RMVB: Why not use it?

[checkers]

I'm too lazy to format a whole new post, so I've stolen my layout from here

Well, there have been some interesting posts in this thread with interesting claims that I wanted to test. The argument comes down to how they compare, using the best tools available. It's no good having some mythical Real encoder that can do 25mb encoding if you don't actually have access to it.
One RV10 encoder I did have acess to however is Real Producer Basic. It's the Real Thing from Real Networks (), the only thing it doesn't let you do is change the predefined profiles. To work around that, I found a profile that used the highest possible quality settings at a lowish bitrate (350kbits), and then encoded the x264 to the same bitrate. No audio encoding was included.

The source:
The file was Honey & Clover II HD - 01 [Solar-IDE][9304C110].mkv, a 720p 24fps "good-enough-for-what-I-need" source which was noise free and with no banding. It's still not licenced, right? Passed it through an AVS script to make it a little more compressible:
directshowsource("Honey & Clover II HD - 01 [Solar-IDE][9304C110].mkv")
killaudio()
bilinearresize(480,272)
Dup(threshold=0.8,blend=true)
removegrain(2)

The x264 Encode:
x264.exe --pass 1 --bitrate 350 --keyint 999 --ref 16 --mixed-refs --no-fast-pskip --bframes 16 --b-pyramid --b-rdo --bime --weightb --direct auto --filter 1,1 --subme 7 --trellis 2 --analyse all --8x8dct --me umh --progress --cqmfile "m4g-lrm.cfg" --thread-input --threads 3 --output NUL "hctest.avs"
x264.exe --pass 2 --bitrate 350 --keyint 999 --ref 16 --mixed-refs --no-fast-pskip --bframes 16 --b-pyramid --b-rdo --bime --weightb --direct auto --filter 1,1 --subme 7 --trellis 2 --analyse all --8x8dct --me umh --progress --cqmfile "m4g-lrm.cfg" --thread-input --threads 3 --output "hctest.mkv" "hctest.avs"
pause
Nothing special here - options were reasonably normal. Special interest is the ...'interesting' GOP size and that's about it. The matrix is sourced from here.

The RV10 Encode:
Done through Real Producer Basic. I used the 350kbits 'download' profile. Nothing interesting otherwise. I enabled 2pass.

The Results
Here is the first 10 seconds of both full encodes. The screenshot is frame #50.

rmvb (full encode)

h264 (full encode)

Of note is the size difference, there's obviously large differences to RC under the hood.

I think the first 10 seconds give a very good indication of the rest of the thing, if you reeeally reeeally need it I will post the whole things.

EDIT: I forgot
The Conclusion
China sucks.

EDIT2:
h264 on the left, RMVB on the right.
Not the same frame... I'm lazy.
Notice the funny colour in the RV encode.
Through this entire sequence the longer it goes on, the blockier and blockier the RV10 becomes. h264 stays nice and consistant quality, not to mention no smudging in sight!

[KonW]

Originally Posted by LS5

RMVB is a container, not a codec. You're talking about the codec, RV10, here. File size is also completely controllable by the encoder, and is not codec or resolution dependent.


Quality differences can be traced back to raws and the skills of the encoder (which includes choices such as resizing).

I believe if they use the RV10..,,at 170 MB they'll have a much betther better better quality one . don't you think so?

[TheFluff]

Originally Posted by KonW

I believe if they use the RV10..,,at 170 MB they'll have a much betther better better quality one . don't you think so?

No, I don't, because while RV10 is probably slightly better than XviD, it's definitely not better than H.264.

Edit: you would do a lot better if you actually explained what the heck you're talking about. 170 MB RV is better than what, exactly?

[ArchMageZeratuL]

The bottom line is that RV10 has an unjustified fame of being very good at low bitrates. As TheFluff said above, it might be better than XviD (I haven't really compared the two), but it's definitely NOT better than H.264 at any bitrates. You want a 50 MB ep? H.264 is better. 100 MB? H.264. 150 MB? H.264. 350? Well, I guess that at 350 MB/ep, anything will look good >_>.

