Back before psychopharmaceuticals, schizophrenia and depression were thought to be very similar or even the same thing. It was only once we had Thorazine (first antipsychotic) and then later tricyclics (first antidepressents) that in the clinical settings schizophrenia and depression began to be sorted much more distinctly, essentially based on the kinds of patients that got better with antipsychotics versus those that got better with antidepressants.
It's pretty common for diagnostic definitions to align with successful treatment methadologies, since "what will help" is the fundamental answer that diagnosis hopes to lead to.
Sounds like we're now getting back to the perspective of a half-century ago.
The Zune HD features 720p on a tiny, tiny, screen...or at home, where I can watch TV on a computer, console, or cable box.
It's not that small a screen for a portable media player, and it'd OLED, not another lousy 24-bit LCDs with horrible banding.
For myself, there's an important third place I'm often in: hotel rooms. Which increasingly have nice HD TVs cursed with only lousy 4:3 SD coax analog inputs getting stretched to 16:9....and an HDMI input (universal in Marriotts now, and in other chains as well).
Being able to carrying around decent HD programming to hook into those sets should be a lovely thing, and free with Media Center recording HD OTA.
Never saw ANY of the benefits/Ultimate Content that was promised.
"Benefits" are a subjective concept, but the actual content is easy to find.
Control Panel > Secuirty > Windows Update
There will be a blue box listing any available Windows Ultimate Extras you don't have installed there.
IIRC, there's Dream Scene (animated backgrounds) and a bunch of content packs, extra Sound Schemes, and the Hold Em Poker and Tinker games. Tinker is pretty cool.
The biggest benefits are having both domain support and Media Center, some Bitlocker stuff, and being able to have lots of languages to switch Windows between.
Since I demo media playback on my work laptop, Domain + Media Center was the critical reason for me to have Ultimate.
As for "if it was good enough in the past..." rule, most web video gets reecoded every 18-24 months to take advantage of more efficient codecs to improve qulaity or reduce bitrates. For real businesses counting.
I recommend people don't focus on YouTube too much as an example of the web video industry. It's very much an anomly in both business model and particualr being subsidized by Google at such a sceale and getting access to Google's very cheap bandwidth. Also, Most YouTube clips aren't watched even a dozen time, so the cost of encoding time can be bigger than the cost of delivering the bits, so they don't tune their encodes for maximum efficiency, but for rapid transcoding.
YouTube does plenty of things that wouldn't make sense for anyone else. They're not really an example of much beyond YouTube.
It's not just about quality, it's about efficiency.
Assuming a threshold for "good enough" quality and buffering time, a less efficient codec means:
Users need 2x the bandwidth to have adequate buffering time
Bandwidth costs are 2x higher.
So, there can be a real impact in terms of reduced audience and increased costs.
Also, comparing Theora with H.264, I think 2x may actually be understating the diffrence.
I made these samples, comparing to Xiph's examples. As you can see, x264 was able to deliver quite a good experience with better per pixel quality, at 640x360 at the bitrate Theora struggled at 400x224 with.
I'd say current versions of Ogg Theora take 100-200% more bandwidth to deliver "good" quality as current versions of x264. Codecs converge at higher bitrates, and no doubt Theora is techically capable of good quality at a sufficiently higher bitrate. But it'll take a lot more bits to get there than other codcs. Theora's bitstream is based on VP3, which is over a decade old now, and we'd generally expect a refined vresion to come out as MPEG-4 part 2 efficiency (like Xvid/Divx without B-frames).
The past discussions were based on a relatively easy clip (Big Buck Bunny), and compared to YouTube encodes in H.264 Main Profile and H.263.
But if you compare what you could do at the same bitrate with a quality-tuned H.264 High Profile, H.264 can do a quite nice 640x360 at the bitrate Theora used for 400x224 AND with higher per-pixel quality. WMV with the current VC-1 implemenation also outperforms Theora (although not by as much as x264).
I made some samples (the.ogv files were made by Xiph, the others were encoded by me to similar specs):
Depends a lot on methadology and demographics. RIAStats shows quite a bit higher, sampling from a bigger range of machines.
However, even they're missing some stuff, as they show the Silverlight 1.0 installed base on MacPPC as 0%.
Sites that adopt Silverlight see their Sliverlight installed base rise very quickly of coures:).
