Penguin
Recent Changes
Blame: pnmscalefixed(1)
EditPageHistoryDiffInfoLikePages
Annotated edit history of pnmscalefixed(1) version 1, including all changes. View license author blame.
Rev Author # Line
1 perry 1 pnmscalefixed
2 !!!pnmscalefixed
3 NAME
4 DESCRIPTION
5 ----
6 !!NAME
7
8
9 pnmscale - scale a PNM file quickly
10 !!DESCRIPTION
11
12
13 __pnmscalefixed__ is the same thing as __pnmscale__
14 except that it uses fixed point arithmetic internally
15 instead of floating point, which makes it run faster. In
16 turn, it is less accurate and may distort the
17 image.
18
19
20 Use the __pnmscale__ man page with __pnmscalefixed__.
21 This man page only describes the difference.
22
23
24 __pnmscalefixed__ uses fixed point 12 bit arithmetic. By
25 contrast, __pnmscale__ uses floating point arithmetic
26 which on most machines is probably 24 bit precision. This
27 makes __pnmscalefixed__ run faster (30% faster in one
28 experiment), but the imprecision can cause distortions at
29 the right and bottom edges.
30
31
32 The distortion takes the following form: One pixel from the
33 edge of the input is rendered larger in the output than the
34 scaling factor requires. Consequently, the rest of the image
35 is smaller than the scaling factor requires, because the
36 overall dimensions of the image are always as requested.
37 This distortion will usually be very hard to
38 see.
39
40
41 __pnmscalefixed__ with the __-verbose__ option tells
42 you how much distortion there is.
43
44
45 The amount of distortion depends on the size of the input
46 image and how close the scaling factor is to an integral
47 1/4096th.
48
49
50 If the scaling factor is an exact multiple of 1/4096, there
51 is no distortion. So, for example doubling or halving an
52 image causes no distortion. But reducing it or enlarging it
53 by a third would cause some distortion. To consider an
54 extreme case, scaling a 100,000 row image down to 50,022
55 rows would create an output image with all of the input
56 squeezed into the top 50,000 rows, and the last row of the
57 input copied into the bottom 22 rows of output.
58
59
60 __pnmscalefixed__ could probably be modified to use 16
61 bit or better arithmetic without losing anything. The
62 modification would consist of a single constant in the
63 source code. Until there is a demonstrated need for that,
64 though, the Netpbm maintainer wants to keep the safety
65 cushion afforded by the original 12 bit
66 precision.
67 ----
This page is a man page (or other imported legacy content). We are unable to automatically determine the license status of this page.