Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > High Efficiency Video Coding (HEVC)

Reply
 
Thread Tools Search this Thread Display Modes
Old 16th October 2018, 04:21   #1  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
Looking for guidance.

The Situation.
To make a long backstory short, basically in the past years i transferred everything i had to digital from DVD, Bluray, tapes, comics, etc. This was a lengthy process as i didn't have a newer CPU and so i mainly left encode jobs going when i went to sleep and had it shutdown after it ended. During this period of time i bought into the idea of bigger is better, everything had to be 1080p for example (which at the time was considered kind of huge compared to dvd or standard quality) and so i ended up with a bunch of "bigger is better" results.

The Problem.
Fast forward to recently and i have since upgraded to an 1800x CPU and have immersed myself into more of an open minded process rather than the bigger is better idea. I've been having space problems where i'm constantly in need of more space. I've come to the conclusion that i made a big mistake with my approach in the past and i've ended up with a bunch of bloated and unnecessarily big files. One of the main reasons i went fully digital is to save physical real world space and i've since gotten rid of a lot of the source material (books/dvds/comics/blurays, etc) and i'm left with what i have now.

The current dilemma.
Okay. So what i've been doing is starting to shrink things down/removing what i see is the "bloat" from that bigger is better idea i was initially working with. The main change being that i've decided i don't need 1080p video and i'm okay with 720p and since that's what takes up the most space i've been working on transforming most of my 1080p content to 720p content either redoing encodes or resizing. In addition to this i'm also shrinking the bitrate down from the very high bitrates originally.

I've done extensive testing on this and i've had varying results from okay to good. Any time you take a source and lower it you're going to have some quality loss that is certain but because of the bloated nature of these big files i feel like i've been able to shrink them with what i consider success relative to real time watching conditions. If one were to analyze them frame by frame up close to the screen there would be some degradation noticeable but when watching them in real time in motion with distance from the screen it's not that noticeable.

I have three current methods:

A custom profile using x265 crf somewhere in the medium/slow preset range.

A two-pass profile using x264 with a very slow/film tune preset.

A crf profile using x264 with a very slow/film tune preset.

I'm running into a bit of a wall here in trying to figure out what is the most efficient path forward. I've had varying results where sometimes i prefer the x265 and other times i prefer the x264 and i keep going back and forth on each. My priority is speed and size over quality but obviously i'd like to have good quality relatively.

My question becomes, when presented with this situation and having x265 around the medium/slow range or using x264 with very slow presets which one becomes preferable in terms of efficiency in size, speed and quality.

Just to offer an example of speed time. On averages with the x265 profile it would usually take around 20-30min for a shorter episode and 45-1hr for a longer episode and you would typically see about half of the bitrate. Movies are more varying because of the difference in length, it's easier to specify an episode because of the similar length.

with the x264 it's been much less consistent with the crf option, i was working on a tv movie last night and i set up the usual very slow/tune film crf profile and set it at about 23.1 and it took about an hour and a half and ended up with a 6gb result which was like two times the size of the original movie. I Have no idea how that even happened. Based on examples like this i find the x265 crf more consistent but longer in speed. Using x265 it probably would've taken at least 2 hours but it would've been consistent in it's result.

What i have liked is the two-pass x264 option. Not only is it somehow usually faster than my crf examples in both x265/x264 but being able to control the output is valuable. The problem obviously comes in it's guesswork in trying to determine what bitrate/size is going to come out with good quality.

With both x264 and x265 using these 3 approaches i feel like i've been able to get near the original quality while successfully shrinking the bloated original files. I just can't come to a definitive conclusion on what i should go with as a standard.

Does very slow x264/tune film compare well to x265 at somewhere near the x265 medium tweaked toward slow, is it preferred when trying to accomplish this to use x265 on x264 being a different codec rather than going from x264 to x264, it would seem logically there would be more degradation taking the same codec and giving it less bitrate would have more of an effect than using a different codec but i don't know if that's how it works for sure, in real time testing it didn't seem to make a huge difference. I have a 1440p monitor so i can stack two 720p videos next to each other and watch them side by side and i can barely see any differences either way. The only difference being that x265 has a softer look maybe but in terms of quality i can't see that much degradation or difference in real time motion watching. That doesn't mean there isn't a difference just because i can't see it though.

