Use Turbo mode?

[ZZZERO]

What are the opinions as to how much improvement in quality is gained by turning turbo off in some of the presets (say unrestricted 2 pass HQ or unrestricted 2 pass EQ modes). Is it negligible or are there many of you that think turbo mode hurts quality enough that you don't use it?

[Sagekilla]

I'm sorry but what are you talking about? Is this a specific GUI you're speaking of?

[Audionut]

how much improvement in quality is gained by turning turbo off



Not much. The speed increase is usually worth the slight quality loss.

edit: although with --b-adapt 2!!!!

[Sharktooth]

I'm sorry but what are you talking about? Is this a specific GUI you're speaking of?
i think he's talking about megui

[EddietheEagle]

It's negligible. I don't worry about it.

[ZZZERO]

Yes. Megui. You say use --b-adapt 2!!!! Is that the "Adapt. B-Frames setting" and set it to "2 - Slow"? The presets I looked at have it set to "1 - Fast" with turbo mode on.

[Audionut]

You say use --b-adapt 2!!!! Is that the "Adapt. B-Frames setting" and set it to "2 - Slow"?

It's totally up to you if you want to use it. And if so, :search:

And yes, that is Adapt. B-Frames setting.


edit: with b-adapt 1, the speed gain is worth setting turbo mode.
With b-adapt 2, that could be different depending on source.

See here.
http://forum.doom9.org/showthread.php?p=1184974#post1184974

Then goto the start of that thread and read.

[Quark.Fusion]

AFAIK b-adapt only matters on first pass, so even in "turbo" mode you can use it if you want it in second pass. Specify --b-frames 3 to speed up it. (Or ever 2, depending on source)

[kemuri-_9]

AFAIK b-adapt only matters on first pass, so even in "turbo" mode you can use it if you want it in second pass. Specify --b-frames 3 to speed up it. (Or ever 2, depending on source)

all frame decisions are only calculated on the first pass.
b-adapt is ignored on the 2nd pass.

any change in settings from 1st pass to 2nd pass that would normally affect frame decisions on a 1st pass will not change the decisions at all.
it uses the 1st pass .stats and does not change it.
(why .stat hacking has worked for ages for specifying frame types on 2nd passes)

so 'turbo'ing on the first pass will make the first pass go faster, but you get worse decisions to use in the 2nd pass.
it's up to you if you desire this quality/time trade-off.

[Ranguvar]

However, the changes are very, very minimal. Even when I use extreme (--me umh (maybe higher) --merange 32 --subme 7 --trellis 2 --no-fast-pskip, etc.) settings, I usually do "turbo" (fast first pass). If you have a very slow AVS script, though, and you don't want to do a lossless pass, then it may be worth it to do a higher quality first pass, since the encoding is mostly limited by the AVS.

[kemuri-_9]

OK, so what are these 'turbo' settings that MeGUI has so i can compare them to the slower settings i use.

that is, all the settings... the default x264 ones, the default megui ones, and then the settings that the turbo mode specifies.

[Ranguvar]

Not sure what MeGUI's turbo is, but this is what I use:

x264.exe --bframes 3 --b-adapt 2 --b-pyramid --deblock -1:-1 --bitrate 6500 --pass 1 --stats "C:\Example.stats" --partitions none --direct auto --weightb --me dia --subme 2 --output NUL --level 3.2 --progress --threads auto "C:\Example.avs"

So basically, deblock, bframes, weightb, b-adapt, b-pyramid, and direct (and AQ) matter. Everything else can be slaughtered. A possible idea for a test, also, is doing the first pass with worse AVS filters. This is more touchy, though. For example, with degraining, do the first pass and try to get the same amount of grain lost, just use fast filters even if they kill a good amount of detail.

[kemuri-_9]

this whole thread is based on megui turbo as far as i can tell, so the settings from that are necessary.

[Quark.Fusion]

In my x264 lossless test --subme 3 was same speed as --subme 2 and in recent builds were some speed optimisations to --subme 1. Also it seems that --b-adapt 2 depends on motion estimation settings.

[cyberbeing]

MeGUI's 1st pass turbo mode is something like this but largely depends on what options are chosen (I haven't listed all options and possibilities):

--pass 1 --bitrate XXXX --bframes XX --b-pyramid (if set for pass 2) --b-adapt 1|2 --direct auto --filter X:X (set the same as pass 2) --subme 1 --partitions none --me dia --threads auto --thread-input --progress --no-dct-decimate (if set for pass 2)

Kemuri, why not just download a build of MeGUI that you can fool around with? It shows you the x264 cli it uses when you change settings.

[Inventive Software]

Wrong forum, for a start! MPEG-4 Encoder GUIs. ;)

@mod: please move this. :)

[ZZZERO]

Really? I thought was asking an encoding question about the turbo settings for a quick first pass and how that affects the quality of the final encode, and that's why I put it in this forum.

[fields_g]

Wrong forum, for a start! MPEG-4 Encoder GUIs. ;)

@mod: please move this. :)

"Turbo Mode" may be a checkbox in a GUI, but the formation and usage discussions belongs here. People who do 2-pass by command line or batch file are benefited by this discussion. The only benefits for a GUI would be developers who want to tweak their turbo option or people who want to try to convince them to do so.

[ZZZERO]

OK, so what are these 'turbo' settings that MeGUI has so i can compare them to the slower settings i use.

that is, all the settings... the default x264 ones, the default megui ones, and then the settings that the turbo mode specifies.

I'd be happy to provide you with those details if I could figure out how to get them. Usually when you change a setting the command line changes in the interface, but when I click or unclick turbo I don't see where it changes so I don't know how to post the difference between turbo on and turbo off here so that I can get some opinions as to how that would affect the final encode. How the command line changes is probably a question I could post in the GUI forum.

