x265 vs x264

[deadrats]

ok guys, here is the original and some test encodes, as i said this is from one of my go to test files as i find it one of the more difficult to get good results with. i used avidemux to open the vc-1/wmv source and cut a 16 second clip that is within this forums rules and used an avi container.

a few notes, the source has an average bit rate of 12mb/s but the portion i cut is listed as 7mb/s and if you calculate the file size it is in fact 7mb/s so all my tests were done with the test encoders being set at average bit rate of 7mb/s. some of the encoders undershot and the source files seems to be primarily composed of b frames.

as for the results to my eyes all the test encodes suck, i tested with xvid and vp8 within Hybrid, ffmpeg mpeg-4 and qs within handbrake and i used xvid4psp to do 8 bit and 10 bit x264 test encodes with all the quality settings maxed out.

i would have tested with vp9 but i can't get it to work on my system, so if Selur or someone else wants to do a test encode with vp9 with all the quality setting maxed out i would be interested in seeing how it turns out.

as it stands now, the test encode i did with Strongene's h265 encoder just beats everything else hands down.

i will also be uploading a few main concept test encodes so you guys can see what a quality encoder can do.

[El Heggunte]

This may sound and may be "off-topic", but I think it's absolutely necessary.

i used avidemux to open the vc-1/wmv source and cut a 16 second clip that is within this forums rules and used an avi container.

Avidemux for dealing with ASF files? VERY-BAD idea. "rws_asa.avi" was muxed and indexed incorrectly. All frames are marked as keyframes, and it's impossible to extract from it a well-formed .VC1 file (elementary stream). Besides, the AVI container is not suitable for Windows Media Audio. You really should have used AsfBin instead.

[ end of just-my-2-shillings ]

[Selur]

I'm a bit confused why you would choose (compared to the source) such a high bit rate for the reencodes?
I mean non of these codecs are normally ment as intermediate format.
Wouldn't something along 1500 or lower been a better choice if we look for compression efficiency?
Or are you aiming to look for transparency at similar bit rate?


btw. strange choice of parameters for x264, looks like:

x264 --profile high --level 5.2 --ref 5 --no-mixed-refs --min-keyint 25 --bframes 5 --b-adapt 2 --qpmax 21 --no-fast-pskip --no-dct-decimate --me umh --subme 11 --trellis 2 --aq-mode 2 --cqm jvt --threads 12

with a target bit rate of 7000.
Especially deactivating mixed refs, seems strange to me, but I'm sure you chose what you deem to be best/maxed out.
It would be nice to see an example of the output of the mysterious H.264 encoder which is better than x264.

btw. personally I would have chosen something more a long the lines of:
http://vimeo.com/66124969

Cu Selur

Ps.: encoding VP9 at FullHD is just too slow for my taste to test at all atm.

[enim]

My chantingel from The City of Angels has provided me few samples which I would like to cite here for purely educational purposes. If I do not post here, probably you will never get them anywhere else. Probably you will never know what Quality means. To some these may help to learn artifacts,to some these may help to understand how they are being sucked, and to some these may help to improve up on. Your inputs and feedbacks are always welcomed.

I could have decided to remain silent, as I already have the things I need.
But, you will never know the difference.


SAMPLE-1
--------
Sample-Frame-433.mp4


If you l@@k @ frame#433 in above sample, it is something like the pic posted as under:



Where as, the encoder chopped off the fingers of forearm with bent wrist. Or, someone's else organ attached to her near elbow. Plus, whole sample contains tons of variety of artifacts. I have already given-up restoration on sample with tons of artifacts and lots of missing details. Still you can try. and waste your energy and time in restoring and re-encoding.

I would be highly interested to know about mathematical model that chops off lots of details or attach someone's else organ to someone else. Do not expect me to tell you the answer of 1+1 coz I do not know either. I luv to do my assignments only.

Following two samples definitely look better than (???) to me for sure.

SAMPLE-2
--------

SAMPLE.GooD.mp4

Good enough. Fair Quality. Preservation of lots of details. No grains!!!
Nice mix of Compression and Quality.


SAMPLE-3
--------

Sample.HiQ.mp4

Simply HiQ.
We everyone would like to have tool like Hanzo's Samurai.

You can download all these samples, and see the difference.
You can also carry on post-mortem, It will lead you in the right direction, I hope.
If so, it will definitely help lots of people who need tools like Hanzo's Samurai.

Otherwise, please do not waste time! please! bcoz...
Most of the time broken and handicapped tools are distributed on very specific and definite purpose.
i HATE broken and handicapped tools.

Your feedback/inputs are always welcomed.

PS: I would like to request my chantingel to provide me both source and full-length encoded files.

[enim]

As one of the foundation stone for HEVC says

HEVC is said to double the data compression ratio compared to H.264/MPEG-4 AVC at the same level of video quality. It can alternatively be used to provide substantially improved video quality at the same bit rate.

I would LuV HEVC on the day on which HEVC can encode above my samples at half bitrate. Otherwise it's wasting cpu power, energy, time and other resources. Till then...

[enim]

As one of the foundation stone for HEVC says

HEVC is said to double the data compression ratio compared to H.264/MPEG-4 AVC at the same level of video quality. It can alternatively be used to provide substantially improved video quality at the same bit rate.

I would LuV HEVC on the day on which HEVC can encode above my last two samples at half bitrate. Otherwise it's wasting cpu power, energy, time and other resources. Till then...

[addu]

Only x265 will be worth when we will see realtime encoding 4k ultrahd on quad or eight core machines. Thenafter all the comparison of such kind is worth mentioning. x264 is more than realtime on my quad core pc with 1080p encoding. At present x265 is too slow for any comparison.

[jamespoo]

been doing some test with the latest naughty build of handbrake and a 1080p full hd sample clip and in my test i can get the bitrate down to around 1800VBR and it looks the same as the original blu-ray 1080p clip but using x264 the bitrate needed to be 3100vbr

i am loving h265

[kotuwa]

Zones
Is there any switch or method in x265 to specify zones like --zones in x264 ?
!?

[Selur]

tom?

[movmasty]

x265

--zones <start frame>,<end frame>,<option>/
<option> is either q=<integer> or b=<float>(bitrate multiplier)

[KarMa]

Only x265 will be worth when we will see realtime encoding 4k ultrahd on quad or eight core machines. Thenafter all the comparison of such kind is worth mentioning. x264 is more than realtime on my quad core pc with 1080p encoding. At present x265 is too slow for any comparison.

Hope you are using the one of the newer releases and not just the one that comes with handbrake (x265 - 1.5)