Through my own extensive daily testing i haven't been able to reach a conclusion on what path to take going forward and so that's why i've now come here to ask for some help or guidance that can maybe help me decide. BECAUSE i don't want to run into another situation again that i did when i bought into the mindset of bigger is better and encoding everything at 1080p and with the highest settings i could no matter the size of the files... it's a messed up situation to find yourself in when you've done everything wrong compared to your current situation and have to redo everything. I don't want to go with x264 and then come to a realization in the future that i should've went with x265 and find myself in a similar position again.

To sum this all up i'm basically trying to determine what will offer the most efficiency between a very slow x264 preset and a medium x265 preset with some slow tweaks, where even though the x264 is "very slow" it's faster using my 1800x CPU than the tweaked medium x265.
Right now i sort of prefer the two-pass x264 very slow as the preferred option as it's faster than x265 and i can add film/grain/animation tunes to it without much problem and if i can find a way to pre- determine bitrates i can have that extra control. From everything i've read though it would appear that x265 perhaps is preferred for this sort of task when trying to achieve lower size files and maintaining a good quality.

(There's no tweaking using the x264 profile it's just set to very slow with a tune, with the x265 i've tweaked the basic medium preset in adding more ref/b frames, added weightb, changing from hex2 to star3-5, upping rd to 4 so it activates rdoquant, upped the lookahead to 60, moved the slices down from 8 to 4, sometimes adjusting deblock to -1 from 0, etc.. just a few tweaks like this nothing that huge. It maybe leans more toward slow than medium without the limitmodes/rect/asymm)

Any advice or help is appreciated on this. Thanks.

Last edited by AVspace; 16th October 2018 at 04:47.
AVspace is offline   Reply With Quote
Old 16th October 2018, 17:41   #2  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,624
Some questions:
  1. Have you done the math to confirm all the power and time and inevitable multigenerational quality loss actually gives you a worthwhile real cost-benefit payoff?
  2. Decide how much quality you want to retain? Is 720p enough for you, forever? How much visible encoding artifacts can you be happy with?
  3. How much smaller do the files need to get to be worth it? 30%? 80%?
  4. Can you live with 1% of the run time of 1% of the files looking significantly more degraded than your target quality? (it can sometimes take 20% more bits to fix the last 1% of 1%)
  5. Can you archive your existing x264 encodes in some kind of long-term storage in case you ever want to go back to those when you get a higher quality display or something? Or is the output of this what you're stuck with forever?
  6. How much CPU time are you willing to spend to so reencode your library (so total duration of content*average fps/encoding fps)? If you've got 1000 hours of 24p content and use settings that encode at 4 fps, It would take more than 8 months of your computer running 24/7. If you only run it for 12 hours a day while you're sleeping, etcetera, that's around 16 months.
  7. And don't forget the higher electricity bills for the compute, and the higher electricity bills to run more AC to cool the extra heat of the encode. For months on end. (If you live in Alaska or Siberia, you would just save on heating, so the economics are better).
  8. Does it make sense to start this now versus waiting for a higher quality and faster encoding version of x265, which would have multiple updates while you are reencoding your library?
This question has come up in different forms with different codecs over the years, and generally it's very hard to make a case for reencoding already lossy encoded files to save space versus just storing what you've got for any content you're likely to watch again. Even with ancient and nuts formats like 25 Mbps DV 480i.

If you've got a situation where that makes sense, we could offer some specific advice. But that would be contingent on a crisp definition of what your goals are.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Instant Video

My Compression Book

Amazon Instant Video is hiring! PM me if you're interested.
benwaggoner is offline   Reply With Quote
Old 17th October 2018, 00:40   #3  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
I've pretty much already decided and concluded that it's worth it. I'm like 6 months into my encoding journey.

As to 720p vs 1080p...

From my analysis and thinking i feel like 720p is the perfect sweet spot for my personal preferences and situation. There are some exceptions for say very cinematic feeling movies like Blade Runner or Lord of the Rings that i might stick with 1080p. I don't watch things on a big TV screen i pretty much do everything from a computer monitor these days. The 720p resolution fits this perfectly as not only can i fit the 720p video on my screen while mutitasking and still having a portion of my screen free but also still looking good in full screen when i want to sit back and focus on the video. The 1080p resolution i just find too big and bloated for these preferences and at my computer screen size i don't really get that much of a difference benefit between the resolutions when sitting back from the screen enough to need the higher resolution.

