Author Topic: Hi10P and 8-bit encodes  (Read 64127 times)

Offline OnDeed

  • Member
  • Posts: 448
  • Uploader account for #OnDeed@irc.rizon.net
Re: Hi10P and 8-bit encodes
« Reply #280 on: December 30, 2011, 01:54:57 PM »
Seeing how are people eager to do this jump into the future (come on, I don't think 6 months is exactly a lot).
"Moving to the future" is fine in my book, but I don't think that a torrent tracker/release aggregation/archival place is a good place to drive that. I would leave that to the elitist-enough fansubbers an rippers.

Since I usually do releases of stuff where there is an alternative choice to download (even if it has worse video), I'm not too heartbroken with making a 10-bit rip (dvdrip, anyway...). However, there is a fansub project I'm part of, and the eventual releases will be 8bit, unless the other people I'm collaborating with strongly veto it. The reason is that if/when we finish it, it is going to be the first complete English release to date. I don't want to cut people away in such case.

TL;DR
I don't think bakabt really needs to take too active role in this transition. Not even leaving a single fallback slot (as many people in the comments seem to be perfectly fine with) is outright bad imho.
(Come on, xvid was put to "equal treatment" policy just recently... about 5 years after the introduction of h.264?)

Offline erejnion

  • Member
  • Posts: 11
  • please excuse my inexperience~
Re: Hi10P and 8-bit encodes
« Reply #281 on: December 30, 2011, 06:58:01 PM »
@0squid0
mkv and mp4 usually use the same compression algorithm for the video in them.

By the way... a lot of people who would be against transiting to equal treatment probably didn't even notice the poll, since they enter here only once per month or something like that, in order to download the next complete series to watch. The guys who are more active in the very site, at least in my opinion, are more likely to be more active in the fansub scene as a whole, thus more likely to prefer 10bit.

Offline OnDeed

  • Member
  • Posts: 448
  • Uploader account for #OnDeed@irc.rizon.net
Re: Hi10P and 8-bit encodes
« Reply #282 on: December 30, 2011, 07:22:10 PM »
By the way... a lot of people who would be against transiting to equal treatment probably didn't even notice the poll, since they enter here only once per month or something like that, in order to download the next complete series to watch. The guys who are more active in the very site, at least in my opinion, are more likely to be more active in the fansub scene as a whole, thus more likely to prefer 10bit.

Hmm, good point.

Offline straypup

  • Member
  • Posts: 3
Re: Hi10P and 8-bit encodes
« Reply #283 on: December 31, 2011, 12:51:40 AM »
Everyone keeps complaining about hardware this and hardware that. Just because you have an older computer does not mean you can not watch a 10-bit video. Like I said earlier, I have a PCI graphics card in my computer and I can watch 10-bit on a pentium 4 system. The main question that was asked was whether 10-bit should replace 8-bit, or be compared and treated the same.

Offline SupraGuy

  • Member
  • Posts: 55
  • Car nut.
Re: Hi10P and 8-bit encodes
« Reply #284 on: December 31, 2011, 05:30:43 AM »
I guess I'll chime in.

Stand-alone media players cost less than $100, sit there on the network, have zero moving parts, and don't have a lot of power dissipation. Connect it to your HDTV (Or 1080p capable LCD monitor) and use it for what it does -- play media. A purpose-built Mini-ITX PC with an IR remote and enough processor and video to playback 1080p Hi10p media costs considerably more than $100, and then it becomes tempting to use it for other unrelated things. It's definitely going to use more power, which means more heat, which means ventilation, particularly if you want it quiet, which means LOTS of ventilation for passive cooling.

My main concern though is that Hi10p support doesn't even seem to be on the radar for the stand-alone media players. I see mention of it on things like the WDTV forums. People with the ASUS O!Play devices (like me) aren't even bothering to ask, because we know that it's just not going to happen. (Which is a shame, because I get really nice playback of other 1080p x264 media from the device.) Now that the Matroska header compression thing is sorted out in the firmware, I quite like it.