[mitsubishi]

If you want to see what meGUI has done, look in the logs. Otherwise you can browse the source to see what it is doing, I think this is the section:

533 if (Turbo)
534 {
535 NbRefFrames = 1;
536 SubPelRefinement = 0;
537 METype = 0; // diamond search
538 I4x4mv = false;
539 P4x4mv = false;
540 I8x8mv = false;
541 P8x8mv = false;
542 B8x8mv = false;
543 AdaptiveDCT = false;
544 MixedRefs = false;
545 BRDO = false;
546 Trellis = false;
547 noFastPSkip = false;
548 WeightedBPrediction = false;
549 biME = false;
550 }
http://megui.svn.sourceforge.net/viewvc/megui/megui/trunk/packages/video/x264/x264Settings.cs?revision=720&view=markup

[kemuri-_9]

hmm... --ref 1 --subme 0 --me dia --analyse none
the rest are x264 defaults (if my head is on straight today)

so the interesting part is the subme 0...
since x264 minimum subme is 1, this actually becomes --subme 1

and we want to try these with b-adapt 2 or not?

[Comatose]

Turbo off (http://img1.hugeup.com/f/09182008/7d2bb8.png) vs Turbo on (http://img1.hugeup.com/f/09182008/298bb.png)

The rest of the settings:
x264 - core 61 r948M 9d4d168 - H.264/MPEG-4 AVC codec - Copyleft 2003-2008 - http://www.videolan.org/x264.html - options: cabac=1 ref=16 deblock=1:0:0 analyse=0x3:0x113 me=umh subme=7 psy_rd=1.0:1.0 brdo=1 mixed_ref=1 me_range=24 chroma_me=1 trellis=2 8x8dct=1 cqm=0 deadzone=21,11 chroma_qp_offset=-4 threads=3 nr=0 decimate=1 mbaff=0 bframes=5 b_pyramid=1 b_adapt=2 b_bias=0 direct=3 wpredb=1 bime=1 keyint=240 keyint_min=24 scenecut=40(pre) rc=2pass bitrate=904 ratetol=3.0 rceq='blurCplx^(1-qComp)' qcomp=1.00 qpmin=10 qpmax=51 qpstep=4 cplxblur=20.0 qblur=0.5 ip_ratio=1.40 pb_ratio=1.30 aq=2:1.00

[Audionut]

One with b-adapt 1 and another with b-adapt 2.

Most realize the speed increase with turbo settings for first pass is worth the slight quality loss.

However b-adapt 2 could have changed that.

Think of it as an extension of the tests you have done in the b-frames thread. Now we find how turbo first pass setting changes things.

[kemuri-_9]

x264 - core 61 r948M 9d4d168 - H.264/MPEG-4 AVC codec - Copyleft 2003-2008 - http://www.videolan.org/x264.html - options: cabac=1 ref=16 deblock=1:0:0 analyse=0x3:0x113 me=umh subme=7 psy_rd=1.0:1.0 brdo=1 mixed_ref=1 me_range=24 chroma_me=1 trellis=2 8x8dct=1 cqm=0 deadzone=21,11 chroma_qp_offset=-4 threads=3 nr=0 decimate=1 mbaff=0 bframes=5 b_pyramid=1 b_adapt=2 b_bias=0 direct=3 wpredb=1 bime=1 keyint=240 keyint_min=24 scenecut=40(pre) rc=2pass bitrate=904 ratetol=3.0 rceq='blurCplx^(1-qComp)' qcomp=1.00 qpmin=10 qpmax=51 qpstep=4 cplxblur=20.0 qblur=0.5 ip_ratio=1.40 pb_ratio=1.30 aq=2:1.00

A. b-adapt 2 has ABSOLUTELY NO effect on second passes

B. most of these settings are severely brought down by the turbo.
*points to wpredb off, trellis off, 1 ref, no analyse, dia me, 1 subme, and no mixed-ref of turbo*

C. you're using an older build, update!

D. what about the first pass where the frame decisions are made? did you turbo that too?

[Comatose]

A. I know, I was just saying it was used with both 1st passes.

B. Yes, but I was trying to show what the effect of turbo is, and thus, those settings/features lowered or turned off.

C. It's an old encode, which is why I took the args from the userdata.

D. They were both full 2pass encodes, independent of each other.

[kemuri-_9]

we're mostly looking for whether turbo is worth the time saving compared to the frame decisions on the first pass.

so from my understanding, most don't turbo 2nd passes....
or am i getting confused by you habitual turbo users? :confused:

[Comatose]

Well, I posted comparison shots in an encode where turbo was used on the first pass, versus one where it wasn't.

[Audionut]

or am i getting confused by you habitual turbo users? :confused:

No, turbo on first pass = more speed - slight quality loss

Turbo both passes = shitty encoder settings. ;)

[kemuri-_9]

time to make sure i got the turbo settings right before i go on a encoding spree

original (same settings from b-adapt 2 thread)

x264_r977 --pass 1 --bitrate 784 --stats x264.stats --min-keyint 1 --keyint 500 --scenecut 50 --ref 16 --mixed-refs
--deblock 0:0 --no-fast-pskip --b-adapt _x_ --bframes _y_ --b-bias 0 --b-pyramid --b-rdo --bime --weightb
--direct auto --subme 7 --direct-8x8 -1 --trellis 2 --analyse all --8x8dct --qpmin 1 --qpmax 51 --qpstep 60
--pbratio 1.4 --ratetol 100.0 --qcomp 0.6 --me umh --merange 32 --psy-rd 0.0:0.0 --thread-input --threads 4
--progress --no-dct-decimate --level 5.1 --log-file x264.1pass.log --output NUL open_lossless.avs


turbo

x264_r977 --pass 1 --bitrate 784 --stats x264.stats --min-keyint 1 --keyint 500 --scenecut 50 --ref 1
--deblock 0:0 --b-adapt _x_ --bframes _y_ --b-bias 0 --b-pyramid --direct auto --subme 1 --direct-8x8 -1
--trellis 0 --analyse none --qpmin 1 --qpmax 51 --qpstep 60 --pbratio 1.4 --ratetol 100.0 --qcomp 0.6
--me dia --merange 32 --psy-rd 0.0:0.0 --thread-input --threads 4 --progress --no-dct-decimate --level 5.1
--log-file x264.1pass.log --output NUL open_lossless.avs


turbo changes

ref: 16 -> 1
subme: 7 -> 1 (0 -> 1 in x264)
me: umh -> dia
analyse: all -> none
8x8dct -> gone
mixed-refs -> gone
brdo -> gone
trellis: 2 -> 0
no-fast-pskip -> gone
weightb -> gone
bime -> gone


did i miss anything?

just thinking of doing 2,4,6, and 16 bframe comparisons this time for b-adapt 1 & 2.