There is the concern that maybe in the future i want to go back to a bigger tv screen and then the 720p resolution isn't nearly as preferable but i don't really see that happening. I'm quite content with my setup. There's also the concern as resolutions continue to get bigger such as the growing 4k and hdr content that 720p is perhaps becoming less future proof. Possibly but i don't see 720p dying out any time soon/there's still a lot of people who watch lower resolution content out there. If anything i feel like 720p will have new life as the go to lower resolution and become the new SD/DVD/480p as things progress forward for all the people looking to lower file size content.

Storage is a big concern. I have a number of external drives including the western digital 8tb ones and it's still not nearly enough. There's no way to archive the old "bigger is better" efficiently. I don't know how anyone does it without running into space efficiency problems. Say like one 25gb bluray file i could instead turn that into the same movie and at least 3 seasons of a series. I just don't find it efficient at all any more to have those big 1080p+ files archived. Archiving to me is more about just having the item in acceptable quality not really about obtaining the best quality. If i want to watch something in the best quality there is streaming and blu-ray, etc.

As to the quality loss when shrinking files down as such i feel like there's a lot of misinformation out there about this topic. It heavily relies on situations like what one is doing to shrink it down and how one is watching the finished result such as how big the screen you're watching is at full screen, etc.

I've done extensive testing and there isn't much quality loss. Definitely not enough to be concerned about. I've taken some 720p source videos with the new versions and watched the entire thing side by side and there wasn't much noticeable difference in quality loss. In a blind test i couldn't even tell which was which.
I've watched a lot of my reencodes back and was more than content with the results. As stated in the original post if one was to analyze the video frame by frame up close in still images there would be some degradation but when it comes to real time motion it's not very noticeable. A lot of the original videos are bloated to begin with. At 1080p resolution there is more noticeable quality loss when doing this which is another reason i find 720p more preferable as it becomes more difficult to achieve this goal at 1080p.

I'm not really wanting to wait until x265 becomes faster, is there a definitive time table of when this is expected to happen? That is something to take into consideration though. If x265 is expected to improve further down the line as such then it might become more preferable in using it now as a standard with that in mind.

My ultimate question just becomes...

Option 1: Take the x265 path/either keeping the current custom profile or perhaps lowering that down to a default medium profile for speed increase. I'm sure that the results of x265 are good enough just not the overall efficiency.

Option 2: Take the x264 very slow path mainly focusing on two-pass where i'm sure the efficiency is good just not the overall results. (And with longevity in mind if x265 continues to improve in efficiency.)

Basically thinking that the "very slow" preset is one of the best, if i can obtain this with great efficiency using x264 is that the better option than needing to use a "lesser" medium -ish x265 profile. If it was the same preset i would likely choose x265 but it's very slow preset x264 vs a medium -ish x265. Is a x265 medium preset comparable to a very slow x264 preset in efficiency of size, speed, quality enough to make that your standard going forward when faced with the two options.

Last edited by AVspace; 17th October 2018 at 00:53.
AVspace is offline   Reply With Quote
Old 17th October 2018, 02:07   #4  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,624
Quote:
Originally Posted by AVspace View Post
As to 720p vs 1080p...

From my analysis and thinking i feel like 720p is the perfect sweet spot for my personal preferences and situation. There are some exceptions for say very cinematic feeling movies like Blade Runner or Lord of the Rings that i might stick with 1080p. I don't watch things on a big TV screen i pretty much do everything from a computer monitor these days. The 720p resolution fits this perfectly as not only can i fit the 720p video on my screen while mutitasking and still having a portion of my screen free but also still looking good in full screen when i want to sit back and focus on the video. The 1080p resolution i just find too big and bloated for these preferences and at my computer screen size i don't really get that much of a difference benefit between the resolutions when sitting back from the screen enough to need the higher resolution.
How long do you want to be able to watch these files? Because UHD monitors have been on the market for several years now, and a quarter of that screen is 1080p. If you don't mind these files only being useful until your next monitor, okay. But I'd never throw out resolution for anything I wanted to watch 2+ years from now.

Quote:
If anything i feel like 720p will have new life as the go to lower resolution and become the new SD/DVD/480p as things progress forward for all the people looking to lower file size content.
720p matters for mobile screen playback, and that's it. No one is making new content in it. <1080p is the new black & white .

