Older blog entries for rillian (starting at number 112)

Today I tried installing the experimental R600/700 3D drivers for Fedora 12. Surprisingly, nothing much broke. And now I have compiz, yay! Like the clutter gdm screen, it's a bit too slow though.

Unfortunately, blender seems to be slower than it was under software rendering. sigh.

In the Future, we will embed machine images in our Ogg files, which, when booted and given network access to the other multiplexed data will decode, render, and export the results data in a variety of JSON responses.

Cute demonstration of chrominance vs luminance sensitivity in the human visual system. With source code!

State of theora

Monty has posted another of his excellent technical updates on the theora encoder rewrite which has been ongoing for more than a year now. It's a good summary of what went into the recent beta releases.

We still have a bit to go before the final 1.1 though; the new rate control still behaves badly on a some clips.

The complete set of status reports:

3 Apr 2009 (updated 3 Apr 2009 at 23:17 UTC) »

State of theora

Yesterday Monty has posted another status update on the new theora encoder. It's a good summary of what went into the recent 1.1 alpha release, and what there is still to do.

The complete set of status reports:

$ history | awk '{a[$2]++}END{for(i in a){print a[i] " " i}}' | sort -rn | head
60 ls
59 cd
52 svn
44 nano
40 make
26 less
26 fgrep
14 wget
11 host
10 rm

Yeah, pretty much.

Commentary commentary

Lawrence Lessig recently posted a summary on the fsfe-uk list about problems that were blocking Theora adoption. Here's a rebuttal, for what it's worth.

His first point, that Theora isn't technically competitive with the lastest batch of encoders for the encumbered MPEG codecs, is entirely true. From Xiph's point of view is that that's a little like saying there's no point in using Linux because it doesn't work as well as Windows, but the technical disparity does need to be addressed.

Monty and Derf have been working on a new encoder the past few months, but there's nothing to show yet. We believe the Theora format has scope to offer similar compression efficiency to h.263 with less complexity. Beyond that, we look to the BBC's Dirac. But in the absence of software to prove the capabilities of the format, one has to take our word, as well as being interested in long term planning, for that to be a meaningful argument.

His second point, about people believing Theora is patented is just FUD as far as I know. We're not aware of any patents. The original developer of the VP3 format which became theora grants rights to any patents they might have on the implementation. Submarine patents are of course always possible, but they affect MPEG and Microsoft codecs just as much as independent designs.

I've heard this argument from two different directions. First from corporations who have already bet on one of the MPEG codecs and want to dissuade any competition, and second from Free Software people, who don't understand how patents work, shrug, say it's all equally bad, and then get behind the proprietary technology.

What happened with the html5 flame fest was that some corporations said they didn't feel the current demand for web content in royalty-free formats justified the additional exposure implementing them would create. That's true so far as it goes, but a very specific statement about their own interests and hardly a reason for anyone else to eschew royalty free formats.

There was a lot of talk at the recent FOMS meeting about how to address the FUD issue and educate the free community about patents. Hopefully some public documentation will come of it. It's been quite difficult to find legal counsel who understands the FLOSS development model well enough to toss ALL the traditional wisdom about patent risk out the window: namely to never do or say anything at all.

All that said, I completely agree with the recommendation that we get people talking. It can only help. Free software can't compete with the installed base of flash video at this point, but we should all be working to offer an alternative for those who can use it, and prepare the toolchain so we can provide the greatest support for software and creative freedom in the next round of web video.

The planet aggregator needs an advogato style trust metric for deciding which feeds to include. Solves that nasty maintainer bottleneck problem.

Ogg podcast feed for Cory Doctorow's reading of Bruce Sterling's The Hacker Crackdown about events leading to the formation of the EFF.


Video and Software Patents

There's an article on linux.com ostensibly about the current state of the Theora reference encoder, and ongoing efforts to improve it.

It's a strange article though. It's misleading to say "no new MPEG-4 licenses are available" when talking about codecs, when the reference is that the systems patent suite is currently unavailable during negotiations. MPEG-4 visual and AVC suites are still available under the normal terms.

And then there's a long, verbatim quote from one of the XviD developers. I'm always interested to hear the attitudes behind such projects expressed clearly, and in this case the thinking seems to be that software patents are evil, he doesn't know anything about them, so everyone should ignore the issue completely because everything potentially infringes. Wow.

There's a huge difference between vague assertions that Linux in general must be infringing hundreds of (unnamed) Microsoft patents, and a very specific list of patents considered by their holders to be required to implement a specific specification. Pretending we can't make distinctions here is just FUD.

I complained about this to a colleague and he said, "Well, it's not a rah-rah theora article, if that's what you wanted." But actually it is. After describing the patent issue as unimportant, it then goes on to say how theora is "good enough" using mp3 as an example. But the codec that's most analogous to mp3 is h.264, not Theora. They are both patented. They both feel like "good quality" at file sizes that are convenient for use. And there is a huge push on to standardize on h.264 (aka MPEG-4 AVC) as a standard codec for from hardware vendors, software vendors and the file-sharing community. If software patents don't matter, and h.264 is technically superior, why does the article recommend theora at all?

The article does acknowledge that software patents can't be ignored is some other jurisdictions (the Xvid developer lives in Germany) and I actually agree that's the reason to use theora over h.264. But it's a big reason, one that isn't going away. And while the mp3 patents start expiring in the next five years, h.264 will be encumbered for a long time to come.

The last section is spot on though. Free formats will succeed by having excellent support in all tools, not by being free. We need to improve the reference implementation, and improve tool support if we want to get somewhere.

103 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!