Hi10p is definitely not automatically better, and makes certain watching choices impractical. In this case,

Another system I have just doesn't have the grunt on its own. With CoreAVC (v2.6) codec, it can handle 1080p x264 reasonably well, with it only glitching on really high compressed areas. As such, I was happy with it. I am NOT going to spend money on upgrading the video card, or in fact anything on that machine. When its time comes, I'm going to hit the HDD with a hammer and junk it. (Probably save the BD-ROM drive and HD tuner that's in it for the next one.) It's the main media player for the living room. It can muddle through MOST Hi10p x264 720p media with minimal lag, but 1080p... Nope. Maybe if I didn't have to deal with Windows overhead (Since a Linux BD player is still vapourware) but as is, it's not going to work. For now, this means that I can't watch this stuff from the couch. This is also where the best sound system is at the moment, at least until I get the rest of the projector system finished.

The machine connected to my projector works just fine. No problems there. Similarly my main machine is also good.

The one is only a matter of time until I have to upgrade the machine for one reason or another, dead processor, mobo, RAM, whatever. At that time, I'll probably shuffle one of my current higher end systems into that duty, and then replace that with something hotter, which is what I usually do. This still leaves me out of watching Hi10p releases in the most comfortable watching environments in the house.

My current solution? Re-encode the video in 8-bit. It's time consuming, but since I get the re-encode started as soon as I finish downloading, doesn't tend to hold me back long, but it's a pain, and then I can't seed the downloads later unless I keep both versions.

Oh, and I've yet to find encoding problems with my re-encodes, like macro blocking, dot crawl or banding -- that wasn't there in the original -- and most of my filesizes are similar, if not smaller. Only a few are a couple of percentage points bigger.

Offline DmonHiro

  • Member
  • Posts: 797
  • Neko The World
Re: Hi10P and 8-bit encodes
« Reply #285 on: December 31, 2011, 07:17:50 AM »
Oh, and I've yet to find encoding problems with my re-encodes, like macro blocking, dot crawl or banding -- that wasn't there in the original -- and most of my filesizes are similar, if not smaller. Only a few are a couple of percentage points bigger.
That's BS. There is no way that a reencode, especially from 10bit to 8bit, is going to be smaller AND have the same quality as the original. Just because YOU can't see the problems does not mean they are not there.

And for the last time: BakaBT has NEVER pandered to the hardware demographic.
Demons run when a good man goes to war. Night will fall and drown the sun, when a good man goes to war. Friendship dies and true love lies, night will fall and the dark will rise, when a good man goes to war. Demons run but count the cost, the battle's won, but the child is lost.

Offline pjkperker

  • Member
  • Posts: 21
Re: Hi10P and 8-bit encodes
« Reply #286 on: December 31, 2011, 07:58:43 AM »
Oh, and I've yet to find encoding problems with my re-encodes, like macro blocking, dot crawl or banding -- that wasn't there in the original -- and most of my filesizes are similar, if not smaller. Only a few are a couple of percentage points bigger.
That's BS. There is no way that a reencode, especially from 10bit to 8bit, is going to be smaller AND have the same quality as the original. Just because YOU can't see the problems does not mean they are not there.
Why are you stopping there, ask him to show some proof.May be he will introduce us to a new generation of video encoding.
By the way... a lot of people who would be against transiting to equal treatment probably didn't even notice the poll, since they enter here only once per month or something like that, in order to download the next complete series to watch. The guys who are more active in the very site, at least in my opinion, are more likely to be more active in the fansub scene as a whole, thus more likely to prefer 10bit.
Very true+ those guys will come after another month and say "The one I downloaded(8-bit) wasn't the best version? So much about BakaBT storing the best quality!"