Quote:
Storage is a big concern. I have a number of external drives including the western digital 8tb ones and it's still not nearly enough. There's no way to archive the old "bigger is better" efficiently. I don't know how anyone does it without running into space efficiency problems. Say like one 25gb bluray file i could instead turn that into the same movie and at least 3 seasons of a series. I just don't find it efficient at all any more to have those big 1080p+ files archived. Archiving to me is more about just having the item in acceptable quality not really about obtaining the best quality. If i want to watch something in the best quality there is streaming and blu-ray, etc.
Do you mean you are starting from the 40 Mbps peak BD rips, without any further encoding? That is archival grade, but you should be able to shrink that to 25-40% with x265 retaining most of the quality and all the pixels.

Quote:
As to the quality loss when shrinking files down as such i feel like there's a lot of misinformation out there about this topic. It heavily relies on situations like what one is doing to shrink it down and how one is watching the finished result such as how big the screen you're watching is at full screen, etc.
https://en.wikipedia.org/wiki/Nyquis...mpling_theorem
Scaling is going to reduce the maximum detail you can capture. Maybe an old movie was pretty soft to begin with. But if the source actually has 1080p detail, it will be lost forever when scaling it down.

Quote:
I'm not really wanting to wait until x265 becomes faster, is there a definitive time table of when this is expected to happen? That is something to take into consideration though. If x265 is expected to improve further down the line as such then it might become more preferable in using it now as a standard with that in mind.
It is always improving. There's no one future version that will change everything.

Quote:
Option 1: Take the x265 path/either keeping the current custom profile or perhaps lowering that down to a default medium profile for speed increase. I'm sure that the results of x265 are good enough just not the overall efficiency.
If you aren't able to use --preset slower at least, I don't know there is much point.

Quote:
Option 2: Take the x264 very slow path mainly focusing on two-pass where i'm sure the efficiency is good just not the overall results. (And with longevity in mind if x265 continues to improve in efficiency.)
Why two pass? do you want to control file size or quality? I'd recommend using --crf with both x264 and x265.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Instant Video

My Compression Book

Amazon Instant Video is hiring! PM me if you're interested.
benwaggoner is offline   Reply With Quote
Old 17th October 2018, 06:08   #5  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
My choice to use 720p isn't the issue haha. I thought about it a lot. I've done a lot of testing and theorizing. It wasn't a decision taken lightly, i went backwards in starting with the best quality i could get in trying to do .iso archived copies to then redoing them with lesser quality x264 encodes and more reasonable size compared to that, and then started to try some 1080p x265 encoding and now i'm finally to 720p... and this seems to be the place that fits for me and what my situation and preferences are. (this was over the course of like 8 years or so)

I agree that monitors continue to improve but i don't want to get some huge monitor either. I'm currently using a 1440p 27" and happy with it. I doubt i'd really want to go much bigger even my next upgrade. When i upgrade my next monitor i'd probably go with a similar size as it's big enough for me currently. With this in mind i don't see how 720p could fail me any time in the near/distant future even going into my next upgrade.

When it comes to things like HDR and 4k. I bought into the whole 3D thing and went and bought a new 3D tv and invested in tons of blu-rays and it was all pretty pointless really. I'm not really a movie or series watcher that cares about the pretty picture and special effects/i'm more of a substance kind of guy and i find that i kept getting caught up in these trends like the idea i needed to have the best archived quality i could when it's not really even what i needed. If even VHS and cassette tapes were still the norm it would fine with me. Was happy watching and listening to them all the same. Why all of a sudden do i need all of this new "best" just because it's what people say i need. I don't. (Admittedly DVD resolution like 300p or 360p whatever it is tends to be too low for me, there is definitely a certain line where even i can't stay relative to my screen. 480p-720p seems to be a pretty righteous sweet spot in what i'm okay with.)


I use crf for x265 but i find it more unreliable with x264 as stated in the original post. I get all kinds of wild results like sometimes twice the size of the original file. Maybe it doesn't work as well on an already encoded file with x264 i don't know but i've had much less reliable results with it compared to x265 crf. I find that i like two-pass with x264 in this capacity. I watch more series than movies and it is easier guesswork when determining a series bitrate and so in that regard two-pass isn't as much of a problem usually. When it comes to movies it does become more of a question mark but it's all a learning process i guess.