[Ranguvar]

Sounds good :) Though IMO --subme 2 or 3 is preferred as there's barely any performance hit (at that speed, speed is usually bottlenecked by something else). But I understand you're trying to test MeGUI's turbo specifically.

Looking forward to results!

[Sharktooth]

im about to change subme to 2 with turbo (n megui)... so please test both subme 1 and subme 2 for turbo first pass.
concerning subme 0, it doesnt exist. megui always used subme 1 for turbo... (SubPelRefinement = 0 -> --subme 1 as you can see in x264Encoder.cs: int subq = xs.SubPelRefinement + 1)

[kemuri-_9]

links for 'hard cpu' 1st pass results
2 & 4 bframes: http://forum.doom9.org/showthread.php?p=1185229#post1185229
6 bframes: http://forum.doom9.org/showthread.php?p=1185387#post1185387
16 bframes: http://forum.doom9.org/showthread.php?p=1184868#post1184868

additional implied turbo alteration
merange: 32 -> 16 (x264: me dia has max 16 merange)
turbo results (subme 1):
b-adapt 1, 2 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.45 size: 24844 PSNR Mean Y:50.43 U:52.66 V:52.78 Avg:50.83 Global:48.96
x264 [info]: slice P:25999 Avg QP:16.45 size: 4239 PSNR Mean Y:47.18 U:50.04 V:50.29 Avg:47.90 Global:47.02
x264 [info]: slice B:3403 Avg QP:22.74 size: 644 PSNR Mean Y:46.87 U:50.81 V:51.27 Avg:47.76 Global:44.60
x264 [info]: consecutive B-frames: 81.6% 4.1% 14.3%
x264 [info]: mb I I16..4: 43.7% 0.0% 56.3%
x264 [info]: mb P I16..4: 10.9% 0.0% 0.0% P16..4: 52.1% 0.0% 0.0% 0.0% 0.0% skip:36.9%
x264 [info]: mb B I16..4: 0.7% 0.0% 0.0% B16..8: 14.2% 0.0% 0.0% direct: 6.0% skip:79.1% L0:29.9% L1:65.6% BI: 4.4%
x264 [info]: final ratefactor: 26.96
x264 [info]: direct mvs spatial:99.0% temporal:1.0%
x264 [info]: SSIM Mean Y:0.9882095
x264 [info]: PSNR Mean Y:47.183 U:50.159 V:50.427 Avg:47.918 Global:46.679 kb/s:777.86
x264 [info]: encoded 29731 frames, 58.47 fps, 777.86 kb/s


b-adapt 2, 2 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.62 size: 24490 PSNR Mean Y:50.37 U:52.62 V:52.76 Avg:50.79 Global:48.87
x264 [info]: slice P:24495 Avg QP:16.14 size: 4460 PSNR Mean Y:47.67 U:50.43 V:50.68 Avg:48.36 Global:47.18
x264 [info]: slice B:4907 Avg QP:22.37 size: 791 PSNR Mean Y:44.77 U:48.87 V:49.18 Avg:45.72 Global:44.81
x264 [info]: consecutive B-frames: 71.5% 14.0% 14.6%
x264 [info]: mb I I16..4: 44.8% 0.0% 55.2%
x264 [info]: mb P I16..4: 10.5% 0.0% 0.0% P16..4: 53.7% 0.0% 0.0% 0.0% 0.0% skip:35.8%
x264 [info]: mb B I16..4: 2.2% 0.0% 0.0% B16..8: 13.9% 0.0% 0.0% direct: 7.1% skip:76.8% L0:34.5% L1:60.8% BI: 4.7%
x264 [info]: final ratefactor: 26.77
x264 [info]: direct mvs spatial:98.6% temporal:1.4%
x264 [info]: SSIM Mean Y:0.9881934
x264 [info]: PSNR Mean Y:47.219 U:50.195 V:50.458 Avg:47.954 Global:46.701 kb/s:781.81
x264 [info]: encoded 29731 frames, 56.77 fps, 781.81 kb/s


b-adapt 1, 4 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.53 size: 24658 PSNR Mean Y:50.38 U:52.61 V:52.76 Avg:50.78 Global:48.88
x264 [info]: slice P:25405 Avg QP:16.36 size: 4316 PSNR Mean Y:47.17 U:50.01 V:50.26 Avg:47.89 Global:47.08
x264 [info]: slice B:3997 Avg QP:23.03 size: 672 PSNR Mean Y:46.86 U:50.85 V:51.35 Avg:47.74 Global:44.49
x264 [info]: consecutive B-frames: 81.0% 4.0% 2.0% 2.0% 10.9%
x264 [info]: mb I I16..4: 44.1% 0.0% 55.9%
x264 [info]: mb P I16..4: 11.1% 0.0% 0.0% P16..4: 52.6% 0.0% 0.0% 0.0% 0.0% skip:36.3%
x264 [info]: mb B I16..4: 0.9% 0.0% 0.0% B16..8: 14.1% 0.0% 0.0% direct: 5.7% skip:79.3% L0:30.4% L1:64.8% BI: 4.8%
x264 [info]: final ratefactor: 26.92
x264 [info]: direct mvs spatial:99.1% temporal:0.9%
x264 [info]: SSIM Mean Y:0.9881600
x264 [info]: PSNR Mean Y:47.164 U:50.154 V:50.430 Avg:47.902 Global:46.641 kb/s:777.07
x264 [info]: encoded 29731 frames, 58.59 fps, 777.07 kb/s