I suspect that the whole thing started from the fact that Real codecs are commonly used for low bitrate... so someone tried it for 100 MB/ep (or whatever they use) and thought it looked great?

[TheFluff]

Also:

Originally Posted by KonW

MORE COLORFUL

tweak(sat=2.0)

is my answer to that.

[epic59]

Originally Posted by KonW

http://share.dmhy.net/

oh my..... 178mb hdtv rips.... those have to look............. winderful...

[LS5]

Originally Posted by ArchMageZeratuL

I suspect that the whole thing started from the fact that Real codecs are commonly used for low bitrate... so someone tried it for 100 MB/ep (or whatever they use) and thought it looked great?

This was probably before H.264 even existed, at which point RealVideo most likely was the best alternative for low bitrate stuff.

Also, seems like the RealVideo format we're referring to isn't RV10 after all, but rather RV40.

RealVideo continued to use H.263 until RealVideo 8, when the company switched to a proprietary video codec. RealVideo codecs are identified by four character codes. RV10 and RV20 are the H.263-based codecs. RV30 and RV40 are RealNetworks' proprietary formats. These identifiers have been the source of some confusion, as people may assume that RV10 is RealVideo version 10, when it is actually the first version of RealVideo. RealVideo 10 uses RV40.

Silly RealNetworks.

[Nicholi]

Originally Posted by KonW

then how to explain the truth that the English subbed one I downloaded is bigger
but with a lesser quality?
Because the English one's subber has a worse RAW?Or the English one's subber's computer or badly set specifications?In my opinion ,unless they use RMVB they won't provide one clear and colorful as the one i posted at a size about 100MB.

Just my personal opinion here in question concerning SHS's fansub of Darker Than Black. Probably the worst sample to test in comparisons against the "almighty" Real codec. Unfortunately since you seem to be new to encoding, not every encoder just takes the source (the RAW) and shoves it into encoder X. Some of them do what is known as filtering, which alters the video from the original. In this case I would contend that SHS's version is completely smoothed out (no fine details remain whatsoever) as well as leaving it slightly blurred in order for them to hit a lower bitrate. Not to mention, Yes, quality can vary quite a bit from encoder to encoder as well as from RAW to RAW. I recently just saw some idiots encoding to PC levels instead of TV. Are we in Sparta? Is this madness?

Originally Posted by KonW

(the avi one is subbed by Shinsen,which i chose randomly,171MB(704x396),the first one 107MB(960x540))
If you download the episode10 Darker Than Black that subbed by thre group KPDM(this is where the first picture from),it will be more easy to find that it is clear that the Rmvb one is much better and clear than the avi one.

Also, Darker Than Black just aired episode 8 in Japan. So unless you just slipped up while trolling or are inherently stupid, I've no idea what this ep10 mumbo jumbo is about.

Gendo: KonW, the truth is... Real Media is the last choice anyone will make because it blurs the video (and we like our video shiny and sharp without blocks).

KonW: You liar!
BANG!

[Oberon]

Originally Posted by KonW

the truth is ,the RMVB is more clear and colorful during the whole playing,and certainly i as well as anyone will get much fun watching the RMVB one,and despite with a smaller file size it is at a higher resolution ,how wonderful,,.
So is there anything more to care about?A smaller size A brighter and more clear sight,,,so that's simple,RMVB.

Are you sure though that both files were made from the same source file? There are different raw-providers with VERY varying quality in some cases. There can be tons of reasons, why the non-rmvb screenshot looks the way it does, starting from the quality of the raw up to the settings and filters used by the encoders, all of which have not necessarily to do with the quality of the codecs themselves. That's why it's not valid, to draw conclusions from two random screenshots. You're comparing apples and oranges. To make a valid point, you'd have to take at least a raw and then encode it with the same quality settings, resolution, whatever, once to rmvb and once to h264 and then compare how those files turned out.

