annotate DOCS/tech/encoding-tips.txt @ 30935:59f25d383d12

rename eatgv -> ffeatgv, lpcm -> fflpcm for consitently naming all ffmpeg codecs requested by compn
author attila
date Wed, 31 Mar 2010 06:33:08 +0000
parents 0f1b5b68af32
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
9012
805b19425c19 some URLs
arpi
parents: 8463
diff changeset
1
805b19425c19 some URLs
arpi
parents: 8463
diff changeset
2 Some important URLs:
805b19425c19 some URLs
arpi
parents: 8463
diff changeset
3 ~~~~~~~~~~~~~~~~~~~~
17679
35c8d3361404 link fix
diego
parents: 16285
diff changeset
4 http://mplayerhq.hu/~michael/codec-features.html <- lavc vs. divx5 vs. xvid
16205
4183c4585f1d URL update
gpoirier
parents: 15015
diff changeset
5 http://www.ee.oulu.fi/~tuukkat/mplayer/tests/rguyom.ath.cx/ <- lavc benchmarks, options
4183c4585f1d URL update
gpoirier
parents: 15015
diff changeset
6 http://ffdshow.sourceforge.net/tikiwiki/tiki-view_articles.php <- lavc for win32 :)
4183c4585f1d URL update
gpoirier
parents: 15015
diff changeset
7 http://www.bunkus.org/dvdripping4linux/index.html <- a nice tutorial
4183c4585f1d URL update
gpoirier
parents: 15015
diff changeset
8 http://forum.zhentarim.net/viewtopic.php?p=237 <- lavc option comparison
4183c4585f1d URL update
gpoirier
parents: 15015
diff changeset
9 http://www.ee.oulu.fi/~tuukkat/mplayer/tests/readme.html <- series of benchmarks
16285
2de4d051dd21 The thread "Call for video encoding settings" has to be easily found until its infos make it to our docs
gpoirier
parents: 16205
diff changeset
10 http://thread.gmane.org/gmane.comp.video.mencoder.user/1196 <- free codecs shoutout and recommended encoding settings
9012
805b19425c19 some URLs
arpi
parents: 8463
diff changeset
11
11423
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
12
9012
805b19425c19 some URLs
arpi
parents: 8463
diff changeset
13 ================================================================================
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
14
11423
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
15
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
16 FIXING A/V SYNC WHEN ENCODING
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
17
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
18 I know this is a popular topic on the list, so I thought I'd share a
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
19 few comments on my experience fixing a/v sync. As everyone seems to
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
20 know, mencoder unfortunately doesn't have a -delay option. But that
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
21 doesn't mean you can't fix a/v sync. There are a couple ways to still
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
22 do it.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
23
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
24 In example 1, we'll suppose you want to re-encode the audio anyway.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
25 This will be essential if your source audio isn't mp3, e.g. for DVD's
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
26 or nasty avi files with divx/wma audio. This approach makes things
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
27 much easier.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
28
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
29 Step 1: Dump the audio with mplayer -ao pcm -nowaveheader. There are
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
30 various options that can be used to speed this up, most notably -vo
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
31 null, -vc null, and/or -hardframedrop. -benchmark also seemed to help
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
32 in the past. :)
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
33
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
34 Step 2: Figure out what -delay value syncs the audio right in mplayer.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
35 If this number is positive, use a command like the following:
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
36
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
37 dd if=audiodump.wav bs=1764 skip=[delay] | lame -x - out.mp3
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
38
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
39 where [delay] is replaced by your delay amount in hundredths of a
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
40 second (1/10 the value you use with mplayer). Otherwise, if delay is
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
41 negative, use a command like this:
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
42
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
43 ( dd if=/dev/zero bs=1764 skip=[delay] ; cat audiodump.wav ) | lame -x - out.mp3
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
44
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
45 Don't include the minus (-) sign in delay. Also, keep in mind you'll
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
46 have to change the 1764 number and provide additional options to lame
13047
4e613c87c2f5 spelling: big-endian and little-endian
diego
parents: 13022
diff changeset
47 if your audio stream isn't 44100/16bit/little-endian/stereo.
11423
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
48
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
49 Step 3: Use mencoder to remux your new mp3 file with the movie:
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
50
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
51 mencoder -audiofile out.mp3 -oac copy ...
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
52
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
53 You can either copy video as-is (with -ovc copy) or re-encode it at
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
54 the same time you merge in the audio like this.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
55
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
56 Finally, as a variation on this method (makes things a good bit faster
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
57 and doesn't use tons of temporary disk space) you can merge steps 1
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
58 and 2 by making a named pipe called "audiodump.wav" (type mkfifo
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
59 audiodump.wav) and have mplayer write the audio to it at the same time
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
60 you're running lame to encode.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
61
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
62 Now for example 2. This time we won't re-encode audio at all. Just
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
63 dump the mp3 stream from the avi file with mplayer -dumpaudio. Then,
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
64 you have to cut and paste the raw mp3 stream a bit...
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
65
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
66 If delay is negative, things are easier. Just use lame to encode
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
67 silence for the duration of delay, at the same samplerate and
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
68 samplesize used in your avi file. Then, do something like:
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
69
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
70 cat silence.mp3 stream.dump > out.mp3
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
71 mencoder -audiofile out.mp3 -oac copy ...
