Sure. but if you're going to say RealVideo 9 is a bad codec, compared to what? it's no fair saying that RealVideo @ 300 Kbps isn't as good as Divx at 3000! Pick a data rate and tell me the codec that offers superior image quality to RealVideo 9 at that data rate.
RealVideo 9 is surprsingly good for modem users. I woudln't watch a music video over it, but it is certainly adequate for simpler kinds of communication.
This is RealPlayer 8, not RealOne (aka RealPlayer 9).
They have clients for Linux, Solaris, Irix, AIX, Unixware, an hp-ux, running on x86, SPARC, MIPS, Alpha, and PPC. I'm not sure if all those binary codecs are available in the OSS stuff yet.
If you need to fit in a bitrate-constrained environment, CBR is the way to go. You could use a VBR with the max bitrate set to the bitrate contraint, but this would mean you'd be using fewer bits than you could with the easier portions of the stream.
Where do you get Vorbis offering better quality at a given data rate? Try doing a 64 Kbps CBR encode (since it is for streaming) to.ogg versus a 64 kbps RealAudio 8 Stereo Music encode. RA8 will sounds a LOT better. And you can still dance to it at 32 Kbps.
I'll skip past most of your whining, and focus on something you're just straight-up technically wrong about.
The RealVideo 9 codec has the best compression effciency (bang for the bit) of any publically available codec in a GM version (I'm discounting Windows Media Video 9, since it is still in beta, but that would be somewhat better in some cases).
Sure, it rarely looks perfect, but it does better than other codecs.
Bear in mind that going from uncompressed video to modem data rates is about a 12000:1 compression! Folks used to watching DivX don't often realize that those files use data rates several times that of typical real-time streaming even over broadband. Also, a file download can have huge data rate spikes that aren't as possible under real-time streaming.
They certainly do want 3rd party help in porting client software to non-mainstream OS's. RealOne is only available for Windows and (in beta) MacOS X. While they've had a *NIX client in the past, I'm guessing that this source release is how they expect to address those platforms in the future.
Which, given how much "if they just released the source, I'd port it" offers media software have received on Slashdot over the years, it's certainly a fair calling of that bluff.
They've said one of the big motivating factors being their OSS initiative is the daunting task of having to port RealPlayer to about 30 different mobile phone platforms. Real simply doesn't have enough engineers to do all that in-house.
Darwin Streaming Server can stream MPEG-1, MPEG-2, and MPEG-4. It can also stream with codecs like H.263 that are compatible with open source solutions. And QuickTime streaming itself is based on RTP/RTSP/SDP, etcetera. The only thing propritary is QuickTime specific codecs, and DSS doesn't know from codecs and clients.
As Charles Wiltgen says, "Darwin Streaming Server is the Apache of Streaming."
Sure. However, MS has historically not tried to obfuscate the bitstream itself, instead applying encryption at the DRM stage. I don't know if that's the case here, though, but I imagine it is.
There are lots of Profiles (defining the tools that can be used) and Levels (defining the contraints for the parmeters of those tools) in MPEG-4. A good initial standard is ISMA Profile 1, which uses Advanced Simple video and AAC-LC audio. So, same video codec as DivX (which can export.mp4, but without audio), and a MUCH better audio codec than MP3/Vorbis.
Alas, QuickTIme doesn't support Advanced Simple, only Simple.
Going forward, I hope to see H.264 take over as the baseline codec in MPEG-4. It should be competitive with today's best propritary codecs (Windows Media Video 9 and RealVideo 9), and have no licensing costs with the baseline profile.
Correct. The bitstream was locked down with WMA2, so WMA9 encoded content will play back just fine to that decoder, and presumably the new open source version of it.
Do note this is only the audio codec. While there are already decoders for the older MS MPEG-4vX codecs, most.wmv files will use the Windows Media Video codec, which AFAIK don't have any kind of open source solution.
This is presumably a decoder compatible with the classic "WMA" codec. This bitstream was locked down with WMA2, although they have made progressive enhancements to the encoder all the way through WMA9. It's really a very good codec for stereo 44.1 16-bit kinds of applications.
Not that there are two NEW WMA codecs introduced in Windows Media 9 that aren't backwards compatible.
WMA 9 Professional aims for higher bitrates (48-768 Kbps) and multichannel (stereo with matrixed support or six discreet channels). And it can do deeper than 16-bit. If the old WMA was a competitor to MP3, WMA9Pro is a competitor to AC-3/Dolby Digital. And like WMA v. MP3, it has about a 50% advantage on compression efficiency.
There is also WMA9 Voice, which is a really good narrowband codec. Despite its name, it does a surprisingly good job with music as well. It operates from 4-20 Kbps.
WMA9 Voice is replacing the ACELP.net voice codec, which Microsoft licensed binary code for, but doesn't have the source code. Thus they couldn't get it for MacOS X or PocketPC.
