Ultimate Profile?

HandBrake for Windows support
Forum rules
An Activity Log is required for support requests. Please read How-to get an activity log? for details on how and why this should be provided.
please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Ultimate Profile?

Post by please_help » Tue Jan 12, 2010 11:50 pm

UPDATE:

I thought I should add some (hopefully) useful information at the top of this thread to save you from having to read it all (I’ve also included this at the current bottom of the thread so you’ll know when this information was produced).

So, if you’re looking for a profile that’s higher than the High Profile then the following may help (sorry for mixing the syntax below, although the settings should still be obvious):

x264 Defaults:

Code: Select all

--subme 7 --bframes 3 --weightb --8x8dct --ref 3 --mixed-refs --trellis 1 --crf 23
The HandBrake High Profile is x264 defaults + crf 20 +:

Code: Select all

b-adapt=2:rc-lookahead=50
If you want to go ‘higher’ then you could add settings from the Slow or Slower x264 presets below to the High Profile (the recommendation is to go for as slow as you can put up with).

x264 Slow (x264 defaults +):

Code: Select all

--me umh --subme 8 --ref 5 --b-adapt 2 --direct auto
x264 Slower (x264 defaults +):

Code: Select all

--me umh --subme 9 --ref 8 --b-adapt 2 --direct auto --partitions all --trellis 2
After a bunch of testing with my computer (4 * 3GHz) and some logical inference, I went for somewhere between the Slow and Slower x264 presets using the following settings (this profile is probably too slow for most people and can in no way be considered 'Ultimate', but I have provided details of results and reasoning in this thread if you're interested).

My proposed profile is the HandBrake High Profile + Strict anamorphic + crf 18 + (copy of the CLI box on the advanced tab below):

Code: Select all

b-adapt=2:rc-lookahead=50:ref=6:me=umh:analyse=all:subq=9:direct=auto:no-fast-pskip=1:no-dct-decimate=1
Cheers

END OF UPDATE[/b]

I hope someone can help with the following.

Thanks in advance, I really appreciate all the work that has been put into HandBrake.

When playing back the largest encode I created with a modified version of the High Profile preset: Strict anamorphic, RF 16, UMH, No Fast P Skip and No DCT Decimate, I got a small amount of tearing with VLC at a couple of points in the movie.

The file size went just over 6GB which is fine for me because the average with this setting seems to be under 3GB (with all subtitles and passthru audio).

Should I reduce the RF and increase other settings to compensate?

I’m basically looking for a ‘perfect’ encode for ‘all’ my movies (I know there’s no such thing, but I want as close as possible with my 4 * 3GHz, 2:1 average encoding or better and not more than 1.5 * movie length encoding time).

I’ve read through x264 presets so I’m now thinking about somewhere between Slow and Slower with RF 18. This gives a decent encoding frame rate, but I’m not sure what settings to increase above Slow. It’s hard to determine the encoding speed versus quality effect of different settings.

I don’t like the fact that DCT-Decimate and Fast-P-Skip might produce blocking artefacts? But they only seem to be added to insane profiles so other factors are considered of higher importance. Although, artefacts would irritate me more than a very slight reduction in quality and I don’t want to check each film thoroughly. So is the following just stupid (it encodes at about the slowest I want for the least compressible movie I’ve tested)?

Code: Select all

b-adapt=2:rc-lookahead=50:me=umh:subq=9:ref=6:direct=auto:no-dct-decimate=1:b-pyramid=1:no-fast-pskip=1:deblock=-1,-1
Should I actually use deblock -1,-1 for films when a few might include a lot of CGI or animation? Could I go back up to RF 16 with the above because it’d produce a smaller encode?

Sorry for the long post and I guess an age-old question.
Last edited by please_help on Sun Jan 17, 2010 8:14 pm, edited 4 times in total.

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Wed Jan 13, 2010 2:35 pm

Could I go back up to RF 16 with the above because it’d produce a smaller encode?
It doesn't produce a smaller encode with strict anamorphic. It's about 100MB larger than:
a modified version of the High Profile preset: Strict anamorphic, RF 16, UMH, No Fast P Skip and No DCT Decimate
Last edited by please_help on Thu Jan 14, 2010 12:21 am, edited 1 time in total.

hunterk
Bright Spark User
Posts: 179
Joined: Tue Jun 03, 2008 2:27 pm

Re: Ultimate Profile?

Post by hunterk » Wed Jan 13, 2010 3:13 pm

Well, I can tell you that b-adapt=2 is a sloooow option that helps quality only a little bit in my experience.

Also, deblock -1,-1 is basically equivalent to running a sharpening filter (think Photoshop) on your movie. Some people like it, but anything other than 0,0 is going to change your output from the original.

Without an activity log, we can't tell whether the tearing you experienced was related to the encoding process or if it's an issue with something else.

As for making a preset that works for everything, you're right to say it's not really possible. I've created 2 custom profiles, 1 for animation and 1 for live action, that work well for me most of the time, but even they require tweaking depending on the source.

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Wed Jan 13, 2010 8:43 pm

Thanks for the response. I hope I can bother you for a bit more information. With reference to the following x264 presets…

Code: Select all

Default: --subme 7 --bframes 3 --weightb --8x8dct --ref 3 --mixed-refs --trellis 1 --crf 23 --threads auto --no-psnr --no-ssim
Slow: --me umh --subme 8 --ref 5 --b-adapt 2 --direct auto
Slower: --me umh --subme 9 --ref 8 --b-adapt 2 --direct auto --partitions all --trellis 2
Film tuning: --deblock -1:-1 --psy-rd 1.0:0.15
b-adapt=2 is a sloooow option that helps quality only a little bit in my experience.
Are you saying that you’d substitute b-adapt=2 for something else that has more effect in contrast to the way the above presets break down?
deblock -1,-1 is basically equivalent to running a sharpening filter
I read kind of the opposite and thought that it’s better described as a blurring filter to get rid of blocking artefacts even at -1,-1. What bothered me in this http://forum.doom9.org/showthread.php?t=109747 explanation from Doom9 regarding the alpha setting was:
The default value of 0 is almost always sufficient to get rid of most blocking (especialy when using a cqm), but leaves the picture noticibly blurier.
Regarding:
Without an activity log, we can't tell whether the tearing you experienced was related to the encoding process
Here it is. I hope you can help.