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
72
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
73 On the other hand, if delay is positive, you'll need to crop off part
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
74 of the mp3 from the beginning. If it's (at least roughly) CBR this is
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
75 easy -- just take off the first (bitrate*delay/8) bytes of the file.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
76 You can use the excellent dd tool, or just your favorite
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
77 binary-friendly text editor to do this. Otherwise, you'll have to
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
78 experiment with cutting off different amounts. You can test with
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
79 mplayer -audiofile before actually spending time remuxing/encoding
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
80 with mencoder to make sure you cut the right amount.
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
81
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
82 I hope this has all been informative. If anyone would like to clean
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
83 this message up a bit and make it into part of the docs, feel free. Of
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
84 course mencoder should eventually just get -delay. :)
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
85
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
86 Rich
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
87
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
88
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
89 ================================================================================
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
90
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
91
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
92 ENCODING QUALITY - OR WHY AUTOMATISM IS BAD.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
93
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
94 Hi everyone.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
95
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
96 Some days ago someone suggested adding some preset options to mencoder.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
97 At that time I replied 'don't do that', and now I decided to elaborate
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
98 on that.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
99
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
100 Warning: this is rather long, and it involves mathematics. But if you
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
101 don't want to bother with either then why are you encoding in the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
102 first place? Go do something different!
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
103
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
104 The good news is: it's all about the bpp (bits per pixel).
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
105
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
106 The bad news is: it's not THAT easy ;)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
107
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
108 This mail is about encoding a DVD to MPEG4. It's about the video
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
109 quality, not (primarily) about the audio quality or some other fancy
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
110 things like subtitles.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
111
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
112 The first step is to encode the audio. Why? Well if you encode the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
113 audio prior to the video you'll have to make the video fit onto one
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
114 (or two) CD(s). That way you can use oggenc's quality based encoding
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
115 mode which is much more sophisticated than its ABR based mode.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
116
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
117 After encoding the audio you have a certain amount of space left to
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
118 fill with video. Let's assume the audio takes 60M (no problem with
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
119 Vorbis), and you aim at a 700M CD. This leaves you 640M for the video.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
120 Let's further assume that the video is 100 minutes or 6000 seconds
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
121 long, encoded at 25fps (those nasty NTSC fps values give me
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
122 headaches. Adjust to your needs, of course!). This leaves you with
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
123 a video bitrate of:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
124
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
125 $videosize * 8
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
126 $videobitrate = --------------
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
127 $length * 1000
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
128
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
129 $videosize in bytes, $length in seconds, $videobitrate in kbit/s.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
130 In my example I end up with $videobitrate = 895.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
131
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
132 And now comes the question: how do I chose my encoding parameters
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
133 so that the results will be good? First let's take a look at a
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
134 typical mencoder line:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
135
10405
a5785a0b9ee1 -dvd -> dvd:// and -vcd -> vcd://
jonas
parents: 9644
diff changeset
136 mencoder dvd://1 -o /dev/null -oac copy -ovc lavc \
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
137 -lavcopts vcodec=mpeg4:vbitrate=1000:vhq:vqmin=2:\
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
138 vlelim=-4:vcelim=9:lumi_mask=0.05:dark_mask=0.01:vpass=1 \
9644
0fe056bdb135 vop -> vf change, small fixes. The Polish documentation should be checked for correctness.
jonas
parents: 9012
diff changeset
139 -vf crop=716:572:2:2,scale=640:480
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
140
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
141 Phew, all those parameters! Which ones should I change? NEVER leave
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
142 out 'vhq'. Never ever. 'vqmin=2' is always good if you aim for sane
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
143 settings - like 'normal length' movies on one CD, 'very long movies'
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
144 on two CDs and so on. vcodec=mpeg4 is mandatory.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
145
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
146 The 'vlelim=-4:vcelim=9:lumi_mask=0.05:dark_mask=0.01' are parameters
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
147 suggested by D Richard Felker for non-animated movies, and they
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
148 improve quality a bit.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
149
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
150 But the two things that have the most influence on quality are
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
151 vbitate and scale. Why? Because both together tell the codec how
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
152 many bits it may spend on each frame for each bit: and this is
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
153 the 'bpp' value (bits per pixel). It's simply defined as
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
154
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
155 $videobitrate * 1000
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
156 $bpp = -----------------------
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
157 $width * $height * $fps
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
158
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
159 I've attached a small Perl script that calculates the $bpp for
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
160 a movie. You'll have to give it four parameters:
9644
0fe056bdb135 vop -> vf change, small fixes. The Polish documentation should be checked for correctness.