Anyway, this decoder should handle 95% of the.wma files out there today, although that percentage might start dropping as WMA9 Pro gets traction. It is certainly a welcome effort!
What about mpeg4ip? It's a complete set of MPEG-4 encoding, streaming, and player tools.
http://sourceforge.net/projects/mpeg4ip/
95% activity on SourceForge!
And it can tune into MPEG-4 streams hosted from Darwin Streaming Server, Envivio, Helix Server, and its own server. Good stuff.
Now if I could only get it to compile on MacOS X for interoperability testing, and so I could get Advanced Simple to work (QuickTime only does Simple).
Helix, groovy as it is, won't solve these problems for you anytime soon.
People will still want Windows Media since it is installed, and Mac people will still want QuickTime. Apple might incorporate Helix playback in QuickTime (unlikely, but possible), but it is very unlikely Microsoft would build.rm playback into the OS.
The only format that works for good fire-and-forget universal playback today is MPEG-1. But that means no real-time streaming, and huge files. There is hope for MPEG-4, which is already supported by both Real and (in a more limited subset) QuickTime, but there isn't any default Windows Media support no or promised.
MPEG-4 is also fragmenting somewhat, between Simple (a baseline) and Advanced Simple (much better, but not supported in QuickTime) today, and MPEG-4 Part 10/H.264 (much better yet, and in products 2003).
Unfortunately, just having a ".mp4" file doesn't tell you what player specs are requried to play it back.
Well, RealNetworks has always had involvement in the standards process, and adopting standards-based technologies. Among high points:
Major force behind creation of RTSP, and adopted it, replacing their former propritary streaming codec. This includes things like SDP (Session Description Protocol). A LOT of what QuickTime Streaming does is based on the standards RealNetworks pushed, and even Windows Media 9 is emphasizing RTSP over MMS.
Early adopter and force behind SMIL, and first adopter of SMIL 2.0 in a streaming product (QuickTime Player is still using 1.0).
First player bundled with MPEG-4 playback (via Envivio plugin).
In talking with lots of RealNetworks employees, I think they understand that they can't monitize file formats anymore, and instead are focusing on building competitive products that can play outside their walled garden. I'm sure they hope that lots of commercial products will be based on Helix, ala MP3 encoders and players.
Sorenson Video 3.1 Pro is very competitive against MPEG-4 Advanced Simple implementations like DivX. It supports 2-pass VBR, B-frames, etcetera.
QuickTime can only decode Simple, not Advanced Simple, so SV3.1 can beat it handily.
Of course, they used Sorenson Video 1, which is much old and lousy. Presumably this is for backwards compatibility (to QT3), instead of the only months old QT6 which is required for MPEG-4.
MPEG-4 is awesome, but it is NOT all about maximum quality at small files. Propritary codecs beat all public implementations of MPEG-4 today on this measure.
Of course, being an ISO standard has other, huge, advangages.
Alas, I don't know of any really good algorithms to fix this problem in post. The best thing to not use composite equipment (not an option or you, alas).
Turning down the sharpness on your TV (a good thing to do with any digital source) can help reduce the effect.
NTSC video will display 480 lines out of 525, regardless of the format, be it VHS, broadcast, or DVD. This is why detail is measured horizontally with analog video.
1. Actually, NTSC VCD is 352x240. This is converted to 320x240 on in-window playback on computers, of course, in order to maintain the 4:3 aspect ratio.
2. Well, we needed SOMETHING to do with Moore's law! Heck, my dual Athlon 2100+ MP machine did a 16-hour render last week...
Yeah, but given CD-R prices, you can just do a single burn (and have an 800 MB disc to boot), and then just toss the disc after your sister comes over. They're darn cheap if you don't need jewel cases for long-term storage.
Indeed. Being able to do a 2-pass encode typically reduces the data rate requirements by 10-50% at the same quality level (depending on how variable the content in the video is).
And yes, MPEG-4 Advanced Simple in Divx is a LOT better than MPEG-2 in terms of compression efficiency. It also has postfiltering (like a deblocking filter) that reduces the appears of artifacts.
VCD's very contraint implementation of MPEG-1 doesn't have anywhere near this kind of flexibility. Of course, it DOES offer wide compatiblity with all kinds of devices.
Slashdot Mirror
Sure. but if you're going to say RealVideo 9 is a bad codec, compared to what? it's no fair saying that RealVideo @ 300 Kbps isn't as good as Divx at 3000! Pick a data rate and tell me the codec that offers superior image quality to RealVideo 9 at that data rate.
RealVideo 9 is surprsingly good for modem users. I woudln't watch a music video over it, but it is certainly adequate for simpler kinds of communication.
This is RealPlayer 8, not RealOne (aka RealPlayer 9).
