally "RGB".
The Confusing Universe of Netpbm Formats
It is easy to get confused about the relationship between the PAM format and PBM, PGM, PPM, and PNM. Here is a little enlightenment:
"PNM" is not really a format. It is a shorthand for the PBM, PGM, and PPM formats collectively. It is also the name of a group of library
functions that can each handle all three of those formats.
"PAM" is in fact a fourth format. But it is so general that you can represent the same information in a PAM image as you can in a PBM,
PGM, or PPM image. And in fact a program that is designed to read PBM, PGM, or PPM and does so with a recent version of the Netpbm
library, will read an equivalent PAM image just fine and the program will never know the difference.
To confuse things more, there is a collection of library routines called the "pam" functions that read and write the PAM format, but also
read and write the PBM, PGM, and PPM formats. They do this because the latter formats are much older and more popular, so this makes it
convenient to write programs that use the newer PAM format.
SEE ALSO
pbm(5), pgm(5), ppm(5), pnm(5), libpnm(3)
31 July 2000 pam(5)