jonas
parents: 9012
diff changeset
161 a) the cropped but unscaled resolution (use '-vf cropdetect'),
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
162 b) the encoded aspect ratio. All DVDs come at 720x576 but contain
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
163 a flag that tells the player wether it should display the DVD at
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
164 an aspect ratio of 4/3 (1.333) or at 16/9 (1.777). Have a look
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
165 at mplayer's output - there's something about 'prescaling'. That's
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
166 what you are looking for.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
167 c) the video bitrate in kbit/s and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
168 d) the fps.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
169
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
170 In my example the command line and calcbpp.pl's output would look
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
171 like this (warning - long lines ahead):
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
172
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
173 mosu@anakin:~$ ./calcbpp.pl 720x440 16/9 896 25
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
174 Prescaled picture: 1023x440, AR 2.33
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
175 720x304, diff 5, new AR 2.37, AR error 1.74% scale=720:304 bpp: 0.164
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
176 704x304, diff -1, new AR 2.32, AR error 0.50% scale=704:304 bpp: 0.167
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
177 688x288, diff 8, new AR 2.39, AR error 2.58% scale=688:288 bpp: 0.181
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
178 672x288, diff 1, new AR 2.33, AR error 0.26% scale=672:288 bpp: 0.185
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
179 656x288, diff -6, new AR 2.28, AR error 2.17% scale=656:288 bpp: 0.190
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
180 640x272, diff 3, new AR 2.35, AR error 1.09% scale=640:272 bpp: 0.206
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
181 624x272, diff -4, new AR 2.29, AR error 1.45% scale=624:272 bpp: 0.211
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
182 608x256, diff 5, new AR 2.38, AR error 2.01% scale=608:256 bpp: 0.230
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
183 592x256, diff -2, new AR 2.31, AR error 0.64% scale=592:256 bpp: 0.236
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
184 576x240, diff 8, new AR 2.40, AR error 3.03% scale=576:240 bpp: 0.259
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
185 560x240, diff 1, new AR 2.33, AR error 0.26% scale=560:240 bpp: 0.267
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
186 544x240, diff -6, new AR 2.27, AR error 2.67% scale=544:240 bpp: 0.275
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
187 528x224, diff 3, new AR 2.36, AR error 1.27% scale=528:224 bpp: 0.303
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
188 512x224, diff -4, new AR 2.29, AR error 1.82% scale=512:224 bpp: 0.312
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
189 496x208, diff 5, new AR 2.38, AR error 2.40% scale=496:208 bpp: 0.347
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
190 480x208, diff -2, new AR 2.31, AR error 0.85% scale=480:208 bpp: 0.359
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
191 464x192, diff 7, new AR 2.42, AR error 3.70% scale=464:192 bpp: 0.402
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
192 448x192, diff 1, new AR 2.33, AR error 0.26% scale=448:192 bpp: 0.417
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
193 432x192, diff -6, new AR 2.25, AR error 3.43% scale=432:192 bpp: 0.432
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
194 416x176, diff 3, new AR 2.36, AR error 1.54% scale=416:176 bpp: 0.490
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
195 400x176, diff -4, new AR 2.27, AR error 2.40% scale=400:176 bpp: 0.509
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
196 384x160, diff 5, new AR 2.40, AR error 3.03% scale=384:160 bpp: 0.583
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
197 368x160, diff -2, new AR 2.30, AR error 1.19% scale=368:160 bpp: 0.609
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
198 352x144, diff 7, new AR 2.44, AR error 4.79% scale=352:144 bpp: 0.707
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
199 336x144, diff 0, new AR 2.33, AR error 0.26% scale=336:144 bpp: 0.741
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
200 320x144, diff -6, new AR 2.22, AR error 4.73% scale=320:144 bpp: 0.778
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
201
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
202 A word for the $bpp. For a fictional movie which is only black and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
203 white: if you have a $bpp of 1 then the movie would be stored
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
204 uncompressed :) For a real life movie with 24bit color depth you
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
205 need compression of course. And the $bpp can be used to make the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
206 decision easier.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
207
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
208 As you can see the resolutions suggested by the script are all
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
209 dividable by 16. This will make the aspect ratio slightly wrong,
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
210 but no one will notice.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
211
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
212 Now if you want to decide which resolution (and scaling parameters)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
213 to chose you can do that by looking at the $bpp:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
214
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
215 < 0.10: don't do it. Please. I beg you!
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
216 < 0.15: It will look bad.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
217 < 0.20: You will notice blocks, but it will look ok.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
218 < 0.25: It will look really good.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
219 > 0.25: It won't really improve visually.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
220 > 0.30: Don't do that either - try a bigger resolution instead.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
221
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
222 Of course these values are not absolutes! For movies with really lots
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
223 of black areas 0.15 may look very good. Action movies with only high
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
224 motion scenes on the other hand may not look perfect at 0.25. But these
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
225 values give you a great idea about which resolution to chose.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
226
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
227 I see a lot of people always using 512 for the width and scaling
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
228 the height accordingly. For my (real-world-)example this would be
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
229 simply a waste of bandwidth. The encoder would probably not even
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
230 need the full bitrate, and the resulting file would be smaller
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
231 than my targetted 700M.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
232
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
233 After encoding you'll do your 'quality check'. First fire up the movie
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
234 and see whether it looks good to you or not. But you can also do a
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
235 more 'scientific' analysis. The second Perl script I attached counts
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
236 the quantizers used for the encoding. Simply call it with
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
237
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
238 countquant.pl < divx2pass.log
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
239
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
240 It will print out which quantizer was used how often. If you see that
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
241 e.g. the lowest quantizer (vqmin=2) gets used for > 95% of the frames
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
242 then you can safely increase your picture size.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
243
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
244 > The "counting the quantesizer"-thing could improve the quality of
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
245 > full automated scripts, as I understand ?