Code: Select all

### Windows GUI 0.9.4 2009112300 
### Running: Microsoft Windows NT 5.1.2600 Service Pack 3 
###
### CPU: Intel(R) Core(TM)2 Quad CPU    Q6600  @ 2.40GHz 
### Ram: 2039 MB 
### Screen: 1280x1024 
### Temp Dir: C:\Documents and Settings\me\Local Settings\Temp\ 
### Install Dir: C:\Program Files\Handbrake 
### Data Dir: C:\Documents and Settings\me\Application Data\HandBrake\HandBrake\0.9.4.0 
#########################################

### CLI Query:  -i "H:\new" -t 1 --angle 1 -c 1-20 -o "H:\new.mkv" -f mkv --strict-anamorphic  --detelecine --decomb -e x264 -q 16 -a 1,2,3 -E ac3,ac3,ac3 -6 auto,auto,auto -R Auto,Auto,Auto -B 32,32,32 -D 0,0,0 --subtitle 1,1,2,3 --subtitle-forced=1 --subtitle-default=1 --markers="C:\Documents and Settings\me\Local Settings\Temp\new-1-chapters.csv" -x b-adapt=2:rc-lookahead=50:me=umh:no-fast-pskip=1:no-dct-decimate=1:b-pyramid=1 -v 1