The real question for any particular site is Silverlight Installed base + conversion rate; if they can get 80% of users who don't have Silverlight to isntall it, from a base of 30%, then they get 86% of their users with Silverlight.
It's perfectly natural to see projects switch technologies periodically. MLB has bounced between technologies for years and they may again based on their experiences this year.
Most of the big companies who have done Silverlight continue to do it, and we're certainly seeing plenty of media companies switching from Flash to Silverlight, and a bunch more once Silverlight 3 is launched with compatibility with F4V (Flash H.264) files.
And from a market share perspective all we need is one great Silverlight site for each user. It's not like someone needs to uninstall Flash to run Silverlight; it's not a zero-sum game.
I'm quite pleased with the current rate of adoption, myself. I'm obviously not going to announce new official numbers here, but there's plenty of sites that track these things that'll give you a sense of the velocity of install rate.
Being an open standard HTML5 is open for development of end-user controls, such as animate only while cursor hovers, sound off till I say so, etc.
If HTML5 enables equivalently rich media players to Flash and Silverlight, HTML5 will be rich enough for publishers to not enable end user controls they don't want enabled.
Or, to put it another way, if HTML5 makes it trivial for end users to skip the commercials in commercial-funded television, no one will be publishing commercial-funded television in HTML5.
There have been 400M downloads of Silverlight so far.
That's more than the total market share of Firefox + Safari + Chrome (+ Linux + Mac + iPhone + Android if you're thinking platforms). So Silverlight's already a bigger audience than every browser NOT IE running on Windows.
Popfly is hardly meant to be the big Silverlight install driver:). In the USA, the highest profile Silverlight projects have probably been Netflix and the Olympics (Beijing and soon Vancouver), with the Masters and NCAA March Madness as recent big ones.
Because so much of media is geoblocked, the big Silverlight drivers vary by market. Russia has RuTube, South Korea has all major broadcasters and the leading search provider. France just had their big Tennis tournament.
Given the original comparisons between Theora and YouTube's H.263 and H.264 encodes, I thought I'd do some samples showing what H.264 is capable of doing. YouTube doesn't use High Profile (no 8x8 blocks) nor CABAC entropy coding, and so leaves a lot of bits on the floor.
Note the third sample uses the high bitrate frame size and the low bitrate data rate, and still outperforms all the YouTube and Theora clips in video and audio quality.
in the Rate-Distorion graph, note they start the plot at 50 Kbps, so look at the actual numbers.
For example, at 40 dB, x264 needs 70 Kbps and Theora needs 120 Kbps.
The gap would be bigger with higher motion, more detail, longer content, and particular when there are buffer constraints. Also, x264 is (properly) tuned for perceptual quality more than strict PSNR accuracy.
Theora suffers from not being a very mature implementation (which Xiph is making great progress on addressing) and being a 90's era codec design (about which Xiph can't do anything without breaking compatibility). And other codecs are getting better as well; if Theroa is refined enough to be reasonably optimal in a year, it'll be competing against the improved H.264 and VC-1 codecs of a year from now, and H.265 not that far away.
There are all kinds of interesting things about Theora, but competitive low bitrate compression efficiency isn't going to be one of them.
Color me skeptical about the impartiality of a company named after one of Microsoft's key embrace-extend efforts, particularly when the issue is public access to media via non-proprietary codecs.
No, I work for Microsoft proper. But I'm speaking as a compression nerd here. Honestly, if Theora was a big sucess that'd be good for Silverlight; we have a codec extensibility model and Flash doesn't. But people seem to think that HTML5 and Theora are going to solve problems that the're not able, or even meant to, solve.
But there's certainly no reason in the world you should or would need to take my word for any of this. x264 and ffmpeg2theora would make a short project out of doing your own comparison.
Make sure to run the lastest Theora alpha 2, and use "-V" to turn on the slowest, highest quality mode.
"Better" is alwasy about context. Try a head-to-head for a scenario of interest to you.
So, Mozilla is the only one who can legally distribute the sources and binaries of Firefox*, under your system.
It's not MY system:). And Mozilla would have to ask them about what counts as a "product" in this context. It could be that a company distributing their own binaries based on Mozilla would need their own license or something. Or could link back to a Mozilla decoder lib or something.