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
246
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
247 Yes, the log file analysis can be used be tools to automatically adjust
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
248 the scaling parameters (if you'd do that you'd end up with a three-pass
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
249 encoding for the video only ;)), but it can also provide answers for
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
250 you as a human. From time to time there's a question like 'hey,
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
251 mencoder creates files that are too small! I specified this bitrate and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
252 the resulting file is 50megs short of the target file size!'. The
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
253 reason is probably that the codec already uses the minimum quantizer
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
254 for nearly all frames so it simply does not need more bits. A quick
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
255 glance at the distribution of the quantizers can be enlightening.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
256
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
257 Another thing is that q=2 and q=3 look really good while the 'bigger'
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
258 quantizers really lose quality. So if your distribution shows the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
259 majority of quantizers at 4 and above then you should probably decrease
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
260 the resolution (you'll definitly see block artefacts).
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
261
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
262
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
263 Well... Several people will probably disagree with me on certain
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
264 points here, especially when it comes down to hard values (like the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
265 $bpp categories and the percentage of the quantizers used). But
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
266 the idea is still valid.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
267
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
268 And that's why I think that there should NOT be presets in mencoder
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
269 like the presets lame knows. 'Good quality' or 'perfect quality' are
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
270 ALWAYS relative. They always depend on a person's personal preferences.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
271 If you want good quality then spend some time reading and - more
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
272 important - understanding what steps are involved in video encoding.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
273 You cannot do it without mathematics. Oh well, you can, but you'll
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
274 end up with movies that could certainly look better.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
275
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
276 Now please shoot me if you have any complaints ;)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
277
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
278 --
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
279 ==> Ciao, Mosu (Moritz Bunkus)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
280
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
281 ===========
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
282 ANOTHER APPROACH: BITS PER BLOCK:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
283
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
284 > $videobitrate * 1000
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
285 > $bpp = -----------------------
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
286 > $width * $height * $fps
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
287
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
288 Well, I came to similar equation going through different route. Only I
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
289 didn't use bits per pixel, in my case it was bits per block (BPB). The block
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
290 is 16x16 because lots of software depends on video width/height being
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
291 divisable by 16. And because I didn't like this 0.2 bit per pixel, when
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
292 bit is quite atomic ;)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
293
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
294 So the equation was something like:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
295
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
296 bitrate
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
297 bpb = -----------------
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
298 fps * ((width * height) / (16 * 16))
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
299
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
300 (width and height are from destination video size, and bitrate is in
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
301 bits (i.e. 900kbps is 900000))
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
302
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
303 This way it apeared that the minimum bits per block is ~40, very
11271
252fb0cf331a spelling fixes, mostly by <ismail.donmez@boun.edu.tr>
diego
parents: 10432
diff changeset
304 good results are with ~50, and everything above 60 is a waste of bandwidth.
22316
f3d7a1b58a82 cosmetics: Fix some common typos, appropiate --> appropRiate,
diego
parents: 17679
diff changeset
305 And what's actually funny is that it was independent of codec used. The
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
306 results were exactly the same, whether I used DIV3 (with tricky nandub's
26908
13061a6174b6 Change spelling of XviD to Xvid as has already been done in the (rest of the)
diego
parents: 26883
diff changeset
307 magick), ffmpeg odivx, DivX5 on Windows or Xvid.
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
308
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
309 Surprisingly there is one advantage of using nandub-DIV3 for bitrate
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
310 starved encoding: ringing almost never apears this way.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
311
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
312 But I also found out, that the quality/BPB isn't constant for
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
313 drastically different resolutions. Smaller picture (like MPEG1 sizes)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
314 need more BPB to look good than say typical MPEG2 resolutions.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
315
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
316 Robert
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
317
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
318
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
319 ===========
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
320 DON'T SCALE DOWN TOO MUCH
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
321
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
322 Sometimes I found that encoding to y-scaled only DVD qualty (ie 704 x
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
323 288 for a 2.85 film) gives better visual quality than a scaled-down
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
324 version even if the quantizers are significantly higher than for the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
325 scaled-down version.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
326 Keep in mind that blocs, fuzzy parts and generaly mpeg artefacts in a
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
327 704x288 image will be harder to spot in full-screen mode than on a
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
328 512x208 image. In fact I've see example where the same movie looks
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
329 better compressed to 704x288 with an average weighted quantizer of
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
330 ~3 than the same movie scaled to 576x240 with an average weighted
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
331 quantizer of 2.4.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
332 Btw, a print of the weighted average quantizer would be nice in
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
333 countquant.pl :)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
334
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
335 Another point in favor of not trying to scale down too much : on hard
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
336 scaled-down movies, the MPEG codec will need to compress relatively
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
337 high frequencies rather than low frequencies and it doesn't like that
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
338 at all. You will see less and less returns while you scale down and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
339 scale down again in desesperate need of some bandwidth :)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
340
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
341 In my experience, don't try to go below a width of 576 without closely
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
342 watching what's going on.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
343
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
344 --
23734
acfe034e5386 ISO8859-1 --> UTF-8
diego
parents: 22316
diff changeset
345 RĂ©mi
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
346
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
347 ===========
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
348 TIPS FOR ENCODING
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
349
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
350 That being said, with video you have some tradeoffs you can make. Most
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
351 people seem to encode with really basic options, but if you play with
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
352 single coefficient elimination and luma masking settings, you can save lots
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
353 of bits, resulting in lower quantizers, which means less blockiness and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
354 less ugly noise (ringing) around sharp borders. The tradeoff, however, is
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
355 that you'll get some "muddiness" in some parts of the image. Play around
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
356 with the settings and see for yourself. The options I typically use for
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
357 (non-animated) movies are:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
358
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
359 vlelim=-4
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
360 vcelim=9
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
361 lumi_mask=0.05
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
362 dark_mask=0.01
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
363
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
364 If things look too muddy, making the numbers closer to 0. For anime and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
365 other animation, the above recommendations may not be so good.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
366
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
367 Another option that may be useful is allowing four motion vectors per
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
368 macroblock (v4mv). This will increase encoding time quite a bit, and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
369 last I checked it wasn't compatible with B frames. AFAIK, specifying
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
370 v4mv should never reduce quality, but it may prevent some old junky
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
371 versions of DivX from decoding it (can anyone conform?). Another issue
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
372 might be increased cpu time needed for decoding (again, can anyone
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
373 confirm?).