### User Query: False

#########################################

[15:52:45] hb_init: checking cpu count
[15:52:45] hb_init: starting libhb thread
HandBrake 0.9.4 (2009112300) - MinGW i386 - http://handbrake.fr
4 CPUs detected
Opening H:\new...
[15:52:45] hb_scan: path=H:\new, title_index=1
[15:52:45] scan: trying to open with libdvdread
libdvdnav: Using dvdnav version 4.1.3
libdvdread: Encrypted DVD support unavailable.
libdvdread: Device (null) inaccessible, CSS authentication not available.
NAME OPEN FAILED
libdvdnav: Unable to find home directorylibdvdnav: DVD disk reports itself with Region mask 0x00000000. Regions: 1 2 3 4 5 6 7 8
libdvdread: Encrypted DVD support unavailable.
libdvdread: Device (null) inaccessible, CSS authentication not available.
[15:52:45] scan: DVD has 1 title(s)
[15:52:45] scan: scanning title 1
[15:52:45] scan: opening IFO for VTS 1
[15:52:45] pgc_id: 1, pgn: 1: pgc: 0x1c9ce00
[15:52:45] scan: vts=1, ttn=1, cells=0->21, blocks=0->3340139, 3340140 blocks
[15:52:45] scan: duration is 02:23:48 (8628000 ms)
[15:52:45] scan: checking audio 1
[15:52:45] scan: id=80bd, lang=English (AC3), 3cc=eng ext=1
[15:52:45] scan: checking audio 2
[15:52:45] scan: id=81bd, lang=English (AC3) (Visually Impaired), 3cc=eng ext=2
[15:52:45] scan: checking audio 3
[15:52:45] scan: id=82bd, lang=English (AC3), 3cc=eng ext=1
[15:52:45] scan: checking subtitle 1
[15:52:45] scan: id=23bd, lang=English, 3cc=eng
[15:52:45] scan: checking subtitle 2
[15:52:45] scan: id=2dbd, lang=English (Closed Caption), 3cc=eng
[15:52:45] scan: checking subtitle 3
[15:52:45] scan: id=2fbd, lang=English, 3cc=eng
[15:52:45] scan: title 1 has 20 chapters
[15:52:45] scan: chap 1 c=0->0, b=0->210629 (210630), 506890 ms
[15:52:45] scan: chap 2 c=1->1, b=210630->349538 (138909), 348502 ms
[15:52:45] scan: chap 3 c=2->2, b=349539->432856 (83318), 196412 ms
[15:52:45] scan: chap 4 c=3->3, b=432857->589334 (156478), 392524 ms
[15:52:45] scan: chap 5 c=4->4, b=589335->771615 (182281), 453667 ms
[15:52:45] scan: chap 6 c=5->5, b=771616->866906 (95291), 237401 ms
[15:52:45] scan: chap 7 c=6->6, b=866907->1016322 (149416), 370518 ms
[15:52:45] scan: chap 8 c=7->7, b=1016323->1273063 (256741), 654915 ms
[15:52:45] scan: chap 9 c=8->8, b=1273064->1475283 (202220), 486746 ms
[15:52:45] scan: chap 10 c=9->10, b=1475284->1681127 (205844), 519755 ms
[15:52:45] scan: chap 11 c=11->11, b=1681128->1843765 (162638), 447610 ms
[15:52:45] scan: chap 12 c=12->12, b=1843766->1974785 (131020), 367635 ms
[15:52:45] scan: chap 13 c=13->14, b=1974786->2166669 (191884), 523950 ms
[15:52:45] scan: chap 14 c=15->15, b=2166670->2365345 (198676), 530908 ms
[15:52:45] scan: chap 15 c=16->16, b=2365346->2486693 (121348), 326676 ms
[15:52:45] scan: chap 16 c=17->17, b=2486694->2675381 (188688), 486896 ms
[15:52:45] scan: chap 17 c=18->18, b=2675382->2859993 (184612), 471928 ms
[15:52:45] scan: chap 18 c=19->19, b=2859994->3033689 (173696), 452816 ms
[15:52:45] scan: chap 19 c=20->20, b=3033690->3184922 (151233), 411627 ms
[15:52:45] scan: chap 20 c=21->21, b=3184923->3340139 (155217), 440611 ms
[15:52:45] scan: aspect = 0
[15:52:45] scan: decoding previews for title 1
[15:52:45] scan: title angle(s) 1
[15:52:45] scan: audio 0x82bd: AC-3, rate=48000Hz, bitrate=192000 English (AC3) (2.0 ch)
[15:52:45] scan: audio 0x80bd: AC-3, rate=48000Hz, bitrate=448000 English (AC3) (5.1 ch)
[15:52:45] scan: audio 0x81bd: AC-3, rate=48000Hz, bitrate=448000 English (AC3) (Visually Impaired) (5.1 ch)
[15:52:45] scan: 10 previews, 720x576, 25.000 fps, autocrop = 76/74/0/2, aspect 16:9, PAR 64:45
[15:52:45] scan: title (0) job->width:720, job->height:304
[15:52:45] libhb: scan thread found 1 valid title(s)
+ title 1:
  + vts 1, ttn 1, cells 0->21 (3340140 blocks)
  + angle(s) 1
  + duration: 02:23:48
  + size: 720x576, pixel aspect: 64/45, display aspect: 1.78, 25.000 fps
  + autocrop: 76/74/0/2
  + chapters:
    + 1: cells 0->0, 210630 blocks, duration 00:08:27
    + 2: cells 1->1, 138909 blocks, duration 00:05:49
    + 3: cells 2->2, 83318 blocks, duration 00:03:16
    + 4: cells 3->3, 156478 blocks, duration 00:06:33
    + 5: cells 4->4, 182281 blocks, duration 00:07:34
    + 6: cells 5->5, 95291 blocks, duration 00:03:57
    + 7: cells 6->6, 149416 blocks, duration 00:06:11
    + 8: cells 7->7, 256741 blocks, duration 00:10:55
    + 9: cells 8->8, 202220 blocks, duration 00:08:07
    + 10: cells 9->10, 205844 blocks, duration 00:08:40
    + 11: cells 11->11, 162638 blocks, duration 00:07:28
    + 12: cells 12->12, 131020 blocks, duration 00:06:08
    + 13: cells 13->14, 191884 blocks, duration 00:08:44
    + 14: cells 15->15, 198676 blocks, duration 00:08:51
    + 15: cells 16->16, 121348 blocks, duration 00:05:27
    + 16: cells 17->17, 188688 blocks, duration 00:08:07
    + 17: cells 18->18, 184612 blocks, duration 00:07:52
    + 18: cells 19->19, 173696 blocks, duration 00:07:33
    + 19: cells 20->20, 151233 blocks, duration 00:06:52
    + 20: cells 21->21, 155217 blocks, duration 00:07:21
  + audio tracks:
    + 1, English (AC3) (5.1 ch) (iso639-2: eng), 48000Hz, 448000bps
    + 2, English (AC3) (Visually Impaired) (5.1 ch) (iso639-2: eng), 48000Hz, 448000bps
    + 3, English (AC3) (2.0 ch) (iso639-2: eng), 48000Hz, 192000bps
  + subtitle tracks:
    + 1, English (iso639-2: eng) (Bitmap)
    + 2, English (Closed Caption) (iso639-2: eng) (Bitmap)
    + 3, English (iso639-2: eng) (Bitmap)
Reading chapter markers from file C:\Documents and Settings\me\Local Settings\Temp\new-1-chapters.csv
Invalid sample rate 0, using input rate 48000
Invalid sample rate 0, using input rate 48000
Invalid sample rate 0, using input rate 48000
[15:52:45] 1 job(s) to process
[15:52:45] starting job
[15:52:45] job configuration:
[15:52:45]  * source
[15:52:45]    + H:\new
[15:52:45]    + title 1, chapter(s) 1 to 20
[15:52:45]  * destination
[15:52:45]    + H:\new.mkv
[15:52:45]    + container: Matroska (.mkv)
[15:52:45]      + chapter markers
[15:52:45]  * video track
[15:52:45]    + decoder: mpeg2
[15:52:45]      + bitrate 9800 kbps
[15:52:45]    + frame rate: same as source (around 25.000 fps)
[15:52:45]    + strict anamorphic
[15:52:45]      + modulus: 0
[15:52:45]      + storage dimensions: 720 * 576 -> 718 * 426, crop 76/74/0/2
[15:52:45]      + pixel aspect ratio: 64 / 45
[15:52:45]      + display dimensions: 1021 * 426
[15:52:45]    + filters
[15:52:45]      + Detelecine (pullup) (default settings)
[15:52:45]      + Decomb (default settings)
[15:52:45]    + encoder: x264
[15:52:45]      + options: b-adapt=2:rc-lookahead=50:me=umh:no-fast-pskip=1:no-dct-decimate=1:b-pyramid=1
[15:52:45]      + quality: 16.00 (RF)
[15:52:45]  * subtitle track 1, English (id 23bd) Picture [VOBSUB] -> Pass-Through 
[15:52:45]  * subtitle track 1, English (id 23bd) Picture [VOBSUB] -> Pass-Through 
[15:52:45]  * subtitle track 2, English (Closed Caption) (id 2dbd) Picture [VOBSUB] -> Pass-Through 
[15:52:45]  * subtitle track 3, English (id 2fbd) Picture [VOBSUB] -> Pass-Through 
[15:52:45]  * audio track 0
[15:52:45]    + decoder: English (AC3) (5.1 ch) (track 1, id 80bd)
[15:52:45]      + bitrate: 448 kbps, samplerate: 48000 Hz
[15:52:45]    + AC3 passthrough
[15:52:45]  * audio track 1
[15:52:45]    + decoder: English (AC3) (Visually Impaired) (5.1 ch) (track 2, id 81bd)
[15:52:45]      + bitrate: 448 kbps, samplerate: 48000 Hz
[15:52:45]    + AC3 passthrough
[15:52:45]  * audio track 2
[15:52:45]    + decoder: English (AC3) (2.0 ch) (track 3, id 82bd)
[15:52:45]      + bitrate: 192 kbps, samplerate: 48000 Hz
[15:52:45]    + AC3 passthrough
libdvdnav: Using dvdnav version 4.1.3
libdvdread: Encrypted DVD support unavailable.
libdvdread: Device (null) inaccessible, CSS authentication not available.
NAME OPEN FAILED
libdvdnav: Unable to find home directorylibdvdnav: DVD disk reports itself with Region mask 0x00000000. Regions: 1 2 3 4 5 6 7 8
libdvdread: Encrypted DVD support unavailable.
libdvdread: Device (null) inaccessible, CSS authentication not available.
[15:52:45] decomb thread started for segment 0
[15:52:45] yadif thread started for segment 1
[15:52:45] yadif thread started for segment 3
[15:52:45] yadif thread started for segment 2
[15:52:45] yadif thread started for segment 0
[15:52:45] decomb thread started for segment 1
[15:52:45] decomb thread started for segment 2
[15:52:45] decomb thread started for segment 3
[15:52:45] reader: first SCR 146 id 224 DTS 16693
[15:52:45] mpeg2: "Chapter 1" (1) at frame 0 time 3600
[15:52:45] encx264: encoding with stored aspect 64/45
[15:52:45] encx264: Encoding at constant RF 16.000000
x264 [warning]: b-pyramid + mb-tree is not supported
x264 [info]: using SAR=64/45
x264 [info]: using cpu capabilities: MMX2 SSE2Fast SSSE3 Cache64
x264 [info]: profile High, level 3.0
No accelerated IMDCT transform found
No accelerated IMDCT transform found
No accelerated IMDCT transform found
[15:52:45] sync: expecting 215724 video frames
[ac3 @ 0x5c998f0]No channel layout specified. The encoder will guess the layout, but it might be incorrect.
[15:52:45] sync: first pts is 3600
[16:01:23] mpeg2: "Chapter 2" (2) at frame 12667 time 45604800
[16:07:06] mpeg2: "Chapter 3" (3) at frame 21379 time 76968000
[16:10:25] mpeg2: "Chapter 4" (4) at frame 26303 time 94694400
[16:16:43] mpeg2: "Chapter 5" (5) at frame 36124 time 130050000