b-adapt 2, 4 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.63 size: 24521 PSNR Mean Y:50.37 U:52.62 V:52.76 Avg:50.79 Global:48.85
x264 [info]: slice P:24227 Avg QP:16.07 size: 4509 PSNR Mean Y:47.71 U:50.46 V:50.71 Avg:48.40 Global:47.22
x264 [info]: slice B:5175 Avg QP:22.45 size: 768 PSNR Mean Y:44.70 U:48.82 V:49.14 Avg:45.65 Global:44.74
x264 [info]: consecutive B-frames: 71.1% 13.0% 10.9% 3.2% 1.7%
x264 [info]: mb I I16..4: 44.8% 0.0% 55.2%
x264 [info]: mb P I16..4: 10.5% 0.0% 0.0% P16..4: 54.0% 0.0% 0.0% 0.0% 0.0% skip:35.5%
x264 [info]: mb B I16..4: 2.2% 0.0% 0.0% B16..8: 14.0% 0.0% 0.0% direct: 7.0% skip:76.9% L0:35.1% L1:60.2% BI: 4.7%
x264 [info]: final ratefactor: 26.75
x264 [info]: direct mvs spatial:98.5% temporal:1.5%
x264 [info]: SSIM Mean Y:0.9881732
x264 [info]: PSNR Mean Y:47.212 U:50.198 V:50.456 Avg:47.949 Global:46.684 kb/s:782.39
x264 [info]: encoded 29731 frames, 46.20 fps, 782.39 kb/s


b-adapt 1, 6 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.55 size: 24632 PSNR Mean Y:50.38 U:52.60 V:52.75 Avg:50.78 Global:48.86
x264 [info]: slice P:25194 Avg QP:16.34 size: 4347 PSNR Mean Y:47.15 U:49.99 V:50.23 Avg:47.87 Global:47.09
x264 [info]: slice B:4208 Avg QP:23.13 size: 667 PSNR Mean Y:46.87 U:50.86 V:51.37 Avg:47.74 Global:44.46
x264 [info]: consecutive B-frames: 80.6% 4.1% 2.0% 1.8% 1.6% 6.8% 3.1%
x264 [info]: mb I I16..4: 44.2% 0.0% 55.8%
x264 [info]: mb P I16..4: 11.1% 0.0% 0.0% P16..4: 52.8% 0.0% 0.0% 0.0% 0.0% skip:36.1%
x264 [info]: mb B I16..4: 0.9% 0.0% 0.0% B16..8: 14.0% 0.0% 0.0% direct: 5.6% skip:79.6% L0:31.9% L1:63.2% BI: 4.9%
x264 [info]: final ratefactor: 26.90
x264 [info]: direct mvs spatial:99.4% temporal:0.6%
x264 [info]: SSIM Mean Y:0.9881312
x264 [info]: PSNR Mean Y:47.148 U:50.140 V:50.416 Avg:47.883 Global:46.619 kb/s:776.87
x264 [info]: encoded 29731 frames, 58.88 fps, 776.87 kb/s


b-adapt 2, 6 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.64 size: 24518 PSNR Mean Y:50.37 U:52.62 V:52.76 Avg:50.78 Global:48.84
x264 [info]: slice P:24182 Avg QP:16.05 size: 4514 PSNR Mean Y:47.72 U:50.46 V:50.71 Avg:48.41 Global:47.23
x264 [info]: slice B:5220 Avg QP:22.51 size: 777 PSNR Mean Y:44.67 U:48.81 V:49.13 Avg:45.63 Global:44.70
x264 [info]: consecutive B-frames: 71.0% 13.1% 10.6% 3.1% 1.3% 0.4% 0.5%
x264 [info]: mb I I16..4: 44.9% 0.0% 55.1%
x264 [info]: mb P I16..4: 10.5% 0.0% 0.0% P16..4: 54.0% 0.0% 0.0% 0.0% 0.0% skip:35.4%
x264 [info]: mb B I16..4: 2.2% 0.0% 0.0% B16..8: 14.0% 0.0% 0.0% direct: 6.9% skip:76.8% L0:35.2% L1:60.1% BI: 4.7%
x264 [info]: final ratefactor: 26.74
x264 [info]: direct mvs spatial:98.7% temporal:1.3%
x264 [info]: SSIM Mean Y:0.9881640
x264 [info]: PSNR Mean Y:47.210 U:50.196 V:50.457 Avg:47.947 Global:46.677 kb/s:782.36
x264 [info]: encoded 29731 frames, 32.58 fps, 782.36 kb/s


b-adapt 1, 16 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.56 size: 24641 PSNR Mean Y:50.37 U:52.60 V:52.74 Avg:50.77 Global:48.86
x264 [info]: slice P:25133 Avg QP:16.35 size: 4357 PSNR Mean Y:47.09 U:49.93 V:50.17 Avg:47.81 Global:47.08
x264 [info]: slice B:4269 Avg QP:22.97 size: 670 PSNR Mean Y:47.16 U:50.93 V:51.46 Avg:47.89 Global:44.51
x264 [info]: consecutive B-frames: 80.5% 4.0% 2.0% 1.8% 1.6% 6.8% 2.1% 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% 0.1% 0.1% 0.2% 0.7%
x264 [info]: mb I I16..4: 44.2% 0.0% 55.8%
x264 [info]: mb P I16..4: 11.1% 0.0% 0.0% P16..4: 53.0% 0.0% 0.0% 0.0% 0.0% skip:36.0%
x264 [info]: mb B I16..4: 0.9% 0.0% 0.0% B16..8: 13.9% 0.0% 0.0% direct: 5.9% skip:79.3% L0:32.1% L1:63.0% BI: 4.9%
x264 [info]: final ratefactor: 26.90
x264 [info]: direct mvs spatial:99.2% temporal:0.8%
x264 [info]: SSIM Mean Y:0.9881331
x264 [info]: PSNR Mean Y:47.134 U:50.105 V:50.381 Avg:47.852 Global:46.620 kb/s:777.29
x264 [info]: encoded 29731 frames, 53.14 fps, 777.29 kb/s


