Lib-Ray video standard: using Google/On2's VP8 video codec
Linked by David Adams on Fri 20th Apr 2012 01:31 UTC, submitted by fsmag
Multimedia, AV "When I started working on a no-DRM, open-standards-based solution for distributing high-definition video on fixed media ('Lib-Ray'), I naturally thought of Theora, because it was developed as a free software project. Several people have suggested, though, that the VP8 codec would be a better fit for my application. This month, I've finally gotten the necessary vpxtools and mkvtoolnix packages installed on my Debian system, and so I'm having a first-look at VP8. The results are very promising, though the tools are somewhat finicky."
E-mail Print r 5   30 Comment(s) http://osne.ws/jy1
Thread beginning with comment 515060
To view parent comment, click here.
To read all comments associated with this story, please click here.
WereCatf
Member since:
2006-02-15

Wow. Thanks for the interest! And thanks to WereCatf for letting me know about this thread.


You're welcome. Good to see that you're not being discouraged easily.

I'm researching that, however, and plan to have a proposal finished in early May.


I have to keep that in mind, I would love to see the finished spec. Though, I do not promise not to point out any flaws if I see any ;)

I'm currently working on the subtitle format (going to be SRT now, in separate files)


This I want some more clarification on: is there some specific reason for those files to be separate instead of inside the Matroska-container? I personally always place cover-images, cast-details, description of the title, all the various subtitles etc. inside Matroska-containers.

Also, do you plan to support italics, bold and non-white text in subtitles? Those are all very useful properties and I feel it would be a rather important shortcoming if the spec didn't support those. The issue, though, is that such properties inside an .SRT would be non-standard and thus those wouldn't show up properly in e.g. VLC Player. Perhaps it might be worth it to research possible alternatives to .SRT?

Of course, I do not expect to put Blu-Ray or Sony out of business with this project. I doubt we'll ever have the volume to make practical embedded players for sale in your local discount department store (not even going to try to beat $75 Blu-Ray players for the mass market).


Before the is finished there really is no point in worrying about such at all, but producing a small no-hassle media player+emulator-gaming console that also happens to support Lib-Ray wouldn't be terribly far-fetched as long as it's actually polished. I have actually been laying out various kinds of plans for such hardware and software for some time now, but since there is no Kickstarter-like service here I haven't been able to actually do anything about it.

Currently, it looks like a minimal "Lib-Ray player" would be an HTPC in the $500 range, but I need to fine that down a bit.


A minimal device capable of playing Lib-Ray would be closer to $60, tbh, if we include the costs of producing a simple plastic container to hold the electronics inside. E.g. Raspberry Pi would hardware-wise be perfectly capable of doing that up to 1080p resolution as the SoC does support VP8-decoding, the only limiting factor is the drivers that do not at the moment have support for that. Ie. if you really wanted a minimal device you'd just need to create something similar to RPi with a SoC that supports VP8-decoding in H/W and make certain that such capabilities are exported through GStreamer and/or OpenMAX.

$500 for a player is definitely way, WAY too much and if you're aiming for such a price-tag you may just shoot yourself in the foot now and save the hassle.

Cheaper systems based on SoC devices intended for the mobile market might be available before long -- it looks like maybe ARM Cortex 9 quad cores or NVIDIA Tegra 3 might be able to handle it.


It seems you're not entirely up to snuff when it comes to SoC - systems. I do not mean that as an insult, I am merely saying that you do not seem to fully understand what even a single-core SoC is capable of nowadays. A quad-core Tegra3 would definitely be an overkill if you merely want a media-player. Of course, if you intend the system to be useable as a desktop PC, too, then it might make sense but then you'd be targeting an entirely different market sector altogether.

Reply Parent Score: 3

Digitante Member since:
2012-04-21

"I'm currently working on the subtitle format (going to be SRT now, in separate files)


This I want some more clarification on: is there some specific reason for those files to be separate instead of inside the Matroska-container? "

Easy-patching. I mentioned that in the SDHC article.
Also -- transparency to the end user.

I personally always place cover-images, cast-details, description of the title, all the various subtitles etc. inside Matroska-containers.


It's true that Matroska supports that, but it makes it more opaque to the end user.

Also, do you plan to support italics, bold and non-white text in subtitles?


Those are the advantages of the Kate ("OggKate") format that I originally specified. In fact, it can handle vector graphics as well.

However, SRT is a far more popular format, and you're the first person who's seen the benefit of the extra features in Kate that I have heard from (though admittedly, that has been a small sample so far).

Will give it further thought.

E.g. Raspberry Pi would hardware-wise be perfectly capable of doing that up to 1080p resolution as the SoC does support VP8-decoding, the only limiting factor is the drivers that do not at the moment have support for that.


Interesting claim. Can you back it up?

The queries I made before in tech forums led me to believe this was not a realistic expectation. In fact, it appeared that no VP8 hardware support existed in any SoC that could also handle the 1920x1080x30fps 12-24mbps video through-put.

However, I'd be very happy to see it work. A player based on a low-cost open hardware platform like RPi would be extremely attractive.

I suspect writing the support code for it is out of my depth, however. I would need to find someone capable of doing that.

Reply Parent Score: 2

WereCatf Member since:
2006-02-15

However, SRT is a far more popular format, and you're the first person who's seen the benefit of the extra features in Kate that I have heard from (though admittedly, that has been a small sample so far).