[16:24:08] mpeg2: "Chapter 6" (6) at frame 47462 time 170866800
[16:28:03] mpeg2: "Chapter 7" (7) at frame 53390 time 192207600
[16:34:16] mpeg2: "Chapter 8" (8) at frame 62642 time 225514800
[16:44:57] mpeg2: "Chapter 9" (9) at frame 79015 time 284457600
[16:53:53] mpeg2: "Chapter 10" (10) at frame 91179 time 328248000
[17:02:06] mpeg2: "Chapter 11" (11) at frame 104177 time 375042593
[17:08:34] mpeg2: "Chapter 12" (12) at frame 115354 time 415279793
[17:14:26] mpeg2: "Chapter 13" (13) at frame 124534 time 448327793
[17:22:47] mpeg2: "Chapter 14" (14) at frame 137627 time 495462593
[17:30:54] mpeg2: "Chapter 15" (15) at frame 150884 time 543187793
[17:36:11] mpeg2: "Chapter 16" (16) at frame 159051 time 572588993
[17:41:32] hb_demux_ps: not a PS packet (bdd09c29)
[17:41:32] hb_demux_ps: not a PS packet (2a661f90)
[17:41:32] hb_demux_ps: not a PS packet (ce9237e4)
[17:41:32] hb_demux_ps: not a PS packet (9b6a52ca)
[17:41:32] hb_demux_ps: not a PS packet (2d7be237)
[17:41:32] hb_demux_ps: not a PS packet (739e62ae)
[17:41:32] hb_demux_ps: not a PS packet (d793b3b2)
[17:41:32] hb_demux_ps: not a PS packet (6889ab7e)
[17:41:32] hb_demux_ps: not a PS packet (e9bd4d05)
[17:41:32] hb_demux_ps: not a PS packet (586ab62d)
[17:41:32] hb_demux_ps: not a PS packet (d65e3c89)
[17:41:32] hb_demux_ps: not a PS packet (8f5c981e)
[17:41:32] hb_demux_ps: not a PS packet (bb3f12fb)
[17:41:32] hb_demux_ps: not a PS packet (41847ff3)
[17:41:32] hb_demux_ps: not a PS packet (c92a94da)
[17:41:32] hb_demux_ps: not a PS packet (bd2b06e9)
[17:41:32] hb_demux_ps: not a PS packet (f2a1f172)
[17:41:32] hb_demux_ps: not a PS packet (56587853)
[17:41:32] hb_demux_ps: not a PS packet (218ac0e5)
[17:41:32] hb_demux_ps: not a PS packet (d046585a)
[17:41:32] hb_demux_ps: not a PS packet (853eff7f)
[17:41:32] hb_demux_ps: not a PS packet (c260a6dc)
[17:41:32] hb_demux_ps: not a PS packet (5c672ea6)
[17:41:32] hb_demux_ps: not a PS packet (7bc1459f)
[17:41:32] hb_demux_ps: not a PS packet (83e61882)
[17:41:32] hb_demux_ps: not a PS packet (0064bfca)
[17:41:32] hb_demux_ps: not a PS packet (eada212a)
[17:41:32] hb_demux_ps: not a PS packet (4ec54281)
[17:41:32] hb_demux_ps: not a PS packet (59aae54d)
[17:41:32] hb_demux_ps: not a PS packet (4e155482)
[17:41:32] hb_demux_ps: not a PS packet (7fcebd5b)
[17:41:32] hb_demux_ps: not a PS packet (cf9a2092)
[17:41:32] hb_demux_ps: not a PS packet (4ffd1c18)
[17:41:32] hb_demux_ps: not a PS packet (9ac7b341)
[17:41:32] hb_demux_ps: not a PS packet (08000280)
[17:41:32] hb_demux_ps: not a PS packet (26e7023d)
[17:41:32] hb_demux_ps: not a PS packet (9db11d73)
[17:41:32] hb_demux_ps: not a PS packet (d9b91837)
[17:41:32] hb_demux_ps: not a PS packet (ba452bd0)
[17:41:32] hb_demux_ps: not a PS packet (b55aa809)
[17:41:32] hb_demux_ps: not a PS packet (af008426)
[17:41:32] hb_demux_ps: not a PS packet (e006007b)
[17:41:32] hb_demux_ps: not a PS packet (5c100010)
[17:41:32] hb_demux_ps: not a PS packet (ab171ef8)
[17:41:32] hb_demux_ps: not a PS packet (418d763e)
[17:41:32] hb_demux_ps: not a PS packet (fac38a29)
[17:41:32] hb_demux_ps: not a PS packet (56c63867)
[17:41:32] hb_demux_ps: not a PS packet (bbad5894)
[17:41:32] hb_demux_ps: not a PS packet (75ee651c)
[17:41:32] hb_demux_ps: not a PS packet (ce598d34)
[17:41:32] hb_demux_ps: not a PS packet (52458379)
[17:41:32] hb_demux_ps: not a PS packet (16a602af)
[17:41:32] hb_demux_ps: not a PS packet (2ed851d2)
[17:41:32] hb_demux_ps: not a PS packet (de274e33)
[17:41:32] hb_demux_ps: not a PS packet (aa940043)
[17:41:32] hb_demux_ps: not a PS packet (8b8bfb43)
[17:41:32] hb_demux_ps: not a PS packet (13169332)
[17:41:32] hb_demux_ps: not a PS packet (191bb005)
[17:41:32] hb_demux_ps: not a PS packet (74ac3a2a)
[17:41:32] hb_demux_ps: not a PS packet (49f514f2)
[17:41:32] hb_demux_ps: not a PS packet (f66a7f8c)
[17:41:32] hb_demux_ps: not a PS packet (2e5db276)
[17:41:32] hb_demux_ps: not a PS packet (d52c0e81)
[17:41:32] hb_demux_ps: not a PS packet (e928d61f)
[17:41:32] hb_demux_ps: not a PS packet (0e5f51f4)
[17:41:32] hb_demux_ps: not a PS packet (0bee5dbe)
[17:41:32] hb_demux_ps: not a PS packet (afd9994b)
[17:41:32] hb_demux_ps: not a PS packet (a18299de)
[17:41:32] hb_demux_ps: not a PS packet (03deca47)
[17:41:32] hb_demux_ps: not a PS packet (b6d4b622)
[17:41:32] hb_demux_ps: not a PS packet (a7b5adf7)
[17:41:32] hb_demux_ps: not a PS packet (b1167be9)
[17:41:32] hb_demux_ps: not a PS packet (258cfb4a)
[17:41:32] hb_demux_ps: not a PS packet (6151af12)
[17:41:32] hb_demux_ps: not a PS packet (50ca3125)
[17:41:32] hb_demux_ps: not a PS packet (01d0d466)
[17:41:32] hb_demux_ps: not a PS packet (e41d1812)
[17:41:32] hb_demux_ps: not a PS packet (da23bb87)
[17:41:32] hb_demux_ps: not a PS packet (9b9ce9dd)
[17:41:32] hb_demux_ps: not a PS packet (3069d688)
[17:41:32] hb_demux_ps: not a PS packet (d6628aa7)
[17:41:32] hb_demux_ps: not a PS packet (7f0bbea1)
[17:41:32] hb_demux_ps: not a PS packet (462f3d74)
[17:41:32] hb_demux_ps: not a PS packet (6be052c4)
[17:41:32] hb_demux_ps: not a PS packet (639e42df)
[17:41:32] hb_demux_ps: not a PS packet (b120c749)
[17:41:32] hb_demux_ps: not a PS packet (9aaa227d)
[17:41:32] hb_demux_ps: not a PS packet (bf50de3a)
[17:41:32] hb_demux_ps: not a PS packet (82b55074)
[17:41:32] hb_demux_ps: not a PS packet (614e7e1e)
[17:41:32] hb_demux_ps: not a PS packet (278bdb55)
[17:41:32] hb_demux_ps: not a PS packet (ad66a63e)
[17:41:32] hb_demux_ps: not a PS packet (0149c91b)
[17:41:32] hb_demux_ps: not a PS packet (0d2ecd9c)
[17:41:32] hb_demux_ps: not a PS packet (a8b019e0)
[17:41:32] hb_demux_ps: not a PS packet (ffffffda)
[17:41:32] hb_demux_ps: not a PS packet (f3576f6e)
[17:41:32] hb_demux_ps: not a PS packet (00f0500a)
[17:41:32] hb_demux_ps: not a PS packet (0769def2)
[17:41:32] hb_demux_ps: not a PS packet (51dca302)
[17:41:32] hb_demux_ps: not a PS packet (def74bf2)
[17:41:32] hb_demux_ps: not a PS packet (a9b9e6e8)
[17:41:32] hb_demux_ps: not a PS packet (04859fa5)
[17:41:32] hb_demux_ps: not a PS packet (d52b77a9)
[17:41:32] hb_demux_ps: not a PS packet (6c41f501)
[17:41:32] hb_demux_ps: not a PS packet (a5e1b6f7)
[17:41:32] hb_demux_ps: not a PS packet (de477f77)
[17:41:32] hb_demux_ps: not a PS packet (f44f4a9d)
[17:41:32] output track 2: ac3 in sync after skipping 576 bytes
[17:41:33] sync: adding 235 ms of silence to audio 2  start 601517873, next 601496640
[17:41:33] output track 1: ac3 in sync after skipping 448 bytes
[17:41:33] output track 0: ac3 in sync after skipping 448 bytes
[17:41:33] sync: adding 267 ms of silence to audio 0  start 601520753, next 601496640
[17:41:33] sync: adding 267 ms of silence to audio 1  start 601520753, next 601496640
[17:44:48] mpeg2: "Chapter 17" (17) at frame 171201 time 616357793
[17:53:14] mpeg2: "Chapter 18" (18) at frame 182995 time 658816193
[18:01:33] mpeg2: "Chapter 19" (19) at frame 194302 time 699521393
[18:08:51] mpeg2: "Chapter 20" (20) at frame 204600 time 736594193
[18:13:24] reader: done. 1 scr changes
[18:13:26] sync: got 215609 frames, 215724 expected
[18:13:26] work: average encoding speed for job is 25.542732 fps
[18:13:28] mux: track 0, 215587 frames, 5495832793 bytes, 5097.68 kbps, fifo 128
[18:13:28] mux: track 1, 269527 frames, 482992384 bytes, 448.00 kbps, fifo 512
[18:13:28] mux: track 2, 269527 frames, 482992384 bytes, 448.00 kbps, fifo 512
[18:13:28] mux: track 3, 269527 frames, 206996736 bytes, 192.00 kbps, fifo 512
[18:13:28] mux: track 4, 0 frames, 0 bytes, 0.00 kbps, fifo 8
[18:13:28] mux: track 5, 0 frames, 0 bytes, 0.00 kbps, fifo 8
[18:13:28] mux: track 6, 1977 frames, 4564194 bytes, 4.23 kbps, fifo 8
[18:13:28] mux: track 7, 2453 frames, 6940658 bytes, 6.44 kbps, fifo 8
[18:13:28] mpeg2 done: 215610 frames
[18:13:28] render: lost time: 82800 (23 frames)
[18:13:28] render: gained time: 82800 (89 frames) (0 not accounted for)
[18:13:28] render: average dropped frame duration: 3600
[18:13:28] encx264: 1 frames had to be split (215585 in, 215651 out)
x264 [info]: frame I:2296  Avg QP:14.47  size: 80551  PSNR Mean Y:47.87 U:50.25 V:50.38 Avg:48.53 Global:48.28
x264 [info]: frame P:82228 Avg QP:17.95  size: 38741  PSNR Mean Y:44.12 U:47.22 V:47.27 Avg:44.90 Global:44.52
x264 [info]: frame B:131063 Avg QP:20.46  size: 16217  PSNR Mean Y:42.52 U:46.79 V:46.77 Avg:43.47 Global:43.09
x264 [info]: consecutive B-frames:  3.3% 19.2% 75.0%  2.4%
x264 [info]: mb I  I16..4:  6.8% 51.0% 42.2%
x264 [info]: mb P  I16..4:  2.0% 18.6%  8.8%  P16..4: 27.9% 25.5% 15.1%  0.0%  0.0%    skip: 2.2%
x264 [info]: mb B  I16..4:  0.4%  4.2%  2.7%  B16..8: 43.7%  6.1%  7.0%  direct:15.2%  skip:20.8%  L0:29.1% L1:29.0% BI:41.9%
x264 [info]: 8x8 transform intra:61.1% inter:38.2%
x264 [info]: coded y,uvDC,uvAC intra: 93.9% 94.9% 84.2% inter: 55.8% 50.3% 27.6%
x264 [info]: i16 v,h,dc,p: 29% 13% 13% 44%
x264 [info]: i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 21% 12% 25%  6%  6%  8%  6%  9%  7%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 23% 12% 15%  7%  8% 11%  7% 10%  7%
x264 [info]: Weighted P-Frames: Y:4.5%
x264 [info]: ref P L0: 64.4%  9.3% 20.3%  5.1%  0.8%
x264 [info]: ref B L0: 91.8%  8.2%
x264 [info]: SSIM Mean Y:0.9843289
x264 [info]: PSNR Mean Y:43.187 U:46.989 V:46.995 Avg:44.072 Global:43.616 kb/s:5098.59
[18:13:29] decomb: deinterlaced 0 | blended 645 | unfiltered 214940 | total 215585
[18:13:29] libhb: work result = 0