b-adapt 2, 16 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:329 Avg QP:14.64 size: 24529 PSNR Mean Y:50.37 U:52.62 V:52.76 Avg:50.78 Global:48.83
x264 [info]: slice P:24114 Avg QP:16.03 size: 4523 PSNR Mean Y:47.73 U:50.47 V:50.72 Avg:48.42 Global:47.24
x264 [info]: slice B:5288 Avg QP:22.54 size: 782 PSNR Mean Y:44.64 U:48.80 V:49.13 Avg:45.60 Global:44.68

x264 [info]: consecutive B-frames: 71.0% 13.0% 10.4% 2.7% 0.7% 0.4% 0.2% 0.2% 0.4% 0.2% 0.3% 0.0% 0.3% 0.0% 0.0% 0.1% 0.1%
x264 [info]: mb I I16..4: 44.9% 0.0% 55.1%
x264 [info]: mb P I16..4: 10.5% 0.0% 0.0% P16..4: 54.1% 0.0% 0.0% 0.0% 0.0% skip:35.4%
x264 [info]: mb B I16..4: 2.2% 0.0% 0.0% B16..8: 14.1% 0.0% 0.0% direct: 6.9% skip:76.8% L0:35.3% L1:59.9% BI: 4.8%
x264 [info]: final ratefactor: 26.74
x264 [info]: direct mvs spatial:98.6% temporal:1.4%
x264 [info]: SSIM Mean Y:0.9881639
x264 [info]: PSNR Mean Y:47.210 U:50.199 V:50.457 Avg:47.946 Global:46.675 kb/s:782.44
x264 [info]: encoded 29731 frames, 9.25 fps, 782.44 kb/s


im going to change subme to 2 with turbo... so please test both subme 1 and subme 2 for turbo first pass.
results coming soon...

[kemuri-_9]

turbo results (subme 2):
b-adapt 1, 2 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:353 Avg QP:14.10 size: 24451 PSNR Mean Y:50.69 U:52.90 V:53.07 Avg:51.11 Global:49.26
x264 [info]: slice P:23554 Avg QP:16.05 size: 4550 PSNR Mean Y:47.51 U:50.19 V:50.45 Avg:48.20 Global:47.38
x264 [info]: slice B:5824 Avg QP:20.25 size: 727 PSNR Mean Y:46.86 U:50.55 V:50.89 Avg:47.73 Global:45.80
x264 [info]: consecutive B-frames: 67.7% 10.2% 22.1%
x264 [info]: mb I I16..4: 46.8% 0.0% 53.2%
x264 [info]: mb P I16..4: 15.0% 0.0% 0.0% P16..4: 51.2% 0.0% 0.0% 0.0% 0.0% skip:33.8%
x264 [info]: mb B I16..4: 0.8% 0.0% 0.0% B16..8: 15.1% 0.0% 0.0% direct: 8.5% skip:75.6% L0:28.7% L1:66.0% BI: 5.3%
x264 [info]: final ratefactor: 23.16
x264 [info]: direct mvs spatial:98.2% temporal:1.8%
x264 [info]: SSIM Mean Y:0.9891060
x264 [info]: PSNR Mean Y:47.422 U:50.296 V:50.568 Avg:48.143 Global:47.034 kb/s:774.44
x264 [info]: encoded 29731 frames, 58.44 fps, 774.44 kb/s


b-adapt 2, 2 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:353 Avg QP:14.20 size: 24140 PSNR Mean Y:50.65 U:52.87 V:53.04 Avg:51.07 Global:49.20
x264 [info]: slice P:20520 Avg QP:15.68 size: 5067 PSNR Mean Y:48.08 U:50.68 V:50.94 Avg:48.74 Global:47.54
x264 [info]: slice B:8858 Avg QP:19.39 size: 931 PSNR Mean Y:46.02 U:49.55 V:49.79 Avg:46.88 Global:46.19
x264 [info]: consecutive B-frames: 50.2% 18.3% 31.5%
x264 [info]: mb I I16..4: 47.2% 0.0% 52.8%
x264 [info]: mb P I16..4: 15.2% 0.0% 0.0% P16..4: 53.9% 0.0% 0.0% 0.0% 0.0% skip:30.9%
x264 [info]: mb B I16..4: 2.2% 0.0% 0.0% B16..8: 15.6% 0.0% 0.0% direct:11.0% skip:71.3% L0:33.4% L1:61.5% BI: 5.0%
x264 [info]: final ratefactor: 22.69
x264 [info]: direct mvs spatial:98.8% temporal:1.2%
x264 [info]: SSIM Mean Y:0.9891799
x264 [info]: PSNR Mean Y:47.495 U:50.370 V:50.625 Avg:48.216 Global:47.109 kb/s:778.92
x264 [info]: encoded 29731 frames, 51.70 fps, 778.92 kb/s


b-adapt 1, 4 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:353 Avg QP:14.19 size: 24170 PSNR Mean Y:50.62 U:52.86 V:53.02 Avg:51.05 Global:49.19
x264 [info]: slice P:22554 Avg QP:15.92 size: 4708 PSNR Mean Y:47.53 U:50.18 V:50.43 Avg:48.21 Global:47.45
x264 [info]: slice B:6824 Avg QP:20.32 size: 753 PSNR Mean Y:46.85 U:50.58 V:50.92 Avg:47.72 Global:45.77
x264 [info]: consecutive B-frames: 66.0% 10.0% 5.2% 5.4% 13.4%
x264 [info]: mb I I16..4: 47.2% 0.0% 52.8%
x264 [info]: mb P I16..4: 15.4% 0.0% 0.0% P16..4: 52.0% 0.0% 0.0% 0.0% 0.0% skip:32.5%
x264 [info]: mb B I16..4: 0.9% 0.0% 0.0% B16..8: 14.6% 0.0% 0.0% direct: 8.1% skip:76.3% L0:30.3% L1:64.0% BI: 5.7%
x264 [info]: final ratefactor: 23.07
x264 [info]: direct mvs spatial:98.6% temporal:1.4%
x264 [info]: SSIM Mean Y:0.9890797
x264 [info]: PSNR Mean Y:47.412 U:50.302 V:50.570 Avg:48.133 Global:47.021 kb/s:773.19
x264 [info]: encoded 29731 frames, 57.97 fps, 773.19 kb/s