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
374
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
375 To get more fair distribution of bits between low-detail and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
376 high-detail scenes, you should probably try increasing vqcomp from the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
377 default (0.5) to something in the range 0.6-0.8.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
378
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
379 Of course you also want to make sure you crop ALL of the black border and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
380 any half-black pixels at the edge of the image, and make sure the final
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
381 image dimensions after cropping and scaling are multiples of 16. Failing to
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
382 do so will drastically reduce quality.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
383
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
384 Finally, if you can't seem to get good results, you can try scaling the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
385 movie down a bit smaller or applying a weak gaussian blur to reduce the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
386 amount of detail.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
387
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
388 Now, my personal success story! I just recently managed to fit a beautiful
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
389 encode of Kundun (well over 2 hours long, but not too many high-motion
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
390 scenes) on one cd at 640x304, with 66 kbit/sec abr ogg audio, using the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
391 options I described above. So, IMHO it's definitely possible to get very
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
392 good results with libavcodec (certainly MUCH better than all the idiot
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
393 "release groups" using DivX3 make), as long as you take some time to play
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
394 around with the options.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
395
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
396
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
397 Rich
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
398
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
399 ============
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
400 ABOUT VLELIM, VCELIM, LUMI_MASK AND DARK_MASK PART I: LUMA & CHROMA
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
401
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
402
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
403 The l/c in vlelim and vcelim stands for luma (brightness plane) and chroma
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
404 (color planes). These are encoded separately in all mpeg-like algorithms.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
405 Anyway, the idea behind these options is (at least from what I understand)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
406 to use some good heuristics to determine when the change in a block is less
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
407 than the threshold you specify, and in such a case, to just encode the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
408 block as "no change". This saves bits and perhaps speeds up encoding. Using
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
409 a negative value for either one means the same thing as the corresponding
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
410 positive value, but the DC coefficient is also considered. Unfortunately
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
411 I'm not familiar enough with the mpeg terminology to know what this means
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
412 (my first guess would be that it's the constant term from the DCT), but it
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
413 probably makes the encoder less likely to apply single coefficient
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
414 elimination in cases where it would look bad. It's presumably recommended
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
415 to use negative values for luma (which is more noticable) and positive for
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
416 chroma.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
417
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
418 The other options -- lumi_mask and dark_mask -- control how the quantizer
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
419 is adjusted for really dark or bright regions of the picture. You're
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
420 probably already at least a bit familiar with the concept of quantizers
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
421 (qscale, lower = more precision, higher quality, but more bits needed to
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
422 encode). What not everyone seems to know is that the quantizer you see
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
423 (e.g. in the 2pass logs) is just an average for the whole frame, and lower
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
424 or higher quantizers may in fact be used in parts of the picture with more
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
425 or less detail. Increasing the values of lumi_mask and dark_mask will cause
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
426 lavc to aggressively increase the quantizer in very dark or very bright
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
427 regions of the picture (which are presumably not as noticable to the human
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
428 eye) in order to save bits for use elsewhere.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
429
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
430 Rich
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
431
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
432 ===================
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
433 ABOUT VLELIM, VCELIM, LUMI_MASK AND DARK_MASK PART II: VQSCALE
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
434
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
435 OK, a quick explanation. The quantizer you set with vqscale=N is the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
436 per-frame quantizer parameter (aka qp). However, with mpeg4 it's
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
437 allowed (and recommended!) for the encoder to vary the quantizer on a
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
438 per-macroblock (mb) basis (as I understand it, macroblocks are 16x16
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
439 regions composed of 4 8x8 luma blocks and 2 8x8 chroma blocks, u and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
440 v). To do this, lavc scores each mb with a complexity value and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
441 weights the quantizer accordingly. However, you can control this
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
442 behavior somewhat with scplx_mask, tcplx_mask, dark_mask, and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
443 lumi_mask.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
444
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
445 scplx_mask -- raise quantizer on mb's with lots of spacial complexity.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
446 Spacial complexity is measured by variance of the texture (this is
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
447 just the actual image for I blocks and the difference from the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
448 previous coded frame for P blocks).
