1Pnmscalefixed User Manual(0)                      Pnmscalefixed User Manual(0)
2
3
4

NAME

6       pnmscale - scale a PNM file quickly
7
8

DESCRIPTION

10       This program is part of Netpbm(1).
11
12       pnmscalefixed  is  the same thing as pnmscale except that it uses fixed
13       point arithmetic internally instead of floating point, which  makes  it
14       run faster.  In turn, it is less accurate and may distort the image.
15
16       Use  the  pnmscale  user manual with pnmscalefixed.  This document only
17       describes the difference.
18
19       pnmscalefixed uses fixed point 12 bit arithmetic.   By  contrast,  pnm‐
20       scale uses floating point arithmetic which on most machines is probably
21       24 bit precision.  This makes pnmscalefixed run faster (30%  faster  in
22       one experiment), but the imprecision can cause distortions at the right
23       and bottom edges.
24
25       The distortion takes the following form: One pixel from the edge of the
26       input  is  rendered  larger  in  the  output  than  the  scaling factor
27       requires.  Consequently, the rest of the  image  is  smaller  than  the
28       scaling  factor  requires,  because the overall dimensions of the image
29       are always as requested.  This distortion will usually be very hard  to
30       see.
31
32       pnmscalefixed  with  the  -verbose option tells you how much distortion
33       there is.
34
35       The amount of distortion depends on the size of the input image and how
36       close the scaling factor is to an integral 1/4096th.
37
38       If  the scaling factor is an exact multiple of 1/4096, there is no dis‐
39       tortion.  So, for example doubling or halving an image causes  no  dis‐
40       tortion.   But  reducing it or enlarging it by a third would cause some
41       distortion.  To consider an extreme case, scaling a 100,000  row  image
42       down  to 50,022 rows would create an output image with all of the input
43       squeezed into the top 50,000 rows, and the last row of the input copied
44       into the bottom 22 rows of output.
45
46       pnmscalefixed could probably be modified to use 16 bit or better arith‐
47       metic without losing anything.  The modification  would  consist  of  a
48       single constant in the source code.  Until there is a demonstrated need
49       for that, though, the Netpbm maintainer wants to keep the safety  cush‐
50       ion afforded by the original 12 bit precision.
51
52       pnmscalefixed does not have pnmscale's -nomix option.
53
54
55
56netpbm documentation           18 November 2000   Pnmscalefixed User Manual(0)