Rip done!
HandBrake has exited.

 ############ End of Log ##############
What are your custom profiles?

And I guess the final question I’m really asking is; if the HandBrake developers were going to add an Ultimate Profile what would it be?

In my opinion it should be somewhere around an average 20-30% hit in speed and 30-40% increase in file size above the High Profile, so what would be the best settings for this?

hunterk
Bright Spark User
Posts: 179
Joined: Tue Jun 03, 2008 2:27 pm

Re: Ultimate Profile?

Post by hunterk » Wed Jan 13, 2010 10:34 pm

b-adapt=2 has reduced my fps by as much as 50%. I've been able to discern a difference in quality between b-adapt=1 and b-adapt=2, but it was slight enough that I didn't think it justified the speed reduction, so I stick with a value of 1. Otherwise, that preset looks pretty similar to what I use.

For deblocking, 0,0 is almost always your best bet. I've never noticed any blurring at default, but I could be wrong. If you do some test encodes with lower (i.e., negative) values, you can see the sharpening happening, though it's hard to notice unless you pick a silly value like -6,-6. Values >1 are definitely blurred. Like the post you linked said: deblocking is one of the more subjective settings, so just play around until you find something that works for you.

The preset you posted uses -1,-1 for film sources, which could be designed to accentuate film grain. This would also make sense in conjunction with the psy-rd setting next to it.

Based on your activity log, the tearing problem might be related to your use of pyramidal b-frames alongside mb-tree (which is turned on by default). If you notice, it threw up a warning right in the middle of the log.

There were a lot of changes to x264 right around the 0.9.4 release related to b-pyramid and mb-tree and these settings were incompatible until svn 3017 (after 0.9.4 release). You might try re-encoding without b-pyramid, which will hurt your complexity a bit (i.e., slightly larger filesize at a given quality).

If anyone disagrees with any of this, feel free to correct me. I'm no expert.

