README
1 MPEG Video Software Statistics Gatherer
2 (Version 2.2; Feb 1, 1995)
3
4 Lawrence A. Rowe, Steve Smoot, Ketan Patel, and Brian Smith
5 Computer Science Division-EECS, Univ. of Calif. at Berkeley
6
7This directory contains a public domain MPEG video statistics gatherer.
8The decoder implements the standard described in the Committee
9Draft ISO/IEC CD 11172 dated December 6, 1991 which is
10sometimes referred to as "Paris Format." The code has been
11compiled and tested on the following platforms:
12
13 HP PA-RISC (HP/UX 9.X) (i.e., HP 9000/7XX and 9000/3XX)
14 Sun Sparc (SunOS 4.X)
15 DECstation 5000 and Alpha
16 Sequent
17 Irix 4.0.5
18 Linux
19
20See the CHANGES file for information on all the improvements over 2.0
21
22See the manual page for information on how to use mpeg_stat.
23
24Send bug reports to mpeg-bugs@plateau.cs.berkeley.edu,
25job offers (PhD in '96) to smoot@cs.berkeley.edu ;-)
26
27Vendors or other organizations interested in supporting
28this research or discussing other aspects of this project should
29contact Professor Larry Rowe at rowe@CS.Berkeley.EDU.
30
31FUTURE PLANS:
32In the next version I'd like to beef up the verification code and do
33something with system layer audio (when present). In addition (major
34though) MPEG-2 would be cool. If you send me code to do any of this, it
35will be much appreciated. (In general, though, I'll only be improving it
36to meet my thesis needs. -srs)
37
38INSTALL:
39 If you have gcc, type "make"
40 See the file INSTALL for slightly more help.
41
42AUDIO (we don't do audio, but Chad Fogg points out):
43 CCETT has been distributing executables only of their Audio bitstream
44 syntax verifier. The site address is: 161.105.2.22
45 Audio conformance bitstreams are also at ftp.tnt.uni-hannover.de
46
47ACKNOWLEDGEMENTS:
48 We gratefully thank Hewlett-Packard, Fujitsu, the Semiconductor
49 Research Corporation for financial support.
50
51 We also want to thank the following people for their help:
52
53 Tom Lane of the Independent JPEG Group provided us with
54 the basic inverse DCT code used by our player.
55 (tom_lane@g.gp.cs.cmu.edu)
56
57 Reid Judd of Sun Microsystems provided advice and assistance.
58
59 Todd Brunhoff of NVR provided advice and assistance.
60
61 Toshihiko Kawai of Sony provided advice and assistance.
62
63USAGE:
64--------------------------------------------------------------------------------
65
66As a sample of the basic information mpeg_stat collects, here is the output of:
67mpeg_stat -quiet flower.mpg
68
69mpeg_stat -- MPEG Analyzer for MPEG I video streams (version 2.2)
70
71Reading flower.mpg
72
73SUMMARY:
74
75Total Bytes read: 719506. Total number of frames: 150. Length is 5.01 sec
76
77Width: 352 Height: 240
78Avg. Frame Size: 4786 bytes + 6 bits (average rate 1147692.76 bits/sec)
79
80Total Compression Rate: 1.89 % of uncompressed 24 bit images
81 = 0.45 bits per pixel
82
83Number of Macroblocks [width * height = sum]: 22 x 15 = 330 per frame
84Skipped Macroblocks = 4389 (9.46%), Coded Macroblocks = 41992 (90.54%)
85 Coded blocks: 34.40% [ 22.30 22.79 23.44 24.21 4.89 2.37 ]
86
87MPEG-Viewer requirements:
88 Pixel aspect ratio of 0.6735
89 Required display speed: 29.97 frames/sec
90 Specified bit rate is 1.00 MBits/sec (2500 * 400bits/sec)
91 Requested buffer size is 8K ints (16 bits/int).
92 And the constrained parameter flag is on.
93 The stream meets the constrained parameter requirements.
94
95Length of vectors in half pixels:
96 Horizontal forward vectors, maximum : 39 average: 6
97 Vertical forward vectors, maximum : 22 average: 0
98
99 Horizontal backward vectors, maximum: 29 average: 5
100 Vertical backward vectors, maximum : 16 average: 0
101
102
103Frame specific information:
104
105 10 I FRAMES, average is:
106 Size: 17323 bytes + 1 bits (24.13%)
107 Compression Rate: 6.84%
108 Q Factor [scales quantization matrix]: 12.51
109
110 40 P FRAMES, average is:
111 Size: 8031 bytes + 0 bits (44.74%)
112 Compression Rate: 3.17%
113 Q Factor [scales quantization matrix]: 12.55
114
115 100 B FRAMES, average is:
116 Size: 2235 bytes + 4 bits (31.13%)
117 Compression Rate: 0.88%
118 Q Factor [scales quantization matrix]: 16.49
119 50.59% interpolated Macro Blocks
120
121