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
449
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
450 tcplx_mask -- raise quantizer on mb's with lots of temporal
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
451 complexity. Temporal complexity is measured according to motion
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
452 vectors.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
453
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
454 dark_mask -- raise quantizer on very dark mb's.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
455
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
456 lumi_mask -- raise quantizer on very bright mb's.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
457 Somewhere around 0-0.15 is a safe range for these values, IMHO. You
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
458 might try as high as 0.25 or 0.3. You should probably never go over
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
459 0.5 or so.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
460
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
461 Now, about naq. When you adjust the quantizers on a per-mb basis like
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
462 this (called adaptive quantization), you might decrease or (more
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
463 likely) increase the average quantizer used, so that it no longer
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
464 matches the requested average quantizer (qp) for the frame. This will
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
465 result in weird things happening with the bitrate, at least from my
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
466 experience. What naq does is "normalize adaptive quantization". That
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
467 is, after the above masking parameters are applied on a per-mb basis,
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
468 the quantizers of all the blocks are rescaled so that the average
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
469 stays fixed at the desired qp.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
470
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
471 So, if I used vqscale=4 with naq and fairly large values for the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
472 masking parameters, I might be likely to see lots of frames using
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
473 qscale 2,3,4,5,6,7 across different macroblocks as needed, but with
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
474 the average sticking around 4. However, I haven't actually tested such
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
475 a setup yet, so it's just speculation right now.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
476
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
477 Have fun playing around with it.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
478
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
479 Rich
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
480
11423
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
481
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
482 ================================================================================
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
483
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
484
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
485 TIPS FOR ENCODING OLD BLACK & WHITE MOVIES:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
486
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
487 I found myself that 4:3 B&W old movies are very hard to compress well. In
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
488 addition to the 4:3 aspect ratio which eats lots of bits, those movies are
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
489 typically very "noisy", which doesn't help at all. Anyway :
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
490
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
491 > After a few tries I am
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
492 > still a little bit disappointed with the video quality. Since it is a
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
493 > "dark" movies, there is a lot of black on the pictures, and on the
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
494 > encoded avi I can see a lot of annoying "mpeg squares". I am using
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
495 > avifile codec, but the best I think is to give you the command line I
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
496 > used to encode a preview of the result:
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
497
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
498 >
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
499 > First pass:
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
500 > mencoder TITLE01-ANGLE1.VOB -oac copy -ovc lavc -lavcopts
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
501 > vcodec=mpeg4:vhq:vpass=1:vbitrate=800:keyint=48 -ofps 23.976 -npp lb
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
502 > -ss 2:00 -endpos 0:30 -vf scale -zoom -xy 640 -o movie.avi
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
503
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
504 1) keyint=48 is way too low. The default value is 250, this is in *frames*
13022
diego
parents: 11423
diff changeset
505 not seconds. Keyframes are significantly larger than P or B frames, so the
diego
parents: 11423
diff changeset
506 less keyframes you have, better the overall movie will be. (huh, like Yoda
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
507 I speak ;). Try keyint=300 or 350. Don't go beyond that if you want
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
508 relatively precise seeking.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
509
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
510 2) you may want to play with vlelim and vcelim options. This can gives you
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
511 a significant "quality" boost. Try one of these couples :
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
512
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
513 vlelim=-2:vcelim=3
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
514 vlelim=-3:vcelim=5
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
515 vlelim=-4:vcelim=7
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
516 (and yes, there's a minus)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
517
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
518 3) crop & rescale the movie before passing it to the codec. First crop the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
519 movie to not encode black bars if there's any. For a 1h40mn movie
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
520 compressed to a 700 MB file, I would try something between 512x384 and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
521 480x320. Don't go below that if you want something relatively sharp when
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
522 viewed fullscreen.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
523
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
524 4) I would recommend using the Ogg Vorbis audio codec with the .ogm
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
525 container format. Ogg Vorbis compress audio better than MP3. On a typical
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
526 old, mono-only audio stream, a 45 kbits/s Vorbis stream is ok. How to
10432
diego
parents: 10405
diff changeset
527 extract & compress an audio stream from a ripped DVD (mplayer dvd://1
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
528 -dumpstream) :
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
529
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
530 rm -f audiodump.pcm ; mkfifo -m 600 audiodump.pcm
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
531 mplayer -quiet -vc null -vo null -aid 128 -ao pcm -nowaveheader stream.dump &
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
532 oggenc --raw --raw-bits=16 --raw-chan=2 --raw-rate=48000 -q 1 -o audio-us.ogg
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
533 +audiodump.pcm &
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
534 wait
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
535
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
536 For a nice set of utilities to manager the .ogm format, see Moritz Bunkus'
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
537 ogmtools (google is your friend).
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
538
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
539 5) use the "v4mv" option. This could gives you a few more bits at the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
540 expense of a slightly longer encoding. This is a "lossless" option, I mean
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
541 with this option you don't throw away some video information, it just
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
542 selects a more precise motion estimation method. Be warned that on some
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
543 very un-typical scenes this option may gives you a longer file than
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
544 without, although it's very rare and on a whole film I think it's always a
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
545 win.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
546
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
547 6) you can try the new luminance & darkness masking code. Play
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
548 with the "lumi_mask" and "dark_mask" options. I would recommend using
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
549 something like :
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
550 lumi_mask=0.07:dark_mask=0.10:naq:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
551 lumi_mask=0.10:dark_mask=0.12:naq:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
552 lumi_mask=0.12:dark_mask=0.15:naq
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
553 lumi_mask=0.13:dark_mask=0.16:naq:
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
554 Be warned that these options are really experimental and the result
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
555 could be very good or very bad depending on your visualization device
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
556 (computer CRT, TV or TFT screen). Don't push too hard these options.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
557
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
558 > Second pass:
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
559 > the same with vpass=2
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
560
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
561 7) I've found that lavc gives better results when the first pass is done
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
562 with "vqscale=2" instead of a target bitrate. The statistics collected
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
563 seems to be more precise. YMMV.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
564
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
565 > I am new to mencoder, so please tell me any idea you have even if it
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
566 > obvious. I also tried the "gray" option of lavc, to encode B&W only,
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
567 > but strangely it gives me "pink" squares from time to time.