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Wed Jan 13, 2010 11:38 pm

Thanks again for the info.
b-adapt=2 has reduced my fps by as much as 50%
Crikey, I didn’t get anything like that reduction in fps.
For deblocking, 0,0 is almost always your best bet…The preset you posted uses -1,-1 for film sources, which could be designed to accentuate film grain.
That’s what I’ve read. I’ve been looking at the x264 defaults, standard presets, tunings and Sharktooth’s profiles. They use -1,-1 for films, -2,-2 for grain and 1,1 for animation. I think I’d rather stick with one ‘Ultimate Profile’ so I guess I’ll go for 0,0.
Based on your activity log, the tearing problem might be related to your use of pyramidal b-frames alongside mb-tree
I noticed that but I assumed that if it’s not supported, it’d just be automatically ignored by the encoder. Also, sorry about this but, the rip that had the tearing I actually did on Linux so there wasn’t a pyramidal b-frames option. I apologise for posting a re-encode with the same settings (except for pyramidal b-frames) but on Windows. I checked the film again and the tearing seems to be gone? So I guess we’ll never know what the problem was.

One question though; does anyone think that a 6GB 2 hour playback with 3 passthru audio tracks and all subtitles in VLC from a hard disk should have a problem?

Any suggestions for an ‘Ultimate Profile’ that would complement Normal and High?

The explanation could be something like the following (updated percentages from my last post):

The High Profile is great for all encodes but if you want to waste some computing time and disk space for a marginal increase in quality then you could go for the Ultimate Profile. The encode will be on average 20% slower (with around a 4 * 3GHz computer, other computers could be a lot slower) and 20% larger than if you use the High Profile (so some encodes could be very close in size to your source).

Weird that I’d write a profile explanation above (and not a very good one at that) but that’s basically what I want.

jbrjake
Veteran User
Posts: 4805
Joined: Wed Dec 13, 2006 1:38 am

Re: Ultimate Profile?

Post by jbrjake » Thu Jan 14, 2010 4:10 pm

....not using b-adapt 2 is a terrible move and I'm puzzled as to why you're gauging it by how it affects quality when you're using a constant quality rate control method.

creamyhorror
Regular User
Posts: 134
Joined: Mon Jan 04, 2010 2:00 pm

Re: Ultimate Profile?

Post by creamyhorror » Thu Jan 14, 2010 4:42 pm

The way I understand it, H.264 always deblocks its encode. 0,0 is just a default value; reducing it to -2,-2 doesn't "sharpen" the image, but rather reduces the strength of the deblock. Too little deblocking and you start seeing blocks clearly; too much, and you get a smudgy mess. I'd recommend going with -3,-3 or -2,-2 on complex/grainy material.

hunterk
Bright Spark User
Posts: 179
Joined: Tue Jun 03, 2008 2:27 pm

Re: Ultimate Profile?

Post by hunterk » Thu Jan 14, 2010 8:54 pm

ah, you guys are right. deblocking -6,-6 is essentially no deblocking, apparently. It always has that over-sharpened look to me, and I had assumed 0,0 was baseline, hence my confusion. :oops:

As for b-adapt, my comparisons were conducted with constant bitrate rather than constant quality, which is why I mentioned its impact on quality. I guess I should go back and compare them with constant quality and look at the file sizes instead of basing it on my own subjective quality comparison.

Here's what the x264 mewiki says about b-adapt:
0. Disabled. Pick B-frames always. This is the same as what the older --no-b-adapt setting did.
1. Old algorithm, faster, speed slightly increases with higher b-frames setting.
2. New algorithm, slower, speed significantly decreases with higher b-frames setting.
Recommendation: Default, use 2 if you have time to waste.
I usually use a lot of b-frames, so this could explain why I have such a tremendous speed hit with b-adapt=2, esp in comparison to b-adapt=1.

jbrjake, could you elaborate a bit on why you recommend against b-adapt=1?

/threadjack

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Thu Jan 14, 2010 8:59 pm

So I guess no one wants to jump in with an actual answer to the fundamental question:

What settings would an Ultimate Profile (or slightly insane profile?) have that would complement the Normal and High settings at the moment?

I understand the simplicity of the current system and the fact that things like the film and television profiles were removed. I also understand that one issue with including an Ultimate Profile is that it could cause people to see the High Profile as substandard.

I probably confused this topic with certain aspects of my previous posts. However, the problem with devising one myself is probably best shown with the following (sorry for mixing the syntax below and I haven’t included all the settings in each, but you should get the idea):

x264 Defaults:

Code: Select all

--subme 7 --bframes 3 --weightb --8x8dct --ref 3 --mixed-refs --trellis 1 --crf 23
x264 Slow (defaults +):

Code: Select all

--me umh --subme 8 --ref 5 --b-adapt 2 --direct auto
x264 Slower (defaults +):

Code: Select all

--me umh --subme 9 --ref 8 --b-adapt 2 --direct auto --partitions all --trellis 2
Old HandBrake Constant Quality (I guess selected to make the best of an older x264 version):

Code: Select all

ref=3:mixed-refs:bframes=3:b-pyramid:weightb:filter=-2,-1:trellis=1:analyse=all:8x8dct:me=umh:subme=9:psy-rd=1,1
New HandBrake High Profile (x264 defaults + crf 20 +):

Code: Select all

b-adapt=2:rc-lookahead=50
Sharktooth’s x264 Unrestricted 1pass Const. Quality HQ:

Code: Select all

--crf 18 --deblock -1:-1 --bframes 3 --b-adapt 2 --b-pyramid --b-bias 0 --scenecut 40 --ref 5 --rc-lookahead 40 --no-mbtree --aq-mode 1 --aq-strength 1.0 --merange 16 --me umh --subme 6 --partitions all --trellis 2 --psy-rd 1.0:0
I don’t think the ‘Ultimate Profile’ should go anywhere near as far as Sharktooth’s insane profile. I don’t even think it needs to go all the way to the HQ profile above. I would just like to get a bit of expert opinion on what a complimentary higher than High Profile would be.

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Fri Jan 15, 2010 12:18 am

So after lots of testing with the least compressible movie I’ve found I can suggest an Ultimate Profile (edit: obviously this isn’t actually an Ultimate Profile, just a suggestion with a silly name that seems to annoy some people). Strict anamorphic with RF 18 and the following:

Code: Select all

b-adapt=2:rc-lookahead=50:ref=6:me=umh:analyse=all:subq=9:direct=auto:no-fast-pskip=1:no-dct-decimate=1
This testing was conducted with a 4*3GHz (overclocked) computer to get a decent encode frame rate that’s just over 20% slower than the movie plays. This is obviously a worst case scenario so most movies would encode a lot quicker. Also, the movie I was testing with will only compress to around 75% of the original with the above settings (with all passthru AC3 tracks and all subtitles included), but again this is a worst case scenario so on average movies would compress to less than 40% of the source size (I haven’t done enough testing yet, so I’ll post more accurate figures later because the compression ratio might actually be much better than stated above).

The following are the reasons for each choice. Can someone check my logic please (it’s only based on other profiles and encoding time)?

Strict anamorphic: I want all the pixels of the movie and I don’t want any resizing to be conducted to make the encoding more efficient because this reduces quality.

