Mercurial > mplayer.hg

comparison DOCS/tech/snow.txt @ 15122:4d25185dedaa

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Fixes suggested by The Wanderer.
author gpoirier
date Mon, 11 Apr 2005 11:13:48 +0000
parents 81ab69124505
children e743f75cdcc4
comparison
equal deleted inserted replaced
15121:8aff72f4082d 15122:4d25185dedaa
8 change, do not rely on it to store videos that you value. 8 change, do not rely on it to store videos that you value.
9 For this reason, MEncoder will not encode without 'vstrict=-1' on the 9 For this reason, MEncoder will not encode without 'vstrict=-1' on the
10 command line. 10 command line.
11 11
12 12
13 OPTIONS OBEYED BY SNOW 13 OPTIONS RECOGNIZED BY SNOW
14 14
15 * vqscale=<0.01-255> 15 * vqscale=<0.01-255>
16 Encoding quality, sane range 1-10, default: 0 (lossless). 16 Encoding quality, sane range 1-10, default: 0 (lossless). May be
17 fractional.
17 A given quality in snow needs a somewhat lower qscale than the same 18 A given quality in snow needs a somewhat lower qscale than the same
18 quality in MPEG-4. 19 quality in MPEG-4.
19 Note that 0 may not be specified; if you want lossless encoding, you 20 Note that 0 may not be specified; if you want lossless encoding, you
20 must leave out vqscale. (Yes, quantizer can be fractional.) 21 must leave out vqscale.
21 22
22 * cmp, subcmp, mbcmp 23 * cmp, subcmp, mbcmp
23 Set the comparison function, default: 0 (SAD). 24 Set the comparison function, default: 0 (SAD).
24 useful values = 0 (SAD), 1 (SSD), 11 (5/3 wavelet), 12 (9/7 wavelet). 25 useful values = 0 (SAD), 1 (SSD), 11 (5/3 wavelet), 12 (9/7 wavelet).
25 Experience shows that SSD is the best most of the time, while SAD is 26 Experience shows that SSD is the best most of the time, while SAD is
32 Wavelet type. 0 (9/7 wavelet), 1 (5/3 wavelet), 2 (13/7 wavelet). 33 Wavelet type. 0 (9/7 wavelet), 1 (5/3 wavelet), 2 (13/7 wavelet).
33 9/7 is probably better for for lossy coding, and 5/3 for lossless. 34 9/7 is probably better for for lossy coding, and 5/3 for lossless.
34 35
35 * last_pred=<0-3> 36 * last_pred=<0-3>
36 Tries a few extra predicted motion vectors before doing EPZS search, 37 Tries a few extra predicted motion vectors before doing EPZS search,
37 default=0. 38 default: 0.
38 This option has negligible effect on both speed and quality of snow, 39 This option has negligible effect on both speed and quality of snow,
39 so just leave it off. (it does, however, help mpeg4.) 40 so just leave it off. (It does, however, help MPEG-4.)
40 41
41 * qpel 42 * qpel
42 Refines motion estimation, default: off. 43 Refines motion estimation, default: off.
43 This setting always helps compressibility, but costs some CPU time 44 This setting always helps compressibility, but costs some CPU time
44 both while encoding and decoding. 45 both while encoding and decoding.
45 46
46 * v4mv 47 * v4mv
47 Allows smaller motion partitions, default: off. 48 Allows smaller motion partitions, default: off.
48 The current MB decision algorithm doesn't make very good use of this: 49 The current MB decision algorithm doesn't make very good use of this:
49 It improves quality, but also increases bitrate. (and you could get 50 It improves quality, but also increases bitrate. (You could get
50 more quality per bitrate by reducing quantizer instead.) 51 more quality per bitrate by reducing quantizer instead.)
51 52
52 53
53 v4mv and the wavelet cmp functions are theoretically good, but in practice 54 v4mv and the wavelet cmp functions are theoretically good, but in practice
54 won't work well until we get OBMC-aware motion estimation. 55 won't work well until we get OBMC-aware motion estimation.