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
568
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
569 Yes, I've seen that too. Playing the resulting file with "-lavdopts gray"
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
570 fix the problem but it's not very nice ...
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
571
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
572 > So if you could tell me what option of mencoder or lavc I should be
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
573 > looking at to lower the number of "squares" on the image, it would be
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
574 > great. The version of mencoder i use is 0.90pre8 on a macos x PPC
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
575 > platform. I guess I would have the same problem by encoding anime
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
576 > movies, where there are a lot of region of the image with the same
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
577 > color. So if you managed to solve this problem...
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
578
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
579 You could also try the "mpeg_quant" flag. It selects a different set of
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
580 quantizers and produce somewhat sharper pictures and less blocks on large
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
581 zones with the same or similar luminance, at the expense of some bits.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
582
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
583 > This is completely off topic, but do you know how I can create good
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
584 > subtitles from vobsub subtitles ? I checked the -dumpmpsub option of
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
585 > mplayer, but is there a way to do it really fast (ie without having to
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
586 > play the whole movie) ?
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
587
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
588 I didn't find a way under *nix to produce reasonably good text subtitles
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
589 from vobsubs. OCR *nix softwares seems either not suited to the task, not
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
590 powerful enough or both. I'm extracting the vobsub subtitles and simply use
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
591 them with the .ogm
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
592
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
593 / .avi :
10405
a5785a0b9ee1 -dvd -> dvd:// and -vcd -> vcd://
jonas
parents: 9644
diff changeset
594 1) rip the DVD to harddisk with "mplayer dvd://1 -dumpstream"
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
595 2) mount the DVD and copy the .ifo file
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
596 2) extract all vobsubs to one single file with something like :
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
597
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
598 for f in 0 1 2 3 4 5 6 7 8 9 10 11 ; do \
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
599 mencoder -ovc copy -oac copy -o /dev/null -sid $f -vobsubout sous-titres
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
600 +-vobsuboutindex $f -ifo vts_01_0.ifo stream.dump
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
601 done
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
602
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
603 (and yes, I've a DVD with 12 subtitles)
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
604 --
23734
acfe034e5386 ISO8859-1 --> UTF-8
diego
parents: 22316
diff changeset
605 RĂ©mi
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
606
11423
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
607
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
608 ================================================================================
dffc970f1922 FIXING A/V SYNC WHEN ENCODING, from an old mail by Rich:
diego
parents: 11271
diff changeset
609
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
610
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
611 TIPS FOR SMOKE & CLOUDS
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
612
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
613 Q: I'm trying to encode Dante's Peak and I'm having problems with clouds,
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
614 fog and smoke: They don't look fine (they look very bad if I watch the
25744
7b9e015bca91 typo fix
diego
parents: 23734
diff changeset
615 movie in TV-out). There are some artifacts, white clouds looks as snow
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
616 mountains, there are things likes hip in the colors so one can see frontier
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
617 curves between white and light gray and dark gray ... (I don't know if you
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
618 can understand me, I want to mean that the colors don't change smoothly)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
619 In particular I'm using vqscale=2:vhq:v4mv
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
620
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
621 A: Try adding "vqcomp=0.7:vqblur=0.2:mpeg_quant" to lavcopts.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
622
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
623 Q: I tried your suggestion and it improved the image a little ... but not
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
624 enough. I was playing with different options and I couldn't find the way.
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
625 I suppose that the vob is not so good (watching it in TV trough the
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
626 computer looks better than my encoding, but it isn't a lot of better).
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
627
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
628 A: Yes, those scenes with qscale=2 looks terrible :-(
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
629
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
630 Try with vqmin=1 in addition to mpeg_quant:vlelim=-4:vcelim=-7 (and maybe
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
631 with "-sws 10 -ssf ls=1" to sharpen a bit the image) and read about vqmin=1
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
632 in DOCS/tech/libavc-options.txt.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
633
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
634 If after the whole movie is encoded you still see the same problem, it will
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
635 means that the second pass didn't picked-up q=1 for this scene. Force q=1
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
636 with the "vrc_override" option.