Offline Daiz

  • Member
  • Posts: 297
  • 10-bit Librarian
    • Underwater
Re: 10-bit and 8-bit encodes
« Reply #287 on: December 31, 2011, 10:32:03 AM »
Stand-alone media players cost less than $100

Yet even before 10-bit they still couldn't render ASS subtitles properly, nor could they probably play anything above Level 4.1 High Profile H.264, and beyond that good luck trying to get ordered chapters to work on them. There has already been lots and lots of 8-bit fansub releases that would be incompatible with your plastic toy.

Fansubbers do not cater to hardware players. You should have known this when you bought a plastic toy with limited capabilities to play fansubs.

Also, if you want the most performance out of your PC for 10-bit playback, use LAV H.264 decoder with xy-vsfilter or alternatively mplayer2 (the faster subtitle rendering with libass might help you quite a bit).
« Last Edit: December 31, 2011, 12:02:16 PM by Daiz »

Offline AceD

  • Member
  • Posts: 2665
    • Facebook
Re: Hi10P and 8-bit encodes
« Reply #288 on: December 31, 2011, 11:52:16 AM »
Nearly all fansubs and BD releases (Coalgirls, THORA etc) are released for PC, and for playback on PC...peoples opinions on holding the push back just because some overpriced shit or console can't play 10-bit should be completely ignored

Offline Puiu

  • Member
  • Posts: 7
Re: Hi10P and 8-bit encodes
« Reply #289 on: December 31, 2011, 02:39:21 PM »
I really can't understand why people want to play 1080p on screens that aren't 1080p. i don't expect to play 1080p on my 2007 laptop and i don't even want to.

Now back on topic:
I noticed when i switched to 10bit fansubs that there was much less banding and some even had better contrast/colors, but i'm not sure if 10bit is the cause - maybe the size has something to do with that, also most releases are the same size or smaller than 8bit ones.

I'm using potplayer beta (the new and much improved kmplayer) with LAV and MadVR. 720p works just fine.

It is time to switch to something new. BakaBT is about having the best! This is the only reason i come here.



PS: complaining that your netbook won't be able to play 1080p? pretty lame.

Offline Hadouken

  • Member
  • Posts: 1441
  • .....
Re: Hi10P and 8-bit encodes
« Reply #290 on: December 31, 2011, 05:00:21 PM »
Talk about being elitist.. lmao.
I'm not really complaining about 10-bit, it's just that the sudden hatred for 8-bit is laughable considering that's what everyone was watching with no problem earlier this year.
.

Offline Aadieu

  • Member
  • Posts: 103
Re: Hi10P and 8-bit encodes
« Reply #291 on: December 31, 2011, 06:29:47 PM »
Ever since somebody on here explained how there is no actual non-proartist equipment out there to actually DISPLAY 10bit colour, and considering how the "less banding" thing seems to not show up in practice*, I fail to see the point of 10bit today at all.

*: maybe groups don't know how to ENcode 10bit properly, maybe existing DEcoders today are screwing it up, or maybe the banding thing is a myth - in any case, has anyone actually SEEN a quality gain? do post comparisons (of *similarly sized* files, from the same source!)

Offline Daiz

  • Member
  • Posts: 297
  • 10-bit Librarian
    • Underwater
Re: 10-bit and 8-bit encodes
« Reply #292 on: December 31, 2011, 06:56:22 PM »
Ever since somebody on here explained how there is no actual non-proartist equipment out there to actually DISPLAY 10bit colour, and considering how the "less banding" thing seems to not show up in practice*, I fail to see the point of 10bit today at all.
Well, congratulations on being an idiot, then. Let me quote my own post from page 3 of this thread:
The bit depth of your monitor is irrelevant - the main benefit of 10-bit is the increased compression due to the higher precision, and that is something that takes effect on the encoding side and isn't in any way affected by the bit depth of the display you are viewing it with.
To add to that, the compression benefit applies no matter what the source bit depth is.