RF 18: It’s incrementally better than the High Profile and is used in Sharktooth’s HQ profile.

b-adapt=2: Because I don’t want to go below the High Profile and it’s in the slow, slower and Sharktooth x264 presets.

rc-lookahead=50: Again, I don’t want to go below the High Profile.

ref=6: Because it’s between slow and slower but is still within the “sane” threshold mentioned in HB docs.

me=umh: Because it’s in the slow, slower, old HB and Sharktooth’s presets.

analyse=all: It’s in the slower, old HB and Sharktooth’s presets.

subq=9: Because it’s in the slower, old HB and Sharktooth’s profiles.

direct=auto: It’s in the slow and slower profiles.

no-fast-pskip=1: Because it was a recommended setting (although not any more) and with this worst case scenario encode it makes 0.0001% difference in encoding time so I’d rather not run the risk of having any blocking.

no-dct-decimate=1: Because I don’t like the fact that it could make a mistake and cause blocking. I don’t mind the decrease in encoding efficiency to prevent this.

Settings I didn’t choose like trellis=2 caused too large a hit in speed. I was trying to go for somewhere between the slow and slower x264 presets using the other profiles in my previous post to help with decisions. For example the old HB preset went for analyse=all but not trellis=2.

I’ve gone for deblocking 0,0 so the profile is compatible with ‘all’ sources.

There is more logic behind this, but I guess I’ll wait to be chastised by developers and moderators before writing any more.
Last edited by please_help on Mon Jan 18, 2010 2:26 am, edited 1 time in total.

hunterk
Bright Spark User
Posts: 179
Joined: Tue Jun 03, 2008 2:27 pm

Re: Ultimate Profile?

Post by hunterk » Fri Jan 15, 2010 2:03 pm

That looks decent to me.

Did you try using trellis=1 to see how it affected speed and output? In my experience, any trellis has been an improvement, though trellis=2 may not be worth it in all cases.

Edit: duh, right. you left it at default of 1.
Last edited by hunterk on Fri Jan 15, 2010 3:15 pm, edited 1 time in total.

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Fri Jan 15, 2010 3:07 pm

Thanks for the reply. The settings provided have been added on top of the High Profile settings and are just a copy of the CLI box at the bottom of the Advanced tab (therefore they include trellis 1).

To reiterate slightly but perhaps in a more concise format:

I’ve gone for the HandBrake High Profile, added the x264 Slow presets and added some aspects from the x264 Slower presets using other profiles, research and encoding time as a guide.

I’ve also added no-fast-pskip=1 and no-dct-decimate=1 for the reasons given above and they are included in slower profiles than provided in my earlier post.

jbrjake
Veteran User
Posts: 4805
Joined: Wed Dec 13, 2006 1:38 am

Re: Ultimate Profile?

Post by jbrjake » Fri Jan 15, 2010 3:41 pm

hunterk wrote:jbrjake, could you elaborate a bit on why you recommend against b-adapt=1?
...because b-adapt 2 lets you use fewer b-frames AND get smaller file sizes AND get higher quality metrics?

TedJ
Veteran User
Posts: 5388
Joined: Wed Feb 20, 2008 11:25 pm

Re: Ultimate Profile?

Post by TedJ » Sat Jan 16, 2010 1:37 am

Is that all? Those guys at libx264 are downright lazy. :P

creamyhorror
Regular User
Posts: 134
Joined: Mon Jan 04, 2010 2:00 pm

Re: Ultimate Profile?

Post by creamyhorror » Sat Jan 16, 2010 8:01 am

please_help wrote:So I guess no one wants to jump in with an actual answer to the fundamental question:

What settings would an Ultimate Profile (or slightly insane profile?) have that would complement the Normal and High settings at the moment?
Why are you calling it an Ultimate Profile when it only fits your own priorities and your own definition of "ultimate"?

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Sat Jan 16, 2010 3:48 pm

That’s kind of a fair point, but not really what I’m saying.

‘Ultimate Profile’ is perhaps a bad choice of words, maybe higher than high or slow would be better. I don’t really care what it’s called, feel free to make suggestions.

Also, I started this thread to get other peoples opinions, not just voice my own (and to get some help). I was only making suggestions of what ‘I’ would be looking for and hoped that other people would share their ideas. No one got involved with an actual profile suggestion, so I decided to make one myself and again hoped that other people would provide suggestions.

At the end of the day I would have preferred the HB developers to make a profile above High that I could have just used instead of starting this thread, but I understand there are reasons behind this.

Cheers

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Sat Jan 16, 2010 5:42 pm

So, in case anyone cares (I’ve repeated info here to save people from having to read this thread)…

After encoding 12 random movies (although this does include the two slowest to encode and least compressible I’ve found), with all English audio tracks (passthru) and all English subtitles included (from sources that only included English subs and audio), I got the following with my proposed ‘Ultimate Profile’ on a 4 * 3GHz computer:

High Profile + Strict anamorphic + crf 18 + (copy of CLI box below)

Code: Select all

b-adapt=2:rc-lookahead=50:ref=6:me=umh:analyse=all:subq=9:direct=auto:no-fast-pskip=1:no-dct-decimate=1
Average encoding speed: 23.9 fps
Slowest encoding speed: 16.4 fps
Fastest encoding speed: 29.3 fps

Average file size: 2.7 GB (which is roughly 60% less than the source on average)
Smallest file size: 1.3 GB (source size 4.4 GB)
Largest file size: 5.2 GB (source size 6.8 GB)

Highest compression: 73% less than source
Lowest compression: 22% less than source

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Sat Jan 16, 2010 11:59 pm

So, if you’re looking for a profile that’s higher than the High Profile then the following may help (sorry for mixing the syntax below, although the settings should still be obvious):

x264 Defaults:

Code: Select all

--subme 7 --bframes 3 --weightb --8x8dct --ref 3 --mixed-refs --trellis 1 --crf 23
The HandBrake High Profile is x264 defaults + crf 20 +:

Code: Select all

b-adapt=2:rc-lookahead=50
If you want to go ‘higher’ then you could add settings from the Slow or Slower x264 presets below to the High Profile (the recommendation is to go for as slow as you can put up with).

x264 Slow (x264 defaults +):

Code: Select all

--me umh --subme 8 --ref 5 --b-adapt 2 --direct auto
x264 Slower (x264 defaults +):

Code: Select all

--me umh --subme 9 --ref 8 --b-adapt 2 --direct auto --partitions all --trellis 2
After a bunch of testing with my computer (4 * 3GHz) and some logical inference, I went for somewhere between the Slow and Slower x264 presets using the following settings (this profile is probably too slow for most people and can in no way be considered 'Ultimate', but I have provided details of results and reasoning in this thread if you're interested).

My proposed profile is the HandBrake High Profile + Strict anamorphic + crf 18 + (copy of the CLI box on the advanced tab below):

Code: Select all

b-adapt=2:rc-lookahead=50:ref=6:me=umh:analyse=all:subq=9:direct=auto:no-fast-pskip=1:no-dct-decimate=1
Cheers[/b]
Last edited by please_help on Sun Jan 17, 2010 8:15 pm, edited 2 times in total.

creamyhorror
Regular User
Posts: 134
Joined: Mon Jan 04, 2010 2:00 pm

Re: Ultimate Profile?