Thanks for providing us with the objective point of view from Microsoft's evangelism department, Ben. Your support for Chris Dibona's point of view pretty much confirms for me he's flat wrong.
I'm actually not an evangelist anymore; I run strategy for Silverlight media technologies.
And as such, have done a lot of Theora encodes to see how interesting it is. It's be trivial to support as a managed decdoer in the the new Silverlight 3 Raw AV pipeline, so it's fine by me if it turns out to be great, but I think peope have very unrealistic expectations for it.
A quick pan isn't that informative without knowing anything about the rate control model they're using. 1 sec of hard video in a 15 sec buffer can be smoothed over.
Seeing as it takes more horsepower to decode H.264 than the older H.263/Sorenensen/Spark/whatever they were using before, I assume it's my aging PC that's at fault. But buying a new PC just to get 30FPS on YouTube is a non-starter, given that the existing one still does everthing else I need it to do at acceptable speed.
That's the big drawback of H.264 - decoder complexity. And YouTube's encodes aren't the hardest either; there's deeper features yet they aren't using.
One of the design goals for what's likely to become H.265 is a 25% improvements in compression efficiency WITH a 50% reduction in decoder complexity.
Still, what bitrate are you downloading? 500 Kbps should be okay even for that ancient box, if you at least have SSE2.
Where's your evidence? Why is Greg's example odd? Have you done a comparable experiment with a different video clip to justify your 2x claim, or more importantly, show that at the same bitrate Theora looks much worse on your clip?
And yes, I've done plenty of my own tests as well on Big Buck Bunny and many other clips.
Why would it, since it didn't at 499Kbps? Or are you claiming that Youtube uses a bad H.264 encoder? Or do you think that example is rigged?
YouTube doesn't use High Profile (no 8x8 blocks or adaptive quantization matricies) or CABAC entropy coding. So they're going to be at least 20% less efficient than the best encodes could be.
Also, they trimmed the clip before the really interesting high motion parts. Most of the shots in the section they did use were static camera, and as animation is noise-free. Toss a nice grainy movie trailer in there and Theora shows basis pattern left and right.
I wish he'd reported the Theora settings used in the encode.
An open-source browser cannot legally read h264 video, that is the real issue that people seem to have trouble to understand.
I keep hearing that, but I don't know why that would be so. MPEG-LA requires a fee for distribution of products. But Mozilla could pay the decoder cap fee (maxes out a $5M/year next year) and allow as many people to download a H.264-enabled Firefox as they want, no?
That is why the HTML standard only mandates a format that is not impaired by any legal restrictions: Theora.
HTML5 does not mandate any codec or format. Ogg with Vorbis and Theora were proposed, but not included in the current draft, due to concerns by (IIRC) Nokia and Apple.
Prove Microsoft's implementation is as secure as the one in Firefox, and I'll listen to you.
And how does one "prove" security again:)?
Anyway, given the WMP OCX component has been in the market and supported code for many years with a good security track record, it seems validation would be more needed for Firefox's brand-new beta code.
...up to a certainly size/bitrate, yes. But I don't know if it would be enough to fill the screen of 480x320 or VGA phone, particuarly at a high quality.
Slashdot Mirror
Interesting.
Back before psychopharmaceuticals, schizophrenia and depression were thought to be very similar or even the same thing. It was only once we had Thorazine (first antipsychotic) and then later tricyclics (first antidepressents) that in the clinical settings schizophrenia and depression began to be sorted much more distinctly, essentially based on the kinds of patients that got better with antipsychotics versus those that got better with antidepressants.
It's pretty common for diagnostic definitions to align with successful treatment methadologies, since "what will help" is the fundamental answer that diagnosis hopes to lead to.
Sounds like we're now getting back to the perspective of a half-century ago.
The Zune HD features 720p on a tiny, tiny, screen...or at home, where I can watch TV on a computer, console, or cable box.
It's not that small a screen for a portable media player, and it'd OLED, not another lousy 24-bit LCDs with horrible banding.
For myself, there's an important third place I'm often in: hotel rooms. Which increasingly have nice HD TVs cursed with only lousy 4:3 SD coax analog inputs getting stretched to 16:9. ...and an HDMI input (universal in Marriotts now, and in other chains as well).
Being able to carrying around decent HD programming to hook into those sets should be a lovely thing, and free with Media Center recording HD OTA.