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
637
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
638 Q: By the way, is there a special difficult in encode clouds or smoke?
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
639
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
640 A: I would say it depends on the sharpness of these clouds / smokes and the
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
641 fact that they are mostly black/white/grey or colored. The codec will do
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
642 the right thing with vqmin=2 for example on a cigarette smoke (sharp) or on
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
643 a red/yellow cloud (explosion, cloud of fire). But may not with a grey and
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
644 very fuzzy cloud like in the chocolat scene. Note that I don't know exactly
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
645 why ;)
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
646
23734
acfe034e5386 ISO8859-1 --> UTF-8
diego
parents: 22316
diff changeset
647 A = RĂ©mi
8463
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
648
e421b4ab7815 encoding tips - collected from mplayer-users list mailings by
arpi
parents:
diff changeset
649
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
650 ================================================================================
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
651
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
652
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
653 TIPS FOR TWEAKING RATECONTROL
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
654
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
655 (For the purpose of this explanation, consider "2nd pass" to be any beyond
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
656 the 1st. The algorithm is run only on P-frames; I- and B-frames use QPs
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
657 based on the adjacent P. While x264's 2pass ratecontrol is based on lavc's,
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
658 it has diverged somewhat and not all of this is valid for x264.)
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
659
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
660 Consider the default ratecontrol equation in lavc: "tex^qComp".
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
661 At the beginning of the 2nd pass, rc_eq is evaluated for each frame, and
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
662 the result is the number of bits allocated to that frame (multiplied by
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
663 some constant as needed to match the total requested bitrate).
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
664
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
665 "tex" is the complexity of a frame, i.e. the estimated number of bits it
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
666 would take to encode at a given quantizer. (If the 1st pass was CQP and
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
667 not turbo, then we know tex exactly. Otherwise it is calculated by
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
668 multiplying the 1st pass's bits by the QP of that frame. But that's not
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
669 why CQP is potentially good; more on that later.)
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
670 "qComp" is just a constant. It has no effect outside the rc_eq, and is
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
671 directly set by the vqcomp parameter.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
672
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
673 If vqcomp=1, then rc_eq=tex^1=tex, so 2pass allocates to each frame the
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
674 number of bits needed to encode them all at the same QP.
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
675 If vqcomp=0, then rc_eq=tex^0=1, so 2pass allocates the same number of
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
676 bits to each frame, i.e. CBR. (Actually, this is worse than 1pass CBR in
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
677 terms of quality; CBR can vary within its allowed buffer size, while
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
678 vqcomp=0 tries to make each frame exactly the same size.)
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
679 If vqcomp=0.5, then rc_eq=sqrt(tex), so the allocation is somewhere
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
680 between CBR and CQP. High complexity frames get somewhat lower quality
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
681 than low complexity, but still more bits.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
682
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
683 While the actual selection of a good value of vqcomp is experimental, the
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
684 following underlying factors determine the result:
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
685 Arguing towards CQP: You want the movie to be somewhere approaching
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
686 constant quality; oscillating quality is even more annoying than constant
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
687 low quality. (However, constant quality does not mean constant PSNR nor
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
688 constant QP. Details are less noticeable in high-motion scenes, so you can
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
689 get away with somewhat higher QP in high-complexity frames for the same
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
690 perceived quality.)
29263
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
691 Arguing towards CBR: You get more quality per bit if you spend those bits
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
692 in frames where motion compensation works well (which tends to be
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
693 correlated with "tex"): A given artifact may stick around several seconds
0f1b5b68af32 whitespace cosmetics: Remove all trailing whitespace.
diego
parents: 26908
diff changeset
694 in a low-motion scene, and you only have to fix it in one frame to improve
15015
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
695 the quality of the whole sequence.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
696
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
697 Now for why the 1st pass ratecontrol method matters:
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
698 The number of bits at constant quant is as good a measure of complexity as
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
699 any other simple formula for the purpose of allocating bits. But it's not
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
700 perfect for predicting which QP will produce the desired number of bits.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
701 Bits are approximately inversely proportional to QP, but the exact scaling
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
702 is non-linear, and depends on the amount of detail/noise, the complexity of
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
703 motion, the quality of previous frames, and other stuff not measured by the
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
704 ratecontrol. So it predicts best when the QP used for a given frame in the
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
705 2nd pass is close to the QP used in the 1st pass. When the prediction is
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
706 wrong, lavc needs to distribute the surplus or deficit of bits among future
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
707 frames, which means that they too deviate from the planned distribution.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
708 Obviously, with vqcomp=1 you can get the 1st pass QPs very close by using
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
709 CQP, and with vqcomp=0 a CBR 1st pass is very close. But with vqcomp=0.5
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
710 it's more ambiguous. The accepted wisdom is that CBR is better for
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
711 vqcomp=0.5, mostly because you then don't have to guess a QP in advance.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
712 But with vqcomp=0.6 or 0.7, the 2nd pass QPs vary less, so a CQP 1st pass
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
713 (with the right QP) will be a better predictor than CBR.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
714
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
715 To make it more confusing, 1pass CBR uses the same rc_eq with a different
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
716 meaning. In CBR, we don't have a real encode to estimate from, so "tex" is
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
717 calculated from the full-pixel precision motion-compensation residual.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
718 While the number of bits allocated to a given frame is decided by the rc_eq
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
719 just like in 2nd pass, the target bitrate is constant (instead of being the
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
720 sum of per-frame rc_eq values). So the scaling factor (which is constant in
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
721 2nd pass) now varies in order to keep the local average bitrate near the
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
722 CBR target. So vqcomp does affect CBR, though it only determines the local
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
723 allocation of bits, not the long-term allocation.
23237af42335 Technical explanation of how to use vqcomp, and why, featured by Loren Merritt
gpoirier
parents: 13199
diff changeset
724
16205
4183c4585f1d URL update
gpoirier
parents: 15015
diff changeset
725 --Loren Merritt