The current profile i use with x265 is pretty close to "slow" it just lacks rect/limitmodes/asym but it also pumps up some things like ref/bframes. It's not quite slow and not quite medium. It provides pretty good results from my eyes it's just not the most efficient when it comes to speed and perhaps has a bit too "soft" of an image at times. I keep messing with it trying to find that sweet spot where it's not as soft but then doesn't have that dirty or plastic look either but haven't found it yet. I tend to like the natural x264 picture a bit more.

Mainly, i'm just looking for some technical guidance from some of you experts. Theoretically and technically is a very slow preset encode using x264 with a tune addition (film/animation/grain) expected to be more efficient than let's just say a slow x265 encode. Also a medium encode because i'd like to know that also. This is where i'm having trouble in decided if i should just stick to x265 when i can use "very slow" on x264 and still have faster encodes that have that more natural x264 image i prefer and also can use grain/animation/film tunes easier than trying to match it with x265. It's not an easy conclusion to come to. That's why i'm hoping someone with the technical expertise can maybe give me an answer in how these options match up technically in expectations.

Last edited by AVspace; 17th October 2018 at 06:18.
AVspace is offline   Reply With Quote
Old 17th October 2018, 08:25   #6  |  Link
Forteen88
Herr
 
Join Date: Apr 2009
Location: North Europe
Posts: 304
The x264-settings for
--tune
aren't that great, especially not the grain-setting (which can be used ONLY if the movie is a grain-fest). You can see some threads here at Doom9 how to set things for specific sources (grain/animation/film) yourself.

For example,
(OBS! I haven't done enough x265-testing myself, these options are mostly taken from advice of Doom9ers)
My general x265-recommendation for high-quality encodes,
x265 --crf 18 --preset slower --profile main10 --no-sao --deblock -1:-1 --qcomp 0.7 --ref 6 --merange 48 --rc-lookahead 48 --bframes 8 --frame-threads 2

My grained recommendation (added to the general recommendation):
--tune grain --no-sao --psy-rdoq 0 --no-strong-intra-smoothing --bframes 3 --ctu 32 --max-tu-size 16 --qg-size 16

My 3D-animation recommendation (added to my general recommendation):
--ref 6 --bframes 16 --aq-strength 0.8 --psy-rd 1 --tu-intra 4 --tu-inter 4

My cartoon/anime-animation recommendation (added to my general recommendation):
--ref 6 --bframes 16 --aq-strength 0.6 --psy-rd 1 --tskip --tskip-fast --tu-inter 4 --tu-intra 4 --limit-tu 4
Forteen88 is offline   Reply With Quote
Old 18th October 2018, 05:04   #7  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
I read through a lot of threads here on the forum. Been lurking. The problem though is most of the ideas/thoughts are based on fresh encodes and trying to achieve the "best" quality whereas i'm trying to take an already fresh encoded file based around this "best quality" mindset and redo it to a more preferable result than the original where i feel these are bloated to begin with as i don't need the "best" quality. Not without quality loss just ideal to my personal preferences and i don't find the quality loss based on my methods to be a problem. I'm also working with lower resolution and hence lower bitrate than what is based in a lot of these theoretical ideas presented. Like in my situation i wouldn't be anywhere near a crf18. It's a whole different level where i am hehe.

It might just be a case where no one can give me the answer that i seek and i need to find out on my own. Maybe i can work out a sort of filter where when i prioritize speed then go with x264 very slow preset and when i want to focus more on quality then go with the slow x265 preset. I suppose that could work in theory. It's not the most ideal way in comparison to setting a standard but maybe a patchwork solution at least to where there's some kind of conclusion to the scenario.

Well, thanks for the responses at least i was able to bounce some ideas off someone other than myself which helps get a clearer picture sometimes.

Last edited by AVspace; 18th October 2018 at 05:39.
AVspace is offline   Reply With Quote
Old 27th October 2018, 00:17   #8  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
Just a final update to this thread.

After some recent testing i've decided and concluded to go with x264 "very slow" for now. I was doing some direct comparisons and i feel like x264 holds up surprisingly strong. On a few of the tests there was clear artifact degradation created in the x265 where there wasn't any in the original and the x264 setting kept that image. This was testing in frame by frame still pictures. In real time motion it's difficult to see a big difference but the fact that x264 was holding it's own gave me more trust in using it. I also just like the look of the picture x264 produces a little more. Haven't been able to quite reproduce that same look in x265.