Never saw ANY of the benefits/Ultimate Content that was promised.
"Benefits" are a subjective concept, but the actual content is easy to find.
Control Panel > Secuirty > Windows Update
There will be a blue box listing any available Windows Ultimate Extras you don't have installed there.
IIRC, there's Dream Scene (animated backgrounds) and a bunch of content packs, extra Sound Schemes, and the Hold Em Poker and Tinker games. Tinker is pretty cool.
The biggest benefits are having both domain support and Media Center, some Bitlocker stuff, and being able to have lots of languages to switch Windows between.
Since I demo media playback on my work laptop, Domain + Media Center was the critical reason for me to have Ultimate.
Of course, "normal" viewing distance is based on the distance where a SD image looked good.
Since 1080p has 6x the pixels of 480p, you'll need to sit 2.4x (square root of 6) closer to see a similar amount of visual detail.
The most important upgrade after getting a HD video source is pushing your couch closer :).
Theora is over a decade old; it definately can beat H.264 in software decoding.
But if you didn't see this:
http://cid-bee3c9ac9541c85b.skydrive.live.com/browse.aspx/.Public/BBB%7C_Compare
As for "if it was good enough in the past..." rule, most web video gets reecoded every 18-24 months to take advantage of more efficient codecs to improve qulaity or reduce bitrates. For real businesses counting.
I recommend people don't focus on YouTube too much as an example of the web video industry. It's very much an anomly in both business model and particualr being subsidized by Google at such a sceale and getting access to Google's very cheap bandwidth. Also, Most YouTube clips aren't watched even a dozen time, so the cost of encoding time can be bigger than the cost of delivering the bits, so they don't tune their encodes for maximum efficiency, but for rapid transcoding.
YouTube does plenty of things that wouldn't make sense for anyone else. They're not really an example of much beyond YouTube.
It's not just about quality, it's about efficiency.
Assuming a threshold for "good enough" quality and buffering time, a less efficient codec means:
Users need 2x the bandwidth to have adequate buffering time
Bandwidth costs are 2x higher.
So, there can be a real impact in terms of reduced audience and increased costs.
Also, comparing Theora with H.264, I think 2x may actually be understating the diffrence.
I made these samples, comparing to Xiph's examples. As you can see, x264 was able to deliver quite a good experience with better per pixel quality, at 640x360 at the bitrate Theora struggled at 400x224 with.
http://cid-bee3c9ac9541c85b.skydrive.live.com/browse.aspx/.Public/BBB%7C_Compare
I'd say current versions of Ogg Theora take 100-200% more bandwidth to deliver "good" quality as current versions of x264. Codecs converge at higher bitrates, and no doubt Theora is techically capable of good quality at a sufficiently higher bitrate. But it'll take a lot more bits to get there than other codcs. Theora's bitstream is based on VP3, which is over a decade old now, and we'd generally expect a refined vresion to come out as MPEG-4 part 2 efficiency (like Xvid/Divx without B-frames).
The past discussions were based on a relatively easy clip (Big Buck Bunny), and compared to YouTube encodes in H.264 Main Profile and H.263.
But if you compare what you could do at the same bitrate with a quality-tuned H.264 High Profile, H.264 can do a quite nice 640x360 at the bitrate Theora used for 400x224 AND with higher per-pixel quality. WMV with the current VC-1 implemenation also outperforms Theora (although not by as much as x264).
I made some samples (the .ogv files were made by Xiph, the others were encoded by me to similar specs):
http://cid-bee3c9ac9541c85b.skydrive.live.com/browse.aspx/.Public/BBB%7C_Compare
Depends a lot on methadology and demographics. RIAStats shows quite a bit higher, sampling from a bigger range of machines.
However, even they're missing some stuff, as they show the Silverlight 1.0 installed base on MacPPC as 0%.
Sites that adopt Silverlight see their Sliverlight installed base rise very quickly of coures :).
The real question for any particular site is Silverlight Installed base + conversion rate; if they can get 80% of users who don't have Silverlight to isntall it, from a base of 30%, then they get 86% of their users with Silverlight.
Why would you have to worry in the first place?
