x265 HEVC Encoder

[LigH]

Current state:

x265 1.8+167-e951ab673b1c (GCC 4.9.2)
x265 1.8+167-e951ab673b1c (GCC 5.2.0)

New documented 'K' frametype ("keyframe", exact type depending on circumstances), several speedups and fixes (mostly in sao).

[microchip8]

Originally Posted by pingfr

Any news? (on the ctu & presets improvements & QoL)?

Cheers.

Impatient much?

[burfadel]

Originally Posted by pingfr

Any news? (on the ctu & presets improvements & QoL)?

Cheers.

A watched pot never boils

[microchip8]

Originally Posted by pingfr

@froggy1: Retarded much?

This is going to get you nowhere

Please don't ask when it's done. It'll be done when it's done

[Blowis]

I wanted to ask is it normal that me=star is faster than umh.
I tested two videos:

1a) Profil medium me= umh : 18.61 fps PSNR : 45.114 SSIM :0.971590
1b) Profil medium me= star : 19.14 fps PSNR : 45.113 SSIM : 0.971609

2a) Profil medium me= umh : 13.23 fps PSNR : 46.188 SSIM : 0.967780
2b) Profil medium me = star : 13.65 fps PSNR : 46.191 SSIM: 0.967778

On a video 2mn subme:star is 7s faster.

Visually I see no difference.

This is different from the x264 ?

Originally Posted by Blowis

I wanted to ask is it normal that me=star is faster than umh.
I tested two videos:...

Interesting, but the basic tradeoff with speed is not quality, it's efficiency (quality @ bit rate). What were the bit rates?

[Blowis]

Originally Posted by x265_Project

Interesting, but the basic tradeoff with speed is not quality, it's efficiency (quality @ bit rate). What were the bit rates?

I use CRF:23 which gives a bitrate 1 724 Kbps and 2 597 Kbps.

After I tested rdpenalty:2 I lost 19% and 29% encoding time.
Visually minimal difference.

Originally Posted by Blowis

I use CRF:23 which gives a bitrate 1 724 Kbps and 2 597 Kbps.

I tested rdpenalty:2 I lost between 19% to 29% encoding time.
Visually minimal difference.

I'm confused. Can you put all the relevant information in a simple table?

command line options changed from default preset
bit rate
Global PSNR
SSIM
FPS

It would also help to know the details of your input video sequence (is it available, what is the pixel resolution, frame rate, and does it have low, medium or high motion)?

Lastly, it would help to know your computer system details (processor, memory, OS).

[Blowis]

Originally Posted by x265_Project

I'm confused. Can you put all the relevant information in a simple table?

command line options changed from default preset
bit rate
Global PSNR
SSIM
FPS

It would also help to know the details of your input video sequence (is it available, what is the pixel resolution, frame rate, and does it have low, medium or high motion)?

Lastly, it would help to know your computer system details (processor, memory, OS).

I use StaxRip.
Video 1a Profil Medium and:
wpp / ctu=64 / min-cu-size=8 / max-tu-size=32 / tu-intra-depth=3 / tu-inter-depth=3 / me=2 / subme=2 / merange=32 / no-rect / no-amp / max-merge=2 / temporal-mvp / early-skip / rdpenalty=0 / no-tskip / no-tskip-fast / strong-intra-smoothing / no-lossless / no-cu-lossless / no-constrained-intra / no-fast-intra / open-gop / no-temporal-layers / interlace=0 / keyint=240 / min-keyint=23 / scenecut=40 / rc-lookahead=20 / lookahead-slices=5 / bframes=4 / bframe-bias=0 / b-adapt=2 / ref=3 / limit-refs=0 / no-limit-modes / weightp / weightb / aq-mode=2 / qg-size=32 / aq-strength=1.00 / cbqpoffs=0 / crqpoffs=0 / rd=3 / psy-rd=1.00 / rdoq-level=2 / psy-rdoq=1.10 / signhide / deblock / sao / no-sao-non-deblock / b-pyramid / cutree / no-intra-refresh / rc=crf / crf=23.0 / qcomp=0.60 / qpmin=0 / qpmax=51 / qpstep=4 / ipratio=1.40 / pbratio=1.30
Video 1b unless such Me=3

The same for video 2a and 2b.