EDIT: I type too slowly.

[jfs]

To close the discussion (as if...) I'l just remind people that there is a "pseudo-free" solution for playing and encoding Real Media formats, Helix DNA. I don't know how well it works nor whether it's even licensed for use on Windows platforms, but it is an option.

[Nicholi]

Originally Posted by jfs

To close the discussion (as if...) I'l just remind people that there is a "pseudo-free" solution for playing and encoding Real Media formats, Helix DNA. I don't know how well it works nor whether it's even licensed for use on Windows platforms, but it is an option.

I've used it before, but only for mild testing purposes to make some samples of Real stuff. Not comparing quality. Once you start encoding in Real though (if you've ever used anything else from MPEG) you see how extremely limited you are in even what you can tell the encoder to do. And I'm not referring to "there aren't options everywhere to tick". I mean even basics things you have almost no control over. Especially with this wonky profile thing they use. And again just to re-emphasize whatever i said 6+ months ago. Real audio is still CBR only. Truly, what the deuce?

[jpwong]

Originally Posted by Skane

From what I understand, most fansubbers follow a self-imposed filesize standard instead of actually compressing the video to its maximum possible compression without sacrificing too much video quality. So comparing the average filesize of current h.264-based files to .rmvb files is kinda meaningless, since they are not following the same "standards".

That's pretty much true. I believe most groups currently target 170MB/230MB since those sizes are close to the older CD standard of 175/233 but waste less space on DVDs.

Anyway, I believe it's been mentioned before, but I think a large part of the whole reason next to no one in the english fansub world would even consider touching real media anythings is the stigma that's attached to it from previous eras where Real Video was to be quite blunt, crap.

First impressions are everything in this world now. It's the same reason there's so much resistance to fluorecent lights in the US right now. Even though all the problems that early versions had are now long gone, that first impression with the lightbulb has created a lasting image of "bad product" in a lot of people's minds.

[False Dawn]

Originally Posted by LS5

The fact that you're unable to find a better quality encode at 100MB doesn't surprise me, really, because practically no English groups release at that file size. >_>

Not quite true. I remember Froth-Bite releasing Ray in mp4 under 100MB, though I don't know what file type they used for that. Decent-enough quality though.

[Skane]

[Froth-Bite] mainly uses .MP4, which uses h.264. They are one of the few groups that have broken away from the 170+ MB mentality, BUT, they still follow a self-imposed standardised filesize these days. Around 140+ MB.

[gg] is another group that released sub-100 MB files, for the series, "Night Head Genesis". I was pretty skeptical when I first saw it, but seeing is believing, and NGH does contain some very pretty graphics.

Cheers.

[cyth]

Originally Posted by Skane

[gg] is another group that released sub-100 MB files, for the series, "Night Head Genesis". I was pretty skeptical when I first saw it, but seeing is believing, and NGH does contain some very pretty graphics.

Night Head Genesis doesn't really have bad art or animation, it's somewhat visually appealing. For some reason, certain shows compress very well and NHG is one of those shows. I remember Samurai Champloo compressing very well, and that one had high motion scenes everywhere. Anyway, NHG has been discussed before because of its astounding compressibility, so nothing new there.

[KonW]

Originally Posted by Nicholi

Just my personal opinion here in question concerning SHS's fansub of Darker Than Black. Probably the worst sample to test in comparisons against the "almighty" Real codec. Unfortunately since you seem to be new to encoding, not every encoder just takes the source (the RAW) and shoves it into encoder X. Some of them do what is known as filtering, which alters the video from the original. In this case I would contend that SHS's version is completely smoothed out (no fine details remain whatsoever) as well as leaving it slightly blurred in order for them to hit a lower bitrate. Not to mention, Yes, quality can vary quite a bit from encoder to encoder as well as from RAW to RAW. I recently just saw some idiots encoding to PC levels instead of TV. Are we in Sparta? Is this madness?