What i was doing was using the staxrip video comparison tool and then comparing clips like say x265 with maxmerge2 and then maxmerge3, with star3 or hex3, loopfillter -3 and +3, etc and then choosing a frame and having it compared to the original and there was so many differences from just the way it shaped the bottom lip, to part of the hair and on and on. Each subtle change seemed to have some effect to the image at around the "slow" setting and below. I found being able to use "very slow" with x264 very satisfying in this regard because there wasn't much to then change at this setting.

Anything slower than "slow" on x265 is just too long in length for my preferences for the encode to finish and i don't find that this setting holds up quite as well as "very slow" x264 when at this setting and looking at my various testing and results. I'll keep an eye on x265's development and when it becomes faster and more efficient to where "slower" and "very slow" settings is a reachable goal and then i'll probably revisit this subject at that time. I think for now in my situation and what i'm trying to achieve x264 with the very slow preset and specific tune is quite an acceptable choice. I'll probably go with crf rather than two-pass as it is more reliable than the guesswork involved with two-pass although i tend to prefer two-pass it's just more difficult to obtain that same reliability, and likely using 10bit rather than 8bit.

So, that seems to be the path of choice for now as concluded until a further time...

Last edited by AVspace; 27th October 2018 at 00:56.
AVspace is offline   Reply With Quote
Old 30th October 2018, 16:55   #9  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 2,624
If you want reliable quality @ perf, I am not surprised if x264 comes out on top if using the same settings on a wide variety of sources.

The more cores and more AVX available, the smaller the perf gap x265 will have, and the more it'll make sense. I generally think that x265 will generally win at --preset slower and beyond.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Instant Video

My Compression Book

Amazon Instant Video is hiring! PM me if you're interested.
benwaggoner is offline   Reply With Quote
Old 30th October 2018, 22:42   #10  |  Link
Forteen88
Herr
 
Join Date: Apr 2009
Location: North Europe
Posts: 304
Quote:
Originally Posted by AVspace View Post
I think for now in my situation and what i'm trying to achieve x264 with the very slow preset and specific tune is quite an acceptable choice. I'll probably go with crf rather than two-pass as it is more reliable than the guesswork involved with two-pass although i tend to prefer two-pass it's just more difficult to obtain that same reliability, and likely using 10bit rather than 8bit.
10-bit in x264? That makes the encode not GPU-decodable, unlike 10-bits x265-encodes (which can be fully GPU-decoded with Nvidia GF 960, or better).
For this to work in x265, you must set maximum --ref 6

Last edited by Forteen88; 31st October 2018 at 07:27.
Forteen88 is offline   Reply With Quote
Old Yesterday, 13:59   #11  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
Quote:
Originally Posted by benwaggoner View Post
If you want reliable quality @ perf, I am not surprised if x264 comes out on top if using the same settings on a wide variety of sources.

The more cores and more AVX available, the smaller the perf gap x265 will have, and the more it'll make sense. I generally think that x265 will generally win at --preset slower and beyond.
Agreed for the most part. I've been mainly using x264 since i made this thread but i still switch to x265 now and again.

From what i've seen in performing and watching the various encoded video results i believe at the settings i'm using for x264 can definitely compare with x265 but there is a clear advantage for x265 in compression. I always see more efficient file size to bitrate ratio with x265 and they both look pretty good relatively. I think once you remove sao/strong intra smoothing and maybe go something like -1 deblock x265 starts to match up closer to x264's picture which x264's "look" might be preferable to people, i know it is for me... that smooth plasticy look that x265 has on default i like x264's default look compared to that but once i switched off some of those smoothing options in x265's defaults it seems to compare better. (at least to my eyes. Those smoothing options might be better to have on for lower resolution content like DVD resolution i'm not sure enough to say always change it.)

When using x264 i'm basically making a conscious decision to sacrifice either some bitrate or file size for the gain in encoding time. When referencing series episodes i generally see about at least a 20 minute difference between my custom very slow x264 and a basic slow x265 profiles as an estimation but it's usually a bit more than 20min. That 20min or so that i save i just find too important to use x265 as my main. I find x265 the most useful when i really want to get the most compression out of my encode and hit a lower file size at a specific quality bitrate.

I still find x265 "slower" preset slightly lacking to where i wouldn't be always confident using it compared to x264 as my main choice. When x265 potentially sees a speed increase in it's future and "slower" presets become more user friendly that's when i'll probably change to x265 being my main choice. If i could close that encoding speed gap while getting the efficiency in compression that x265 brings that would be nice. That would require quite a speed boost improvement though because currently at slower preset in x265 it takes me soooo long in my encodes compared to x264 encodes. It might be a while hehe.