My Cpu I7 4770K@4.3Ghz DDR 8Go@2.4Ghz SSD 512go Samsung 840Pro Windows 10

[Motenai Yoda]

what x265_Project is trying to say is that you'll get the same quality but different bitrate, for the same bitrate you'll get different quality.

[LigH]

@ Blowis

Originally Posted by x265_Project

command line options changed from default preset

That means the command line in your encoder call, not the whole set of internal parameters stored in the auxiliary data in the video stream (that's simply too much to recognize at one glance).

A patch with new performance presets was pushed today. We look forward to your feedback.

[Motenai Yoda]

The table shows 2 row for limit-refs, but I'm pretty sure the first was for limit-modes, and limit-modes there aren't at all in the code too, only defaulted at 0.

Anyway I didn't get how to enable the 2 pass mode for cfr+vbv, if I have to.

Originally Posted by Motenai Yoda

The table shows 2 row for limit-refs, but I'm pretty sure the first was for limit-modes, and limit-modes there aren't at all in the code too, only defaulted at 0.

Anyway I didn't get how to enable the 2 pass mode for cfr+vbv, if I have to.

Thanks for the feedback. We'll get this cleaned up before the patch is committed (tomorrow).

[LigH]

Originally Posted by x265_Project

A patch with new performance presets was pushed today.

Just when I am away from my usual building PC ... well, I will try to publish one as soon as I rebuilt my compiling environments elsewhere.

P.S.:

Originally Posted by x265_Project

... before the patch is committed (tomorrow).

Oh, then I have a little time, ok.

[LigH]

Merry Christmas (or similar holidays) with a current build with updated preset defaults (e.g. limit-modes/refs; please RTFM).

x265 1.8+187-da48f2690076 (GCC 4.9.2)
x265 1.8+187-da48f2690076 (GCC 5.2.0)

[Ajvar]

Hey guys, I've been away from my horse-powere laptop and so from encoding videos. Can somebody write a short-list of what drastically changed since 1.7.1? In efficiency/speed. Thanks.

[Selur]

"what drastically changed", the presets changed, pme and pmode were added but those only seem to be only really useful on systems 4+ core systems, all in all I wouldn't call any changed 'drastic'

[Blowis]

Hi,
I just tested the last days made. I chose the Medium profile I modified all its: --crf 22 --early-skip --aq-mode 2 --me star --merange 44 --rc-lookahead 60 --keyint 240 --psy-rd 1.3.

wpp / ctu=64 / min-cu-size=8 / max-tu-size=32 / tu-intra-depth=1 / tu-inter-depth=1 / me=3 / subme=2 / merange=44 / no-rect / no-amp / max-merge=2 / temporal-mvp / early-skip / rdpenalty=0 / no-tskip / no-tskip-fast / strong-intra-smoothing / no-lossless / no-cu-lossless / no-constrained-intra / no-fast-intra / open-gop / no-temporal-layers / interlace=0 / keyint=240 / min-keyint=23 / scenecut=40 / rc-lookahead=60 / lookahead-slices=5 / bframes=4 / bframe-bias=0 / b-adapt=2 / ref=3 / limit-refs=3 / no-limit-modes / weightp / weightb / aq-mode=2 / qg-size=32 / aq-strength=1.00 / cbqpoffs=0 / crqpoffs=0 / rd=3 / psy-rd=1.30 / rdoq-level=0 / psy-rdoq=0.00 / signhide / deblock / sao / no-sao-non-deblock / b-pyramid / cutree / no-intra-refresh / rc=crf / crf=21.0 / qcomp=0.60 / qpmin=0 / qpmax=51 / qpstep=4 / ipratio=1.40 / pbratio=1.30


But I see --Limit-modes : 3 that normally there should be more than 0. You can set it to 0 or must be done manually.

[foxyshadis]

--limit-modes has no argument, it's either on or off. --limit-refs is 0-3.

Changes since 1.8: Modified presets, most are faster, higher quality, or both. Limit-modes. Intra refresh. 2-pass CRF. Monochrome (i400). Big psy-rd changes, possibly improvements. Lots of 12-bit speedups & bug fixes, and some other misc speedups and fixes for 8 & 10. Analysis save/load fixes. Level 8.5 (unlimited) allowed. i64x64 disabled until it's fixed.

See the 1.8 announcement for changes since 1.7.