forked from barak/minidjvu
-
Notifications
You must be signed in to change notification settings - Fork 6
/
NEWS
160 lines (131 loc) · 4.96 KB
/
NEWS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
---
0.9m07
---
Allow directory as an input command line argument. Encoder will process all
supported images (TIFF, BMP, PBM) found in it (non recursive) in alphabetical
order.
Change the way DIRM saved to bundled multipage document to workaround problems
caused by limitation on number simultaniously opened file handles.
bugfix: correct reading chunk id of the image from the settings file.
---
0.9m06
---
Use DjVuLibre's connected components detection algorithm as old contour walking
approach sometimes failed to detect characters in table rects.
Enforce lossless classification for the not-a-letter elements. Now pieces of the
graphics, like table lines, may be losslessly compressed in shared dictionary.
Empty Sjbz chunk is not saved for empty pages anymore.
bugfix: now images with same basename (but different paths) may be used in
one bundled document. Their chunk ids (constructed from basename by default)
are ensured to be unique. Settings file may be used to control page chunk id
of the corresponding image.
---
0.9m05
---
Replace classifier with a more accurate version. No ceache is longer
used in it so RAM consumption become smaller. -C option has been removed.
So now only one classifier is supported.
bugfix: missing AT&T id in single-page djvus
bugfix: long horisontal lines was splitted to a too small pieces
bugfix: Fixed few minor bugs
---
0.9m04
---
Fixed a bug with case-sensitive filesystems
---
0.9m03
---
Multiple bugs fixed.
---
0.9m02
---
Added settings file support that may be passed via -S key. Settings file allows to
fine-tune DjVu encoding process and set per-page/per-djbz parameters.
Documentation and translation is updated.
---
0.9m01
---
A forked modified version. From now on fork version will contain version of minidjvu
that it based on + "m" + version of modification.
---
0.9
---
Relicensed to GPL3+.
---
0.8
---
Pattern matching algorithm is adjusted to prevent some undesired
substitutions (for example, quite different shapes sometimes were unified
if minidjvu could somehow link them with a chunk of intermediate forms).
bugfix: specifying aggression level above 100 actually resulted in producing
larger files.
Letter averaging now works also in multipage mode.
minidjvu now produces multipage DJVU documents (either bundled or indirect)
when processing images in the multipage mode.
A new file naming convention for the multipage mode: pages are now generated with the
same base names as the original images, just the extension is changed to
'djvu'. A numerical suffix is added to the base name in case of a name clash.
So the 'pattern' parameter is gone: instead the same command line syntax is used
both for single page and multipage mode.
Localization via gettext (currently doesn't work under Windows) implemented
and a Russian man page added.
Configure scripts rewritten from scratch and added to the source tree.
The minidjvu library has been separated from the executable (once again).
However under Windows (mingw32) a single statically linked binary is generated
by default.
Multipage TIFF files now directly supported.
The Windows installable package is now called just "minidjvu", as all "friends" are gone.
---
0.7
---
bugfix: no more attempts to substitute pictures on multipage compression
bugfix: no longer writes full paths to INCL chunks
Memory leaks (hopefully) eliminated.
Slightly increased aggression.
A javascripted page illustrating using the most common options;
under Windows, that will be a GUI (not too pretty, but better than nothing).
---
0.6
---
bugfix: new pattern matcher works multipagely
Letter averaging. Only single-page for now.
(I'm still not completely convinced that it's the Right Thing.)
pages-per-dict now defaults to 5.
Directory layout changed.
Configure scripts missing. I wonder whether they really should be back.
(All that's because I have to use Windows due to hardware crash.)
---
0.5
---
Improved pattern matcher.
---
0.4
---
bugfix: no more "some weird error happened" messages on color TIFFs
Binaries are built in bin/ now, where they should be
Experimental multipage compression (cannot bundle pages yet, however)
----
0.33
----
bugfix: no more crashes on pages with gray background
bugfix: configure script now works with gcc3 (renamed all to C++)
---
0.3
---
Lossy compression
New "erosion" algorithm (gains about 10% in file size)
TIFF support (through libtiff)
Command-line interface improved
man page added
---
0.2
---
Added lossless compression (runs slow)
Added "smooth" filter (gains about 5% in file size)
Added minimal error-handling
Documentation is now nonzero!
---
0.1
---
Initial release.
Supports only decompression of black-and-white single-page DjVu files.