b-adapt 2, 4 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:353 Avg QP:14.23 size: 24078 PSNR Mean Y:50.64 U:52.87 V:53.04 Avg:51.07 Global:49.18
x264 [info]: slice P:20170 Avg QP:15.58 size: 5122 PSNR Mean Y:48.13 U:50.71 V:50.96 Avg:48.79 Global:47.59
x264 [info]: slice B:9208 Avg QP:19.46 size: 950 PSNR Mean Y:45.97 U:49.51 V:49.76 Avg:46.83 Global:46.13
x264 [info]: consecutive B-frames: 49.8% 17.1% 25.8% 4.7% 2.6%
x264 [info]: mb I I16..4: 47.5% 0.0% 52.5%
x264 [info]: mb P I16..4: 15.2% 0.0% 0.0% P16..4: 54.1% 0.0% 0.0% 0.0% 0.0% skip:30.7%
x264 [info]: mb B I16..4: 2.4% 0.0% 0.0% B16..8: 15.9% 0.0% 0.0% direct:10.9% skip:70.8% L0:33.9% L1:60.9% BI: 5.1%
x264 [info]: final ratefactor: 22.64
x264 [info]: direct mvs spatial:98.9% temporal:1.1%
x264 [info]: SSIM Mean Y:0.9891569
x264 [info]: PSNR Mean Y:47.493 U:50.365 V:50.616 Avg:48.208 Global:47.098 kb/s:777.72
x264 [info]: encoded 29731 frames, 36.55 fps, 777.72 kb/s


b-adapt 1, 6 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:353 Avg QP:14.19 size: 24174 PSNR Mean Y:50.62 U:52.87 V:53.03 Avg:51.05 Global:49.19
x264 [info]: slice P:22411 Avg QP:15.92 size: 4735 PSNR Mean Y:47.50 U:50.15 V:50.39 Avg:48.18 Global:47.45
x264 [info]: slice B:6967 Avg QP:20.29 size: 752 PSNR Mean Y:46.95 U:50.67 V:51.00 Avg:47.81 Global:45.80
x264 [info]: consecutive B-frames: 65.6% 10.0% 5.1% 5.3% 9.0% 3.9% 1.0%
x264 [info]: mb I I16..4: 47.2% 0.0% 52.8%
x264 [info]: mb P I16..4: 15.5% 0.0% 0.0% P16..4: 52.2% 0.0% 0.0% 0.0% 0.0% skip:32.3%
x264 [info]: mb B I16..4: 0.9% 0.0% 0.0% B16..8: 14.6% 0.0% 0.0% direct: 8.1% skip:76.4% L0:30.7% L1:63.5% BI: 5.7%
x264 [info]: final ratefactor: 23.06
x264 [info]: direct mvs spatial:98.6% temporal:1.4%
x264 [info]: SSIM Mean Y:0.9890773
x264 [info]: PSNR Mean Y:47.409 U:50.301 V:50.568 Avg:48.130 Global:47.019 kb/s:773.51
x264 [info]: encoded 29731 frames, 58.34 fps, 773.51 kb/s


b-adapt 2, 6 bframes:

x264 [info]: slice I:353 Avg QP:14.23 size: 24093 PSNR Mean Y:50.64 U:52.87 V:53.04 Avg:51.07 Global:49.17
x264 [info]: slice P:20096 Avg QP:15.56 size: 5131 PSNR Mean Y:48.15 U:50.72 V:50.97 Avg:48.80 Global:47.60
x264 [info]: slice B:9282 Avg QP:19.51 size: 961 PSNR Mean Y:45.94 U:49.49 V:49.75 Avg:46.80 Global:46.09
x264 [info]: consecutive B-frames: 49.8% 17.1% 25.3% 4.1% 1.6% 1.0% 1.1%
x264 [info]: mb I I16..4: 47.5% 0.0% 52.5%
x264 [info]: mb P I16..4: 15.2% 0.0% 0.0% P16..4: 54.2% 0.0% 0.0% 0.0% 0.0% skip:30.6%
x264 [info]: mb B I16..4: 2.4% 0.0% 0.0% B16..8: 16.0% 0.0% 0.0% direct:10.9% skip:70.7% L0:34.0% L1:60.8% BI: 5.2%
x264 [info]: final ratefactor: 22.63
x264 [info]: direct mvs spatial:98.8% temporal:1.2%
x264 [info]: SSIM Mean Y:0.9891465
x264 [info]: PSNR Mean Y:47.487 U:50.362 V:50.613 Avg:48.203 Global:47.087 kb/s:777.64
x264 [info]: encoded 29731 frames, 24.78 fps, 777.64 kb/s


b-adapt 1, 16 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:353 Avg QP:14.21 size: 24128 PSNR Mean Y:50.61 U:52.85 V:53.01 Avg:51.04 Global:49.19
x264 [info]: slice P:22374 Avg QP:15.93 size: 4741 PSNR Mean Y:47.44 U:50.09 V:50.34 Avg:48.13 Global:47.45
x264 [info]: slice B:7004 Avg QP:20.30 size: 750 PSNR Mean Y:47.09 U:50.73 V:51.10 Avg:47.89 Global:45.81
x264 [info]: consecutive B-frames: 65.6% 10.0% 5.1% 5.3% 9.0% 3.9% 0.0% 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% 0.1% 0.1% 0.1% 0.6%
x264 [info]: mb I I16..4: 47.2% 0.0% 52.8%
x264 [info]: mb P I16..4: 15.5% 0.0% 0.0% P16..4: 52.3% 0.0% 0.0% 0.0% 0.0% skip:32.2%
x264 [info]: mb B I16..4: 0.9% 0.0% 0.0% B16..8: 14.5% 0.0% 0.0% direct: 8.2% skip:76.4% L0:31.0% L1:63.3% BI: 5.7%
x264 [info]: final ratefactor: 23.06
x264 [info]: direct mvs spatial:98.6% temporal:1.4%
x264 [info]: SSIM Mean Y:0.9890744
x264 [info]: PSNR Mean Y:47.395 U:50.273 V:50.549 Avg:48.104 Global:47.017 kb/s:773.16
x264 [info]: encoded 29731 frames, 58.53 fps, 773.16 kb/s