It's perfectly natural to see projects switch technologies periodically. MLB has bounced between technologies for years and they may again based on their experiences this year.
http://news.cnet.com/8301-1023_3-10215761-93.html
Most of the big companies who have done Silverlight continue to do it, and we're certainly seeing plenty of media companies switching from Flash to Silverlight, and a bunch more once Silverlight 3 is launched with compatibility with F4V (Flash H.264) files.
And from a market share perspective all we need is one great Silverlight site for each user. It's not like someone needs to uninstall Flash to run Silverlight; it's not a zero-sum game.
I'm quite pleased with the current rate of adoption, myself. I'm obviously not going to announce new official numbers here, but there's plenty of sites that track these things that'll give you a sense of the velocity of install rate.
Being an open standard HTML5 is open for development of end-user controls, such as animate only while cursor hovers, sound off till I say so, etc.
If HTML5 enables equivalently rich media players to Flash and Silverlight, HTML5 will be rich enough for publishers to not enable end user controls they don't want enabled.
Or, to put it another way, if HTML5 makes it trivial for end users to skip the commercials in commercial-funded television, no one will be publishing commercial-funded television in HTML5.
There have been 400M downloads of Silverlight so far.
That's more than the total market share of Firefox + Safari + Chrome (+ Linux + Mac + iPhone + Android if you're thinking platforms). So Silverlight's already a bigger audience than every browser NOT IE running on Windows.
Popfly is hardly meant to be the big Silverlight install driver :). In the USA, the highest profile Silverlight projects have probably been Netflix and the Olympics (Beijing and soon Vancouver), with the Masters and NCAA March Madness as recent big ones.
Because so much of media is geoblocked, the big Silverlight drivers vary by market. Russia has RuTube, South Korea has all major broadcasters and the leading search provider. France just had their big Tennis tournament.
Given the original comparisons between Theora and YouTube's H.263 and H.264 encodes, I thought I'd do some samples showing what H.264 is capable of doing. YouTube doesn't use High Profile (no 8x8 blocks) nor CABAC entropy coding, and so leaves a lot of bits on the floor.
Note the third sample uses the high bitrate frame size and the low bitrate data rate, and still outperforms all the YouTube and Theora clips in video and audio quality.
http://cid-bee3c9ac9541c85b.skydrive.live.com/browse.aspx/.Public/BBB%7C_Compare?lc=1033
This?
http://people.xiph.org/~greg/video/ytcompare/comparison.html
I read it, and have commented on it at length throughout this thread.
Basically the article briefly says that YouTube's H.264 is better than Theora, and then goes on at length showin how Theroa is better than H.263.
Xiph's own data shows H.264 has a big bitrate advantage at the same quality level even in a test that should favor Theora.
http://web.mit.edu/xiphmont/Public/theora/demo7.html
in the Rate-Distorion graph, note they start the plot at 50 Kbps, so look at the actual numbers.
For example, at 40 dB, x264 needs 70 Kbps and Theora needs 120 Kbps.
The gap would be bigger with higher motion, more detail, longer content, and particular when there are buffer constraints. Also, x264 is (properly) tuned for perceptual quality more than strict PSNR accuracy.
Theora suffers from not being a very mature implementation (which Xiph is making great progress on addressing) and being a 90's era codec design (about which Xiph can't do anything without breaking compatibility). And other codecs are getting better as well; if Theroa is refined enough to be reasonably optimal in a year, it'll be competing against the improved H.264 and VC-1 codecs of a year from now, and H.265 not that far away.
There are all kinds of interesting things about Theora, but competitive low bitrate compression efficiency isn't going to be one of them.
Color me skeptical about the impartiality of a company named after one of Microsoft's key embrace-extend efforts, particularly when the issue is public access to media via non-proprietary codecs.
No, I work for Microsoft proper. But I'm speaking as a compression nerd here. Honestly, if Theora was a big sucess that'd be good for Silverlight; we have a codec extensibility model and Flash doesn't. But people seem to think that HTML5 and Theora are going to solve problems that the're not able, or even meant to, solve.
But there's certainly no reason in the world you should or would need to take my word for any of this. x264 and ffmpeg2theora would make a short project out of doing your own comparison.
Make sure to run the lastest Theora alpha 2, and use "-V" to turn on the slowest, highest quality mode.
"Better" is alwasy about context. Try a head-to-head for a scenario of interest to you.
So, Mozilla is the only one who can legally distribute the sources and binaries of Firefox*, under your system.