Last edited by AVspace; Yesterday at 14:29.
AVspace is offline   Reply With Quote
Old Yesterday, 14:01   #12  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
Quote:
Originally Posted by Forteen88 View Post
10-bit in x264? That makes the encode not GPU-decodable, unlike 10-bits x265-encodes (which can be fully GPU-decoded with Nvidia GF 960, or better).
For this to work in x265, you must set maximum --ref 6
Not sure what you mean. Almost everything i've read stated that 10bit encoding if it's compatible with your needs is preferable. I've seen no issues when encoding and playing back videos in 10bit. If you mean decoding in the encode or decoding while playing back the video it hasn't been any problem for me.

Last edited by AVspace; Yesterday at 14:08.
AVspace is offline   Reply With Quote
Old Yesterday, 15:34   #13  |  Link
hello_hello
Registered User
 
Join Date: Mar 2011
Posts: 3,710
Quote:
Originally Posted by AVspace View Post
Not sure what you mean. Almost everything i've read stated that 10bit encoding if it's compatible with your needs is preferable. I've seen no issues when encoding and playing back videos in 10bit. If you mean decoding in the encode or decoding while playing back the video it hasn't been any problem for me.
I'm pretty sure HEVC had a 10 bit profile from the get-go, so it's well supported. For AVC, not so much.....

Not that I looked particularly hard, but I saw no mention of 10 bit AVC support here,yet I did see support for 12 bit HEVC.
AVC is stuck at 8 bit in respect to hardware player support. I doubt it'll ever change.
hello_hello is offline   Reply With Quote
Old Yesterday, 18:47   #14  |  Link
AVspace
Registered User
 
Join Date: Oct 2018
Posts: 14
Quote:
Originally Posted by hello_hello View Post
I'm pretty sure HEVC had a 10 bit profile from the get-go, so it's well supported. For AVC, not so much.....

Not that I looked particularly hard, but I saw no mention of 10 bit AVC support here,yet I did see support for 12 bit HEVC.
AVC is stuck at 8 bit in respect to hardware player support. I doubt it'll ever change.
I see. As stated that just seems to be a compatibility concern.

That's not an issue for me as i do everything from a PC, in that situation where compatability isn't an issue from what i've read 10bit is preferable.

One source example,
"Yes, you can get better compression efficiency (i.e. better quality at same bitrate) with 10-Bit, even when your source is "only" 8-Bit and the final output is 8-Bit too, simply because the encoder's and decoder's internal precision is higher (-> less rounding errors). On the other hand support for 10-Bit or even 12-Bit H.264 doesn't exist in any (consumer) playback device at the moment. Even software decoders are still very rare currently. So this is something you probably don't want to use yet. Also high-bitdepth H.264 will be significant slower to encode/decode. It was said that 8-Bit basically is a trade-off for speed" --LordMulder

I don't have any issues with support and the encoding with it doesn't seem slow to me at all and so i haven't had any issues.

(I don't know how much this would help me in the future but i plan to upgrade to a 10bit monitor eventually. I haven't really looked into it yet and how that all works but it could be some potential future proofing, but, even if it's not if i'm able to get a little extra efficiency from using it then it's worth it. I also use things like "no pskip" and still okay with my speed. It's pretty fast relative to my other comparison option x265 as explained in the post. At least 20min difference on episodes but usually more and that's with 10bit and things like no pskip and subme11, etc. I'm still working on a main configuration in trying to get all the bonus quality i can get while maintaining time speed over the x265 slow profile. My current one is looking good but i feel like i could still squeeze a bit more in there and so i'm still experimenting with that. Oh i should point out and that's all including the removal of rect from x265 slow which increases the speed. If that was on it would be more like 35m, although with limit modes enabled it sometimes show rect as on in the logfile so i'm not sure how that works exactly but usually when i remove rect from x265's slow profile it increases the speed by about 15min and that's all considered against x265 slow with rect unchecked and so that's giving some benefit to the x265 side of things in the comparison).

Side note: Is there an ongoing discussion where questions can be asked about x264 settings? For whatever reason i couldn't seem to find it haha. The x264 forum looked to only have a few threads on it.

Last edited by AVspace; Yesterday at 19:39.
AVspace is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 01:36.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2018, vBulletin Solutions Inc.