b-adapt 2, 16 bframes:

x264 [info]: using cpu capabilities: MMX2 SSE2Fast
x264 [info]: slice I:353 Avg QP:14.24 size: 24079 PSNR Mean Y:50.64 U:52.87 V:53.04 Avg:51.07 Global:49.16
x264 [info]: slice P:20017 Avg QP:15.53 size: 5140 PSNR Mean Y:48.17 U:50.73 V:50.98 Avg:48.82 Global:47.62
x264 [info]: slice B:9361 Avg QP:19.54 size: 970 PSNR Mean Y:45.91 U:49.47 V:49.73 Avg:46.78 Global:46.06
x264 [info]: consecutive B-frames: 49.7% 17.0% 25.1% 3.8% 1.0% 0.6% 0.5% 0.4% 0.4% 0.4% 0.2% 0.2% 0.1% 0.2% 0.0% 0.1% 0.3%
x264 [info]: mb I I16..4: 47.5% 0.0% 52.5%
x264 [info]: mb P I16..4: 15.2% 0.0% 0.0% P16..4: 54.3% 0.0% 0.0% 0.0% 0.0% skip:30.5%
x264 [info]: mb B I16..4: 2.4% 0.0% 0.0% B16..8: 16.1% 0.0% 0.0% direct:10.9% skip:70.6% L0:34.3% L1:60.6% BI: 5.1%
x264 [info]: final ratefactor: 22.62
x264 [info]: direct mvs spatial:98.9% temporal:1.1%
x264 [info]: SSIM Mean Y:0.9891365
x264 [info]: PSNR Mean Y:47.487 U:50.362 V:50.613 Avg:48.203 Global:47.082 kb/s:777.23
x264 [info]: encoded 29731 frames, 6.78 fps, 777.23 kb/s

[cyberbeing]

One thing that stands out in your above test when using subme 2 in a "turbo" pass is a huge increase (+70-80%) in the use of B-frames and a +7% increase in I-frames (353 vs 329) and a rise in both PSNR and SSIM.

[Sharktooth]

i spotted a visual difference between turbo and non-turbo encoded files. so i did a quick visual test and i saw subme 2 in first pass mitigated that difference a lot, that's why i asked for a complete test and updated megui with subme 2 for "turbo" first pass.

[cyberbeing]

I just never realized that subme affected adaptive b-frame placement and improved detection of scene changes. It makes me wonder how much more b-frames and I-frames would continue to increase when subme 7 is used.

[Dark Shikari]

I just never realized that subme affected adaptive b-frame placement and improved detection of scene changes. It makes me wonder how much more b-frames and I-frames would continue to increase when subme 7 is used.Its actually a not-entirely-intended situation, and only affected by subme 1 vs 2. The reason is because subme1 uses mbcmp=SAD and subme2 and above uses mbcmp=SATD.

[Ranguvar]

Woot! I'm lucky! I've been using subme 2 instead of 1, for the hell of it :p

[Comatose]

How would --ref 1 (which MeGUI uses) affect the first pass, theoretically?

[qyqgpower]

I've tested several cases in order to find out a optimal frame decision setting of 1st pass for --b-adapt 2.

source is the opening of CLANNAD(DVD), lot of small moving objects and some fade scenes.

let's only compare the frame decision and speed, CPU usage is around 50% on my quad core@3GHz.

full quality(ideal frame decision)
--pass 1 --bitrate 2065 --keyint 240 --min-keyint 24 --ref 10 --mixed-refs --no-fast-pskip --bframes 16 --b-adapt 2 --b-pyramid --b-rdo --bime --weightb --direct auto --subme 7 --trellis 2 --psy-rd 0.0:0 --partitions all --8x8dct --me tesa --merange 32 --threads 3 --thread-input --sar 1:1 --progress --no-dct-decimate --output NUL
x264 [info]: slice I:25 Avg QP:13.64 size: 17373 PSNR Mean Y:52.42 U:54.60 V:54.78 Avg:52.75 Global:51.30
x264 [info]: slice P:979 Avg QP:17.47 size: 16141 PSNR Mean Y:45.88 U:47.90 V:48.09 Avg:46.43 Global:44.71
x264 [info]: slice B:1149 Avg QP:18.97 size: 5754 PSNR Mean Y:45.43 U:47.69 V:47.92 Avg:46.03 Global:44.24
encoded 2153 frames, 1.37 fps, 2035.61 kb/s


old turbo subme1
--pass 1 --bitrate 2065 --keyint 240 --min-keyint 24 --bframes 16 --b-adapt 2 --b-pyramid --direct auto --subme 1 --partitions none --me dia --threads 3 --thread-input --progress --no-dct-decimate --output NUL
x264 [info]: slice I:24 Avg QP:15.70 size: 17849 PSNR Mean Y:50.45 U:53.45 V:53.70 Avg:50.99 Global:48.41
x264 [info]: slice P:1081 Avg QP:19.43 size: 15963 PSNR Mean Y:43.73 U:46.68 V:46.86 Avg:44.47 Global:42.30
x264 [info]: slice B:1048 Avg QP:21.30 size: 4855 PSNR Mean Y:43.07 U:46.46 V:46.73 Avg:43.90 Global:41.70
encoded 2153 frames, 10.13 fps, 2028.89 kb/s