Post by creamyhorror » Sun Jan 17, 2010 3:59 am

please_help wrote: Also, I started this thread to get other peoples opinions, not just voice my own (and to get some help). I was only making suggestions of what ‘I’ would be looking for and hoped that other people would share their ideas. No one got involved with an actual profile suggestion, so I decided to make one myself and again hoped that other people would provide suggestions.
I'm guessing people have already found settings that work for them (I bet most people have their own preferred higher-than-High-Profile profile). Your profile works for your particular priorities, but other people might want even slower options, or faster ones, or ones catered to animation or grainy sources. I have no problem with you recommending settings that you like, but calling it "Ultimate" just sounds grandiose (especially when you seem to be quite new to x264 encoding).

By the way, the "Post Your "Best Settings" here and Why" thread in the General forum was locked, and it's pretty similar to this thread.
At the end of the day I would have preferred the HB developers to make a profile above High that I could have just used instead of starting this thread, but I understand there are reasons behind this.
They're best positioned to answer why they didn't want to include slower presets. I don't know myself, but I never use inbuilt presets in programs anyway.

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Sun Jan 17, 2010 4:43 pm

Fair enough, no point in flogging a dead horse. I’ll just give it one more kick for luck:
I'm guessing people have already found settings that work for them (I bet most people have their own preferred higher-than-High-Profile profile).
Yes, I’m sure a lot of them have. I wanted to start a discussion about what would compliment the current Normal and High Profile i.e. be a similar step up to a higher-than-High-Profile as it is from the Normal to High Profile. Either way, this was an epic fail.
Your profile works for your particular priorities, but other people might want even slower options, or faster ones, or ones catered to animation or grainy sources.
Well it appears to work for me but I’m not convinced that it satisfies even my own criteria in the most efficient and effective way, given no real feedback. Also, my priorities were open for discussion within the context of the question I posed. As for people wanting slower, faster, animation, or grainy profiles, this is true, although it falls outside the scope of my question except when posed in such a way as to relate to a complimentary higher-than-High-Profile that would cater to ‘most’ sources and ‘average’ computing power. Therefore, I don’t think my proposed profile is complimentary enough. The answer is probably to go for something like the x264 Slow presets and maybe a slightly higher crf? But again, I’ve epically failed to induce any discussion about this.
I have no problem with you recommending settings that you like
That’s very noble of you.
but calling it "Ultimate" just sounds grandiose
Yep, I guess a very bad choice of words. It took me all of a split second to come up with that one. But as you can see from the title of the thread, I did add a question mark, so I wasn’t saying “check me out and my amazing settings”, I was posing a question. Apparently I went about it in completely the wrong way. Oh well, life goes on.
especially when you seem to be quite new to x264 encoding
You should have just used “newbie”. New-ish, yes, completely uninformed, no.
the "Post Your "Best Settings" here and Why" thread in the General forum was locked, and it's pretty similar to this thread
I don’t think so. I’m the only person who has posted a ‘profile’ here and my intention was to come up with a single complimentary higher-than-High-Profile. But I failed, so I just posted what I came up with, which is almost certainly too slow to be considered complimentary to the Normal and High Profile.
They're best positioned to answer why they didn't want to include slower presets.
Yes they are, but I guess it’s fair enough that the question “what would it be if they had?” is being universally ignored seeing as I obviously went about it in completely the wrong way.
I never use inbuilt presets in programs anyway
That’s your prerogative, I rarely use all the settings in a preset myself, but I like the fact that people who know a lot about a program have taken time to provide what they consider to be good settings. At the very least it provides a good starting point. It also saves the less informed from spending a lot of time learning about all the settings. The people best equipped to make these standardised profiles and ‘best’ settings decisions are generally the people who built the program.

Cheers

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Sun Jan 17, 2010 9:58 pm

So I thought I should mention that I’ve updated the description of my suggested profile with the following because some people seem to have the wrong idea about this thread and my intentions:
this profile is probably too slow for most people and can in no way be considered 'Ultimate'
Either way the following question remains mostly (or more accurately, completely) unanswered:

What settings would compliment the Normal and High Profile to produce a higher than High Profile for general use that would satisfy people with some disk space and processing time to spare?

This is still up for discussion. Although, I have a sneaking suspicion that it never will be discussed in this thread. Probably my fault; newbie, posts are too long, I wasn't concise, I didn't make my point well enough, it's not the right forum etc etc.

Cheers

Statick
Posts: 19
Joined: Sat Dec 27, 2008 7:42 pm

Re: Ultimate Profile?

Post by Statick » Mon Jan 18, 2010 1:12 am

i'd just like to add that i've found this thread particularly useful in putting together a new "better-than-high" preset for myself. as my HTPC is pretty slow (Athlon 3000), i average around 3.5fps on a HQ encode, so testing settings is a slow and laborious process. some of the discussion here has helped me reach settings i'm happy with, and reach a better understanding of what certain settings do, far more quickly than i've previously managed with google and forum searches. i don't understand why the OP is getting flak for this thread at all, maybe i'm missing something.

cheers

creamyhorror
Regular User
Posts: 134
Joined: Mon Jan 04, 2010 2:00 pm

Re: Ultimate Profile?

Post by creamyhorror » Mon Jan 18, 2010 1:55 am

please_help wrote: What settings would compliment the Normal and High Profile to produce a higher than High Profile for general use that would satisfy people with some disk space and processing time to spare?
Off the top of my head your proposal looks fine, except that I'd add bframes=5 or 6 for more efficient compression and deblock -2,-2 or -3,-3 for greater detail/sharpness at the risk of block artifacts. Then it's about the same as what I use.

please_help
Posts: 29
Joined: Sun Jan 10, 2010 6:33 pm

Re: Ultimate Profile?

Post by please_help » Mon Jan 18, 2010 3:44 am

Thanks for the sentiment Statick, I believe it was my bad choice of words and overly verbose posts. But I’m glad you found this thread useful rather than just annoying.

Thanks for getting involved creamyhorror and thanks for sharing your preferences. I wasn’t sure whether the hit in speed by increasing bframes was actually worth it for general use? After looking at my logs it seems like 3 consecutive bframes are used fairly rarely, generally less than 10% of the time for the random selection of movies I tested and pretty close to the percentage of time that none are used at all.

With regards to deblocking, I read that deblocking decreases with higher bit rate so if one were to go for crf 18 then deblocking at 0,0 would be similar to deblocking -1,-1 (or something) at crf 20? I guess this is down to personal taste, but I’d rather not risk any blocking artifacts and if I stick to my reason for starting this topic and assume that the HB developers got it right, I would (tentatively) suggest 0,0 for a general use higher-than-high profile?

At the risk of putting my neck on the line again and writing too much in this post (as usual) I would also suggest that the answer to my complimentary higher-than-high profile question is probably to trust the x264 developers and just go for the x264 Slow preset (which would be incrementally slower than the High Profile) with something like crf 18?

Post Reply