Will give it further thought.


It just happens that in most cases such functionality isn't taken advantage of in any meaningful ways, but I have also seen cases where they have been used to atleast some advantage. For example, hearing-impaired subtitling could present sound-effects and such with different characteristics than speech, thereby making it easier to follow and clearer.

Interesting claim. Can you back it up?

The queries I made before in tech forums led me to believe this was not a realistic expectation. In fact, it appeared that no VP8 hardware support existed in any SoC that could also handle the 1920x1080x30fps 12-24mbps video through-put.


http://wiki.webmproject.org/hardware/arm-socs lists some of the SoCs that are already shipping and which support VP8. There's more of them coming in the future, and I can't remember which manufacturer it was, but someone was claiming their next chip would do 4k resolution decoding. Ie. there's quite good support now and it's only getting better in the near future.

RPi uses the Broadcom chip mentioned on that list. It does not mention the maximum bandwidth the SoCs can handle since such details are usually only given after you've signed NDA, but atleast in the case of BCM2835 I think I saw a Broadcom engineer claim it can do up to 24mbps. (I could remember wrong so don't take it as a fact, but that's how I remember it. You could possibly just try and ask them if you're interested.)

However, I'd be very happy to see it work. A player based on a low-cost open hardware platform like RPi would be extremely attractive.


The SoC is not open hardware, none of them are, so that could pose a problem. I am not aware of a single good-quality SoC that wouldn't require you signing NDA in order to gain full programming information and tools.

I suspect writing the support code for it is out of my depth, however. I would need to find someone capable of doing that.


I doubt that would be an issue once you settle on the spec, there's always someone out there to aid. I have been doing some coding here and there, including a few kernels and such, but I'm really rusty these days and I'm better at just bitching about things and trying to appear smarter than I actually am, so I wouldn't likely be able to aid much.

Reply Parent Score: 3

Digitante Member since:
2012-04-21

E.g. Raspberry Pi would hardware-wise be perfectly capable of doing that up to 1080p resolution as the SoC does support VP8-decoding, the only limiting factor is the drivers that do not at the moment have support for that.


This doesn't necessarily refute what you say, but this is what it says on the Raspberry Pi wiki at e-Linux:


GPU

The RaspberryPi appears to handle h264 1080p movie from USB to HDMI at least 4MB/s.

The Admin "JamesH" said it would handle "basically 1080p30, high profile, >40Mb/s." (5MB/s) in h264

And about WVGA(480p30) or 720p20 in VP8/WEBM


Based on that, it can't do it. But I guess you mean this can be fixed with software? Put another way, could you do it?

Reply Parent Score: 2

WereCatf Member since:
2006-02-15

Based on that, it can't do it. But I guess you mean this can be fixed with software? Put another way, could you do it?


I do not know if the people behind RPi have managed to fix the issue, but when I was still lurking around the forums the issue with VP8 decoding was that it was not implemented; it was all done in software. I assume that that 720p20 mentioned there is based on software-decoding.

EDIT: I cannot find any suitable WebM-video to download anywhere to test on my ARM-devices. Urgh.

Edited 2012-04-21 20:55 UTC

Reply Parent Score: 2

Alfman Member since:
2011-01-28

WereCatf,

"A minimal device capable of playing Lib-Ray would be closer to $60, tbh..."

You seem to know alot about this, so I'm not trying to doubt your figure but I'm currently looking for "living room suitable hardware" capable of running a linux media center and I haven't found anything for less than $200 and then some, and that's for a very low end system.

The main thing I'm looking to do is stream video over the network and hook up some kind of remote.

The only stuff I've found in the sub-hundred dollar range is mass produced proprietary devices. I could probably buy something and jailbreak it, but I don't like *having* to jailbreak it and voiding the warranty (* However I'll listen to anyone's suggests for this too). The Rasberry Pi is intriguing but it's neither proven nor available yet.

I'd earnestly like to hear what else were you could be referring to under a $100 price point that has similar capabilities?

Edited 2012-04-22 03:58 UTC

Reply Parent Score: 2

WereCatf Member since:
2006-02-15

I have also been in the market for a suitable HTPC based on ARM-architechture, but more-or-less all the currently-shipping ones are either aimed for tinkerers or try to pack too many things in the package, making it too expensive. RPi would almost work for me as it is supported these days by XBMC media center, but the lack of optical audio output is the thing that breaks the thing.

That said, I do not know of any currently-shipping ARM HTPC below the $100 price point other than the RPi. I am merely saying that it is entirely possible to create one, no one just seems all that interested in doing that.

As such I have only three recommendations if you want a HTPC on the cheap: Apple TV 2 (can only do 720p, not 1080p), RPi (will take a few more months to get), or wait for someone to realize there is demand for such and to start manufacturing something. Apple TV 3 can do 1080p, but I do not know if it has been jailbroken yet or if XBMC even has support for it.

I have actually tried to e-mail various manufacturers and hint that there is interest in sub-$100 HTPC-devices, but none of them have so much as bothered to respond at all. I can't start manufacturing such myself either because I do not have the required legal expertise or financial resources -- yet another reason why Kickstarter available here in Europe would be great.

Reply Parent Score: 3

Read more of this thread... >