do post comparisons (of *similarly sized* files, from the same source!)
Here's two:
http://blisswater.info/comparison/elephantsdream/ (a comparison at exactly the same bitrate from a lossless source)
http://screenshotcomparison.com/comparison/66418 (Exactly the same source and encoding settings with the only difference being the bit depth - the 10-bit video turned out 15% smaller, too)

Offline brunoais

  • Developer
  • Member
  • Posts: 1481
  • It's juice and jam time!
Re: Hi10P and 8-bit encodes
« Reply #293 on: December 31, 2011, 07:02:42 PM »
I believe we only need an extra slot in the C part.
The reason? 10-bit releases uses more CPU than 8-bit releases so we will offer, at least, 1 8-bit release in the weakest "department". This can be used, as an example, in situations where the viewer uses a computer which is also occupied doing other things and the member wants to view the video while the computer is working on other stuff (as an example).

The perfect for me would be to make available the best 8-bit release in slot C and, if there's no 8-bit for slot C use the best 8-Bit for slot B. This way we can somewhat please the ones that what to watch the videos in their PDA, tablet, etc... , for example (have a 8-bit release, when possible),  and we won't have too many torrents for each show.
Want more smilies in bakabt? Check this: http://forums.bakabt.me/index.php?topic=28322.0
Please!!! I need hentai recommendation here: http://forums.bakabt.me/index.php?topic=28566
BLOG POST
BBT Ako & Riko Suminoe Fanclub#000004

Offline OnDeed

  • Member
  • Posts: 448
  • Uploader account for #OnDeed@irc.rizon.net
Re: Hi10P and 8-bit encodes
« Reply #294 on: December 31, 2011, 07:06:32 PM »
There are some wierd differences in the Moe Onna comparison... was that with equal filtering? At least #5 seems to have some hue shift.

P.S. You meantioned the hardware players being "incompatible" with ass. They might not use fonts and ignore tags, but is that really critical? Usually it should be more than okay to watch&read subtitles without that - in some cases of lousy font choice it would be even beneficial. While it might hurt the TSer/styler's feelings, such stuff is not overly important.

@ Brunoais
There is a good reason for 8bit HD too - many people can play high bitrate 1080p via gpu/igp decoders, while they might not be able to play such 10bit (weak cpu) or even any 10bit at all (standalones).

Offline Daiz

  • Member
  • Posts: 297
  • 10-bit Librarian
    • Underwater
Re: 10-bit and 8-bit encodes
« Reply #295 on: December 31, 2011, 07:19:01 PM »
At least #5 seems to have some hue shift.
I used CCCP's ffdshow to decode the 10-bit video back when I did that comparison some 6 months ago, which had the minor color conversion swscale issue with 10-bit material. The filtering is otherwise identical, except 16-bit input was used for the 10-bit encode whereas dithered 8-bit input (from the same debanding filter) was used for the 8-bit one (feeding 16-bit input to 8-bit x264 gave worse results banding-wise).

Also, I would say that having typesetting fuck up could end up badly - depending on whether a player would simply not display multiple lines at the same time or show them unstyled onscreen all at the same time, you'd either get a screen full of text at times or alternatively miss out on important signs, or both.
« Last Edit: December 31, 2011, 07:22:00 PM by Daiz »

Offline Aadieu

  • Member
  • Posts: 103
Re: 10-bit and 8-bit encodes
« Reply #296 on: December 31, 2011, 07:41:39 PM »
http://screenshotcomparison.com/comparison/66418 (Exactly the same source and encoding settings with the only difference being the bit depth - the 10-bit video turned out 15% smaller, too)

PHAIL. What a case of "OMFG ooh wow my banding used to be 67% dark bands 33% light bands.... and it's now 33% light bands and 67% dark bands, in the exact same places and amounts... fap fap fap 10bit ruels!"