They have clients for Linux, Solaris, Irix, AIX, Unixware, an hp-ux, running on x86, SPARC, MIPS, Alpha, and PPC. I'm not sure if all those binary codecs are available in the OSS stuff yet.
If you need to fit in a bitrate-constrained environment, CBR is the way to go. You could use a VBR with the max bitrate set to the bitrate contraint, but this would mean you'd be using fewer bits than you could with the easier portions of the stream.
Where do you get Vorbis offering better quality at a given data rate? Try doing a 64 Kbps CBR encode (since it is for streaming) to .ogg versus a 64 kbps RealAudio 8 Stereo Music encode. RA8 will sounds a LOT better. And you can still dance to it at 32 Kbps.
Of course, they do have binary code for all their codecs for MacOS X at least. And for SPARC and some other platforms through RealVideo 8.
Anyone know what the range of binary platforms supported is?
I'll skip past most of your whining, and focus on something you're just straight-up technically wrong about.
The RealVideo 9 codec has the best compression effciency (bang for the bit) of any publically available codec in a GM version (I'm discounting Windows Media Video 9, since it is still in beta, but that would be somewhat better in some cases).
Sure, it rarely looks perfect, but it does better than other codecs.
Bear in mind that going from uncompressed video to modem data rates is about a 12000:1 compression! Folks used to watching DivX don't often realize that those files use data rates several times that of typical real-time streaming even over broadband. Also, a file download can have huge data rate spikes that aren't as possible under real-time streaming.
They certainly do want 3rd party help in porting client software to non-mainstream OS's. RealOne is only available for Windows and (in beta) MacOS X. While they've had a *NIX client in the past, I'm guessing that this source release is how they expect to address those platforms in the future.
Which, given how much "if they just released the source, I'd port it" offers media software have received on Slashdot over the years, it's certainly a fair calling of that bluff.
They've said one of the big motivating factors being their OSS initiative is the daunting task of having to port RealPlayer to about 30 different mobile phone platforms. Real simply doesn't have enough engineers to do all that in-house.
Presumably really the Windows Media MPEG-4v1 and v2 codecs. The current "Windows Media Video" (without the MPEG-4) line started with V7.
You can't even encode files to those old v1 and v2 with the WM SDK for the past few versions.
Darwin Streaming Server can stream MPEG-1, MPEG-2, and MPEG-4. It can also stream with codecs like H.263 that are compatible with open source solutions. And QuickTime streaming itself is based on RTP/RTSP/SDP, etcetera. The only thing propritary is QuickTime specific codecs, and DSS doesn't know from codecs and clients.
As Charles Wiltgen says, "Darwin Streaming Server is the Apache of Streaming."
Sure. However, MS has historically not tried to obfuscate the bitstream itself, instead applying encryption at the DRM stage. I don't know if that's the case here, though, but I imagine it is.
Actually, WMA9 Pro is something else. It's a compressed 2 or 6 channel codec, competing with Dolby Digital.
A lossless codec should be relatively easy to reverse engineer, since you know EXACTLY what the correct output is!
Well, how about ".mp4!"
.mp4, but without audio), and a MUCH better audio codec than MP3/Vorbis.
There are lots of Profiles (defining the tools that can be used) and Levels (defining the contraints for the parmeters of those tools) in MPEG-4. A good initial standard is ISMA Profile 1, which uses Advanced Simple video and AAC-LC audio. So, same video codec as DivX (which can export
Alas, QuickTIme doesn't support Advanced Simple, only Simple.
Going forward, I hope to see H.264 take over as the baseline codec in MPEG-4. It should be competitive with today's best propritary codecs (Windows Media Video 9 and RealVideo 9), and have no licensing costs with the baseline profile.
Correct. The bitstream was locked down with WMA2, so WMA9 encoded content will play back just fine to that decoder, and presumably the new open source version of it.
Do note this is only the audio codec. While there are already decoders for the older MS MPEG-4vX codecs, most .wmv files will use the Windows Media Video codec, which AFAIK don't have any kind of open source solution.
.wma files.
This is mainly useful for playing
This is presumably a decoder compatible with the classic "WMA" codec. This bitstream was locked down with WMA2, although they have made progressive enhancements to the encoder all the way through WMA9. It's really a very good codec for stereo 44.1 16-bit kinds of applications.
.wma files out there today, although that percentage might start dropping as WMA9 Pro gets traction. It is certainly a welcome effort!
Not that there are two NEW WMA codecs introduced in Windows Media 9 that aren't backwards compatible.
WMA 9 Professional aims for higher bitrates (48-768 Kbps) and multichannel (stereo with matrixed support or six discreet channels). And it can do deeper than 16-bit. If the old WMA was a competitor to MP3, WMA9Pro is a competitor to AC-3/Dolby Digital. And like WMA v. MP3, it has about a 50% advantage on compression efficiency.