-----------------
Also, Darker Than Black just aired episode 8 in Japan. So unless you just slipped up while trolling or are inherently stupid, I've no idea what this ep10 mumbo jumbo is about.
--------
Gendo: KonW, the truth is... Real Media is the last choice anyone will make because it blurs the video (and we like our video shiny and sharp without blocks).
BANG!

As a matter of fact,in anime you'll find 90% of the video is static for which the using of a high bitrate is quite a waste of space,that's the perfect time that the Real comes in,by using a low bitrate at static part and high at parts with heated motions,we get a small file with a high resolution and also good quality in video..
with the extension RMVB.and this can hardly be done by any other avis(suppose there's no H.264 existed),,,,that's the reason we have a larger file with a worse quality and a smaller file with a better satisfaction if you watch it,,,
-------
the picture i posted is from the KPDM subgroup(Darker than Black 07)
and since it's already been several week ago(no seeds now), if you want to download one subbed by KPDM to check the quality by BT,you must choose the one that will be aired later....the number 10 actually means nothing, i just chose a number larger than 08 randomly,,,,,

-----------
at least for anime and most of the films ,Rmvb won't disappoint you,,it won't blurs the video even sometimes it is at a lower bitrate(because at that time the video is at a lower frame rate,)...

[KonW]

Originally Posted by jpwong

That's pretty much true. I believe most groups currently target 170MB/230MB since those sizes are close to the older CD standard of 175/233 but waste less space on DVDs.

Anyway, I believe it's been mentioned before, but I think a large part of the whole reason next to no one in the english fansub world would even consider touching real media anythings is the stigma that's attached to it from previous eras where Real Video was to be quite blunt, crap.

First impressions are everything in this world now. It's the same reason there's so much resistance to fluorecent lights in the US right now. Even though all the problems that early versions had are now long gone, that first impression with the lightbulb has created a lasting image of "bad product" in a lot of people's minds.

well ,,that's sad,,
smaller size will mean shorter download time,,,,,

I'm too lazy to format a whole new post, so I've stolen my layout from here

Well, there have been some interesting posts in this thread with interesting claims that I wanted to test. The argument comes down to how they compare, using the best tools available. It's no good having some mythical Real encoder that can do 25mb encoding if you don't actually have access to it.
One RV10 encoder I did have acess to however is Real Producer Basic. It's the Real Thing from Real Networks (), the only thing it doesn't let you do is change the predefined profiles. To work around that, I found a profile that used the highest possible quality settings at a lowish bitrate (350kbits), and then encoded the x264 to the same bitrate. No audio encoding was included.

The source:
The file was Honey & Clover II HD - 01 [Solar-IDE][9304C110].mkv, a 720p 24fps "good-enough-for-what-I-need" source which was noise free and with no banding. It's still not licenced, right? Passed it through an AVS script to make it a little more compressible:
directshowsource("Honey & Clover II HD - 01 [Solar-IDE][9304C110].mkv")
killaudio()
bilinearresize(480,272)
Dup(threshold=0.8,blend=true)
removegrain(2)

Actually i can't see much difference,,,,

i want to know the size comparison between the 2,it would be great if the H264 can give a fille as small as the RV does,

[kivine]

xvid/divx vs. RV . RV typically wins.

h264/x264 vs. RV. H264 typically wins

the only problem i see with RV is that there is always some smudging or over smoothen video

[ffdshow]

I did some tests months before, the conclusion was:

- most chn releases are done in 1 pass encode to save time, so its pretty worthless to compare
- with some changes in setting, rv is actually quite good on low bitrate
- the video quality is close to normal xvid but not really better, 90Mb rv is like 160Mb Xvid, but I see some frames dropped
- In my opinion, rv is not good enough on high quality release, its not worthy to use more MBs for quality either. Though, I can understand why chn groups has no real success switching to xvid/h264.

RV isn't bad.

[Nicholi]

RV isn't bad.
It's just worse. lol