It's not MY system :). And Mozilla would have to ask them about what counts as a "product" in this context. It could be that a company distributing their own binaries based on Mozilla would need their own license or something. Or could link back to a Mozilla decoder lib or something.
IANAL, thank goodness.
Thanks for providing us with the objective point of view from Microsoft's evangelism department, Ben. Your support for Chris Dibona's point of view pretty much confirms for me he's flat wrong.
I'm actually not an evangelist anymore; I run strategy for Silverlight media technologies.
And as such, have done a lot of Theora encodes to see how interesting it is. It's be trivial to support as a managed decdoer in the the new Silverlight 3 Raw AV pipeline, so it's fine by me if it turns out to be great, but I think peope have very unrealistic expectations for it.
Well, I like to think I've gotten past the point of claiming :).
http://www.bing.com/search?q=%22world's+greatest+compressionist%22&src=IE-SearchBox&Form=IE8SRC
http://www.focalpress.com/Book.aspx?id=7224&terms=%22ben+waggoner%22
(finishing up the second edition right now).
A quick pan isn't that informative without knowing anything about the rate control model they're using. 1 sec of hard video in a 15 sec buffer can be smoothed over.
Yep, didn't catch that until after that last post.
Interesting that they cut it off right before the challenging high motion parts, though.
Seeing as it takes more horsepower to decode H.264 than the older H.263/Sorenensen/Spark/whatever they were using before, I assume it's my aging PC that's at fault. But buying a new PC just to get 30FPS on YouTube is a non-starter, given that the existing one still does everthing else I need it to do at acceptable speed.
That's the big drawback of H.264 - decoder complexity. And YouTube's encodes aren't the hardest either; there's deeper features yet they aren't using.
One of the design goals for what's likely to become H.265 is a 25% improvements in compression efficiency WITH a 50% reduction in decoder complexity.
Still, what bitrate are you downloading? 500 Kbps should be okay even for that ancient box, if you at least have SSE2.
Where's your evidence? Why is Greg's example odd? Have you done a comparable experiment with a different video clip to justify your 2x claim, or more importantly, show that at the same bitrate Theora looks much worse on your clip?
Xiph's own metrics show a 2x advantage on even a very easy short clip:
http://web.mit.edu/xiphmont/Public/theora/demo7.html
And yes, I've done plenty of my own tests as well on Big Buck Bunny and many other clips.
Why would it, since it didn't at 499Kbps? Or are you claiming that Youtube uses a bad H.264 encoder? Or do you think that example is rigged?
YouTube doesn't use High Profile (no 8x8 blocks or adaptive quantization matricies) or CABAC entropy coding. So they're going to be at least 20% less efficient than the best encodes could be.
Also, they trimmed the clip before the really interesting high motion parts. Most of the shots in the section they did use were static camera, and as animation is noise-free. Toss a nice grainy movie trailer in there and Theora shows basis pattern left and right.
I wish he'd reported the Theora settings used in the encode.
An open-source browser cannot legally read h264 video, that is the real issue that people seem to have trouble to understand.
I keep hearing that, but I don't know why that would be so. MPEG-LA requires a fee for distribution of products. But Mozilla could pay the decoder cap fee (maxes out a $5M/year next year) and allow as many people to download a H.264-enabled Firefox as they want, no?
That is why the HTML standard only mandates a format that is not impaired by any legal restrictions: Theora.
HTML5 does not mandate any codec or format. Ogg with Vorbis and Theora were proposed, but not included in the current draft, due to concerns by (IIRC) Nokia and Apple.
But the cost of the H.264 licenses are vanishingly small compared to the extra bandwidth cost of using Theora for a company like Google or Apple.
Using H.264 is free as long as clips are under 12 minutes. A number which may sound familiar to YouTube users :).
Here's info on the MPEG-LA licensing terms: http://www.mpegla.com/avc/avc-agreement.cfm
Prove Microsoft's implementation is as secure as the one in Firefox, and I'll listen to you.
And how does one "prove" security again :)?
Anyway, given the WMP OCX component has been in the market and supported code for many years with a good security track record, it seems validation would be more needed for Firefox's brand-new beta code.
...up to a certainly size/bitrate, yes. But I don't know if it would be enough to fill the screen of 480x320 or VGA phone, particuarly at a high quality.