new turbo subme2
--pass 1 --bitrate 2065 --keyint 240 --min-keyint 24 --bframes 16 --b-adapt 2 --b-pyramid --direct auto --subme 2 --partitions none --me dia --threads 3 --thread-input --progress --no-dct-decimate --output NUL
x264 [info]: slice I:26 Avg QP:14.79 size: 17604 PSNR Mean Y:51.37 U:54.08 V:54.31 Avg:51.85 Global:49.99
x264 [info]: slice P:1030 Avg QP:19.35 size: 16229 PSNR Mean Y:43.87 U:46.79 V:47.00 Avg:44.61 Global:42.45
x264 [info]: slice B:1097 Avg QP:21.00 size: 5163 PSNR Mean Y:43.38 U:46.62 V:46.89 Avg:44.18 Global:41.99
encoded 2153 frames, 7.96 fps, 2034.74 kb/s

normal subme6+hex
--pass 1 --bitrate 2065 --keyint 240 --min-keyint 24 --bframes 16 --b-adapt 2 --b-pyramid --direct auto --psy-rd 0.0:0 --partitions none --threads 3 --thread-input --progress --no-dct-decimate --output NUL
x264 [info]: slice I:27 Avg QP:14.56 size: 17692 PSNR Mean Y:51.48 U:54.17 V:54.40 Avg:51.97 Global:50.14
x264 [info]: slice P:991 Avg QP:19.16 size: 16147 PSNR Mean Y:44.13 U:47.07 V:47.27 Avg:44.88 Global:42.78
x264 [info]: slice B:1135 Avg QP:20.84 size: 5568 PSNR Mean Y:43.56 U:46.64 V:46.93 Avg:44.33 Global:42.17
encoded 2153 frames, 7.75 fps, 2031.30 kb/s

slower subme6+hex+bframe related options (pretty close now)
--pass 1 --bitrate 2065 --keyint 240 --min-keyint 24 --bframes 16 --b-adapt 2 --b-pyramid --b-rdo --bime --weightb --direct auto --psy-rd 0.0:0 --partitions none --threads 3 --thread-input --progress --no-dct-decimate --output NUL
x264 [info]: slice I:27 Avg QP:14.40 size: 18063 PSNR Mean Y:51.63 U:54.20 V:54.45 Avg:52.11 Global:50.23
x264 [info]: slice P:979 Avg QP:19.16 size: 16824 PSNR Mean Y:44.12 U:46.97 V:47.16 Avg:44.85 Global:42.76
x264 [info]: slice B:1147 Avg QP:20.87 size: 5109 PSNR Mean Y:43.85 U:46.90 V:47.17 Avg:44.62 Global:42.48
encoded 2153 frames, 7.25 fps, 2032.98 kb/s <- speed is 5 times more than full quality

lower ME to dia (bad result)
--pass 1 --bitrate 2065 --keyint 240 --min-keyint 24 --bframes 16 --b-adapt 2 --b-pyramid --b-rdo --bime --weightb --direct auto --psy-rd 0.0:0 --partitions none --me dia --threads 3 --thread-input --sar 1:1 --progress --no-dct-decimate --output NUL
x264 [info]: slice I:26 Avg QP:14.52 size: 17864 PSNR Mean Y:51.61 U:54.17 V:54.41 Avg:52.08 Global:50.16
x264 [info]: slice P:1002 Avg QP:19.22 size: 16749 PSNR Mean Y:44.11 U:46.96 V:47.15 Avg:44.83 Global:42.72
x264 [info]: slice B:1125 Avg QP:20.90 size: 4950 PSNR Mean Y:43.81 U:46.91 V:47.15 Avg:44.59 Global:42.43
encoded 2153 frames, 7.42 fps, 2032.78 kb/s

rise to subme7+umh+no-fast-pskip, no changes in frame decision

best result: subme6+hex+bframe related options for 1st pass.

[shon3i]

Since somebody says that subme 3 have same speed as subme 2, can we use subme 3 as default for turbo on?

[Sharktooth]

why???

@qyqgpower: --b-adapt 2 and --bframes 16 is a no go...
also there's no best as per forum rules... lower the number of b-frames to 3 or 4 and you'll see a HUGE difference in speed...

[ajp_anton]

I don't think "--bframes 16" had to do anything with the point of his post. He was just showing how different 1st pass settings affect the frame decision.

[ZZZERO]

Would b-adapt 2 and --bframes 3 and subme 2 be a good turbo setting? The hit to the speed from b- adapt 2 seems to kick in with high numbers of b frames, but with 3 it might still be fast enough for turbo but with better decisions.

[kemuri-_9]

the bframe count needs to be the same for both passes,
so whatever max bframes you want on the 2nd pass, you also need on the first pass.
like i use 16, so the first pass must be 16 as well.

[qyqgpower]

why???

@qyqgpower: --b-adapt 2 and --bframes 16 is a no go...
also there's no best as per forum rules... lower the number of b-frames to 3 or 4 and you'll see a HUGE difference in speed...
sorry to use the term of "best".
but as ajp_anton has said, my intension was to find out a optimal 1st pass setting for frame decision, so how much --bframes would some one use with --b-adapt 2 is up to him.
But he would get much more accurate frame decision with subme6+hex+bframe related options than only subme2, and without lose too much speed in 1st pass.

[Sharktooth]

nope. the speed is related to the number of b-frames, that's why i told you to lower the number to 3 or 4...
16 b-frames and b-adapt 2 will kill the encoding speed and will produce false results since subme 2 is way much faster than subme 6... try it.

[ajp_anton]

Again, the point wasn't the speed. It was that higher subme, even above 2 -> better frame decision.

[Sagekilla]

Sharktooth: I believe you mean 16 b-frames and b-adapt 2 will kill encoding speed :) I rarely notice any speed difference from 1-16 b's on --badapt 1

[Sharktooth]

yep, it was a typo. fixed.
@ajp_anton: still his test is invalid, since most ppl wont use --bframes 16 along with --b-adapt 2. so in most cases, subme 2 at first pass gives the user a huge speed avantage for a really minor quality loss and in most cases there will be a huge difference in speed from his tests. hence is up to the user to choose what to do. if you want the ubermaximum quality it's obvious you wont use faster first pass settings... for that reason the Insane presets for megui do not have turbo enabled.