Btw, what encoders/releases are being compared here? And 15% smaller - is that 15% smaller than Mazui-tight 240-270mb, or Bloatgirls sizes, where you can easily shave off 40-70% and still get the same quality?

Offline AceD

  • Member
  • Posts: 2665
    • Facebook
Re: Hi10P and 8-bit encodes
« Reply #297 on: December 31, 2011, 07:57:08 PM »
You respond with some stupid ass quote you made up, when all he did was answer the question you asked, and then you proceed to ask a question where the answer is blatantly in the quote your replying too...try reading it properly

i don't even....
« Last Edit: December 31, 2011, 08:05:54 PM by AceD »

Offline Daiz

  • Member
  • Posts: 297
  • 10-bit Librarian
    • Underwater
Re: 10-bit and 8-bit encodes
« Reply #298 on: December 31, 2011, 08:17:53 PM »
PHAIL. What a case of "OMFG ooh wow my banding used to be 67% dark bands 33% light bands.... and it's now 33% light bands and 67% dark bands, in the exact same places and amounts... fap fap fap 10bit ruels!"

Are you blind or have a shit display or something? There's a pretty notable difference in banding, especially in the first and fifth shots.

Btw, what encoders/releases are being compared here? And 15% smaller - is that 15% smaller than Mazui-tight 240-270mb, or Bloatgirls sizes, where you can easily shave off 40-70% and still get the same quality?

Both are my own encodes, with the 8-bit encode being ~153 MB in size and 10-bit encode being ~131 MB in size (both video only). For a somewhat less direct comparison, an 8-bit release encode (with different filtering) I did for the same episode was 265 MB (with audio).

Offline Aadieu

  • Member
  • Posts: 103
Re: Hi10P and 8-bit encodes
« Reply #299 on: December 31, 2011, 08:29:38 PM »
PHAIL. What a case of "OMFG ooh wow my banding used to be 67% dark bands 33% light bands.... and it's now 33% light bands and 67% dark bands, in the exact same places and amounts... fap fap fap 10bit ruels!"

Are you blind or have a shit display or something? There's a pretty notable difference in banding, especially in the first and fifth shots.

Btw, what encoders/releases are being compared here? And 15% smaller - is that 15% smaller than Mazui-tight 240-270mb, or Bloatgirls sizes, where you can easily shave off 40-70% and still get the same quality?

Both are my own encodes, with the 8-bit encode being ~153 MB in size and 10-bit encode being ~131 MB in size (both video only). For a somewhat less direct comparison, an 8-bit release encode (with different filtering) I did for the same episode was 265 MB (with audio).

Thanks.

Don't you think 131 vs. 153 MB video streams (btw, that for a full ep or ordered chapters) is rather bitrate-starved to be making comparisons? Though, admittedly, both look pretty crisp.

Already good vision raised to perfection via contacts, 46" FullHD display. Nope, I see the same amount of banding, in the same places. Different-looking banding, yeah, one has more prominent dark strips in the banding, while the other has more prominent light strips in the banding... but the same locations and quantity of banding nonetheless.

You respond with some stupid ass quote you made up, when all he did was answer the question you asked, and then you proceed to ask a question where the answer is blatantly in the quote your replying too...try reading it properly

i don't even....


Sarcasm... is the autism getting in the way of comprehension?

There is no answer there. Unless 8-bit and 10-bit are encoder groups, don't think so lol. Neither have I ever seen anything tagged [Daiz]. So, unless this person is affiliated with some major fansubbing or RAW group or is somehow else a well-known authority, what results *he* personally got from comparing his own 8-bit and 10-bit encodes - rather similar levels of quality, minute differences in banding type but not quantity, purported 15% size reduction but no info on filesizes [at the time the comment was first written] - don't say much of anything. Plus, it's screens that he himself selected to emphasize the difference, presumably best seen in these very scenes. And it's not really visible even then.
« Last Edit: December 31, 2011, 08:31:11 PM by Aadieu »