There is also WMA9 Voice, which is a really good narrowband codec. Despite its name, it does a surprisingly good job with music as well. It operates from 4-20 Kbps.
WMA9 Voice is replacing the ACELP.net voice codec, which Microsoft licensed binary code for, but doesn't have the source code. Thus they couldn't get it for MacOS X or PocketPC.
Anyway, this decoder should handle 95% of the
What about mpeg4ip? It's a complete set of MPEG-4 encoding, streaming, and player tools.
http://sourceforge.net/projects/mpeg4ip/
95% activity on SourceForge!
And it can tune into MPEG-4 streams hosted from Darwin Streaming Server, Envivio, Helix Server, and its own server. Good stuff.
Now if I could only get it to compile on MacOS X for interoperability testing, and so I could get Advanced Simple to work (QuickTime only does Simple).
Helix, groovy as it is, won't solve these problems for you anytime soon.
.rm playback into the OS.
People will still want Windows Media since it is installed, and Mac people will still want QuickTime. Apple might incorporate Helix playback in QuickTime (unlikely, but possible), but it is very unlikely Microsoft would build
The only format that works for good fire-and-forget universal playback today is MPEG-1. But that means no real-time streaming, and huge files. There is hope for MPEG-4, which is already supported by both Real and (in a more limited subset) QuickTime, but there isn't any default Windows Media support no or promised.
MPEG-4 is also fragmenting somewhat, between Simple (a baseline) and Advanced Simple (much better, but not supported in QuickTime) today, and MPEG-4 Part 10/H.264 (much better yet, and in products 2003).
Unfortunately, just having a ".mp4" file doesn't tell you what player specs are requried to play it back.
Well, RealNetworks has always had involvement in the standards process, and adopting standards-based technologies. Among high points:
Major force behind creation of RTSP, and adopted it, replacing their former propritary streaming codec. This includes things like SDP (Session Description Protocol). A LOT of what QuickTime Streaming does is based on the standards RealNetworks pushed, and even Windows Media 9 is emphasizing RTSP over MMS.
Early adopter and force behind SMIL, and first adopter of SMIL 2.0 in a streaming product (QuickTime Player is still using 1.0).
First player bundled with MPEG-4 playback (via Envivio plugin).
In talking with lots of RealNetworks employees, I think they understand that they can't monitize file formats anymore, and instead are focusing on building competitive products that can play outside their walled garden. I'm sure they hope that lots of commercial products will be based on Helix, ala MP3 encoders and players.
Sorenson Video 3.1 Pro is very competitive against MPEG-4 Advanced Simple implementations like DivX. It supports 2-pass VBR, B-frames, etcetera.
QuickTime can only decode Simple, not Advanced Simple, so SV3.1 can beat it handily.
Of course, they used Sorenson Video 1, which is much old and lousy. Presumably this is for backwards compatibility (to QT3), instead of the only months old QT6 which is required for MPEG-4.
MPEG-4 is awesome, but it is NOT all about maximum quality at small files. Propritary codecs beat all public implementations of MPEG-4 today on this measure.
Of course, being an ISO standard has other, huge, advangages.
This is exactly what I'm talking about.
Alas, I don't know of any really good algorithms to fix this problem in post. The best thing to not use composite equipment (not an option or you, alas).
Turning down the sharpness on your TV (a good thing to do with any digital source) can help reduce the effect.
I exactly am talking about horizontal resolution.
NTSC video will display 480 lines out of 525, regardless of the format, be it VHS, broadcast, or DVD. This is why detail is measured horizontally with analog video.
Folks,
Say, did anything ever come of On2's VPVision? It was an open-source PVR product based on their VP3 codec (which is now being used in Ogg Theora).
http://www.on2.com/vpvision/vpvision.html
Downloadable source and everything! Only for Windows, though.
Anyone ever check it out?
1. Actually, NTSC VCD is 352x240. This is converted to 320x240 on in-window playback on computers, of course, in order to maintain the 4:3 aspect ratio.
2. Well, we needed SOMETHING to do with Moore's law! Heck, my dual Athlon 2100+ MP machine did a 16-hour render last week...
Yeah, but given CD-R prices, you can just do a single burn (and have an 800 MB disc to boot), and then just toss the disc after your sister comes over. They're darn cheap if you don't need jewel cases for long-term storage.
Indeed. Being able to do a 2-pass encode typically reduces the data rate requirements by 10-50% at the same quality level (depending on how variable the content in the video is).
And yes, MPEG-4 Advanced Simple in Divx is a LOT better than MPEG-2 in terms of compression efficiency. It also has postfiltering (like a deblocking filter) that reduces the appears of artifacts.
VCD's very contraint implementation of MPEG-1 doesn't have anywhere near this kind of flexibility. Of course, it DOES offer wide compatiblity with all kinds of devices.