Doom10 Forum: Digital Video Discussion
September 07, 2010, 05:00:32 AM *
Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with username, password and session length
News: Of course it's better, it's one more. Read the rules!
 
   Home   Help Search Login Register  
Pages: [1]
  Print  
Author Topic: VBV, higher bitrate at the same crf?  (Read 825 times)
STaRGaZeR
Member

Online Online

Posts: 7


View Profile
« on: March 11, 2010, 05:48:39 PM »

I've been playing with x264 with different settings to see the results. One of the test really took my attention. Command line:

Code:
x264 --crf 15 --fps 29.97 --keyint 29 --me tesa --subme 10 --ref 4 --bframes 3 --deblock -1:-1 --psy-rd 0.0:0.0 --trellis 2 --slices 4 --aq-mode 2 --b-adapt 2 --weightp 0 --profile high --level 4.1 --min-keyint 1 --b-pyramid strict --aud --direct auto --partitions all --colorprim "bt709" --transfer "bt709" --colormatrix "bt709" --fullrange "on" --sar 1:1 test.mkv -o 2tesasub10.2x.mkv

With the exact same settings, source, etc. but this time with --vbv-maxrate 40000 --vbv-bufsize 30000 resulted in 425 kbps more. Results without VBV:

Code:
E:\tests>x264 --crf 15 --fps 29.97 --keyint 29 --me tesa --subme 10 --ref 4 --bframes 3 --deblock -1:-1 --psy-rd 0.0:0.0 --trellis 2 --slices 4 --aq-mode 2 --b-adapt 2 --weightp 0 --profile high --level 4.1 --min-keyint 1 --b-pyramid strict --aud --direct auto --partitions all --colorprim "bt709" --transfer "bt709" --colormatrix "bt709" --fullrange "on" --sar 1:1 test.mkv -o 2tesasub10.22.mkv
ffms [info]: 1680x1050p 1:1 @ 30000/1001 fps (cfr)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 FastShuffle SSE4.2
x264 [info]: profile High, level 4.1
x264 [info]: frame I:117   Avg QP:14.04  size:130464
x264 [info]: frame P:1225  Avg QP:18.13  size: 44200
x264 [info]: frame B:1620  Avg QP:21.11  size: 17882
x264 [info]: consecutive B-frames: 17.1% 11.9% 27.3% 43.7%
x264 [info]: mb I  I16..4: 27.9% 43.1% 29.1%
x264 [info]: mb P  I16..4:  7.1% 12.5%  4.2%  P16..4: 20.6% 11.4%  3.6%  1.2%  0
.5%    skip:39.0%
x264 [info]: mb B  I16..4:  1.5%  1.8%  0.6%  B16..8: 31.8%  3.0%  3.0%  direct:
 3.1%  skip:55.3%  L0:39.2% L1:42.4% BI:18.4%
x264 [info]: 8x8 transform intra:49.2% inter:55.5%
x264 [info]: direct mvs  spatial:99.8% temporal:0.2%
x264 [info]: coded y,uvDC,uvAC intra: 48.0% 44.1% 20.5% inter: 14.7% 7.2% 0.6%
x264 [info]: i16 v,h,dc,p: 40% 19% 10% 31%
x264 [info]: i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 26% 17% 22%  5%  6%  7%  5%  6%  6%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 26% 18% 18%  6%  7%  8%  6%  6%  6%
x264 [info]: ref P L0: 85.5% 12.8%  1.6%
x264 [info]: ref B L0: 93.0%  7.0%
x264 [info]: ref B L1: 94.8%  5.2%
x264 [info]: kb/s:7963.20

encoded 2962 frames, 7.22 fps, 7963.25 kb/s

Results with VBV:

Code:
E:\tests>x264 --crf 15 --fps 29.97 --keyint 29 --me tesa --subme 10 --ref 4 --bframes 3 --deblock -1:-1 --psy-rd 0.0:0.0 --trellis 2 --slices 4 --aq-mode 2 --b-adapt 2 --weightp 0 --vbv-maxrate 40000 --vbv-bufsize 30000 --profile high --level 4.1 --min-keyint 1 --b-pyramid strict --aud --direct auto --partitions all --colorprim "bt709" --transfer "bt709" --colormatrix "bt709" --fullrange "on" --sar 1:1 test.mkv -o 2tesasub10.23.mkv
ffms [info]: 1680x1050p 1:1 @ 30000/1001 fps (cfr)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 FastShuffle SSE4.2
x264 [info]: profile High, level 4.1
x264 [info]: frame I:117   Avg QP:13.78  size:133504
x264 [info]: frame P:1225  Avg QP:17.75  size: 46395
x264 [info]: frame B:1620  Avg QP:20.53  size: 19323
x264 [info]: consecutive B-frames: 17.1% 11.9% 27.3% 43.7%
x264 [info]: mb I  I16..4: 27.4% 42.3% 30.3%
x264 [info]: mb P  I16..4:  6.8% 12.5%  4.5%  P16..4: 20.3% 11.7%  3.8%  1.3%  0
.5%    skip:38.5%
x264 [info]: mb B  I16..4:  1.5%  1.8%  0.7%  B16..8: 32.1%  3.2%  3.2%  direct:
 3.4%  skip:54.0%  L0:38.9% L1:42.3% BI:18.9%
x264 [info]: 8x8 transform intra:49.0% inter:53.4%
x264 [info]: direct mvs  spatial:99.8% temporal:0.2%
x264 [info]: coded y,uvDC,uvAC intra: 49.1% 44.8% 21.6% inter: 15.3% 7.6% 0.8%
x264 [info]: i16 v,h,dc,p: 40% 19% 10% 31%
x264 [info]: i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 26% 17% 22%  5%  6%  7%  5%  6%  6%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 26% 18% 18%  6%  7%  8%  6%  6%  6%
x264 [info]: ref P L0: 85.5% 12.9%  1.6%
x264 [info]: ref B L0: 92.9%  7.1%
x264 [info]: ref B L1: 94.9%  5.1%
x264 [info]: kb/s:8398.69

encoded 2962 frames, 6.90 fps, 8398.75 kb/s

I'd like to know where those 435 kbps come from. Thanks!
Logged

Specs, with: ATI HD5850 - Lastest Catalyst - Vista x64 - MPC-HC x86, EVR Sync

Windows Explorer sucks in Windows 7
Dark Shikari
x264 developer
Administrator
Member
*****
Offline Offline

Posts: 447


View Profile
« Reply #1 on: March 11, 2010, 06:14:57 PM »

Probably because of the way that per-row VBV causes the frame quantizers to be rounded to integer values.
Logged

STaRGaZeR
Member

Online Online

Posts: 7


View Profile
« Reply #2 on: March 12, 2010, 07:44:14 PM »

If that's the case, where are those bits going? If I use a higher crf to compensate so I get the same filesize with vbv enabled and disabled, will both have the same quality or the one with vbv will be worse, assuming the vbv limit was never reached? If it was reached it'll have worse quality in that part of course.
Logged

Specs, with: ATI HD5850 - Lastest Catalyst - Vista x64 - MPC-HC x86, EVR Sync

Windows Explorer sucks in Windows 7
Dark Shikari
x264 developer
Administrator
Member
*****
Offline Offline

Posts: 447


View Profile
« Reply #3 on: March 13, 2010, 05:32:45 AM »

If that's the case, where are those bits going? If I use a higher crf to compensate so I get the same filesize with vbv enabled and disabled, will both have the same quality or the one with vbv will be worse, assuming the vbv limit was never reached? If it was reached it'll have worse quality in that part of course.
It should have roughly the same if the limit is never reached (or gotten close to reaching).
Logged

Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.11 | SMF © 2006-2009, Simple Machines LLC Valid XHTML 1.0! Valid CSS!