Bounties at Xiph.Org Options

[QuantumKnot]

http://wiki.xiph.org/Bounties

I wonder if it is a good or bad sign that nearly all of these bounties are related to Vorbis development.

If you can think of some good things to add, feel free to add them. I'm itching to look into some of them but my plate is totally full till March next year.

[rjamorim]

Maybe we should start a fund to bring Monty back to coding Vorbis

Oh, wait, I'm afraid it has already been proven that doesn't work.

This post has been edited by rjamorim: Oct 20 2004, 04:28

[OggZealot]

Maybe they could add a bounty for the one who will built the official win32 Vorbis-tools 1.1 (<== Joke)

no seriously, I think it's a good idea but the bounties are a bit low (specially considering the MPC donation) ... at first look I would give some cash for someone to code priority 7-1-3 ... but I am convinced Monty will code priority 1-3 himself when Theora 1 will be out ... this is a logic reward between Xiph coders ... (Theora devs code Video for Xiph , then Monty tunes Audio for Theora as a reward for his friends)

so in the end I would only be REALLY interested to give for priority N°7, specially as it would save space on my HD & HD space is money ... so in a way I would get my money back

... but this topic leads me to another wondering I already had before:
Why Xiph doesn't join Gnome ?

Ok Gnome is a desktop, a gui, & Xiph encoders are command line ... but Gnome is actually more than a desktop ... if Beep (Xmms) & Rythmbox are part of Gnome I don't see no reason why Xiph couldn't be incorporated in Gnome ... in fact in the heart of most linux users Xiph is already un-noticably a part of Gnome.

Xiph is actually pure open source ... & if you look at the linux world:
distro:
there is Debian in the one hand & Red Hat/Mandrake/Suse & co on the other hand,
desktop:
Gnome in the one hand & KDE in the other hand ...

Red Hat & Co, KDE are all "semi-commercial" while Debian Gnome is the "true" linux

I don't use (yet) linux myself but for me it's clear
wma = Microsoft
m4a = Apple
ogg = Linux, & specially Debian Gnome

... when I look at some Gnome bounty & compare the price with the bitrate peeling ... it becomes clear that if joining Gnome can increases the Xiph bounties price ... I vote for Xiph in Gnome ...

specially if it can boost Xiph codecs integration to the desktop ... a simple exemple that comes to my mind is windows explorer vorbisext shell program extension integration in all Gnome based file browser ... so there are potential advantages in joining Gnome ...

... another thing I have in mind since some time is that Xiph should swallow W3C png ...
Xiph is the protector of multimedia on the net ...
for me multimedia:
Audio ==> Video <== Image
Vorbis==> Theora <== OpenSource JPG (Remember Forgent)
Flac ==> CorePNG <== PNG

so Xiph is missing 1/3 of what it claims to defend

Debian Gnome/Xiph/W3C ==> Same Fight,
Debian Gnome = Open Application
Xiph = Open Multimedia Format
W3C = Open Web Standard

for me
Xiph should join Gnome
Xiph should swallow PNG (a bit like OOo has swallow a part of xml with .sxw)
Xiph should do the same for JPG as W3C did for GIF (& yes even if JPG is open, for JPG it's not a patent issue it's a technology issue ... don't wanna use JPG1 in 2010 just because it's open, an open source alternative to JPG MUST be developped because JPG2000 stinks (very slow loading) & is NOT open, not because JPG is "open")

... last idea I was wondering was:
Xiph obviously don't care about Win32 build, so why not releasing Xiph codecs as "open shareware" on windows & on linux as real open source ... I mean X-Chat already do that ... the code is open & everyone is free to make a free binary & distribute it, but if you want the official win32 built you have to pay for the time the dev spent developing a software for which he doesn't care as win32 is not his native OS ... when X-Chat dev did that some months ago I was very angry ... but with time I understund & started using Silverex unofficial X-Chat win32 build ... I mean with the actual situation of Xiph not releasing official win32 Vorbis-tools 1.1 ... it's the same ... I use unofficial win32 Vorbis-tools 1.1 Case build ... so Xiph releasing Vorbis-tools 1.1 as shareware wouldn't change anything for me ...

Indeed I dislike the idea much ... but staying on Xiph IRC channels for some time I know for sure they don't care at all about win32 ... almost all their dev are on linux (& as far as I understund Debian Gnome Linux ...) so instead of having this ridiculous situation where 1.1 is out for advanced users aware of unofficial build & new win32 users still use 1.01 at least the situation would be clear ... want win32 1.1 pay for it ... or use linux ... most Xiph dev are Linux zealot ... so come on Xiph, if you want to force win32 Xiph users to use Linux ... be honest ... TELL IT LOUD ...

finally, 3 bounties on 7 linked to Apple Itunes-Ipod-Quicktime ... this is pathetic ... I don't care at all about Apple software supporting Xiph, this is an abheration & a waste of money, there should be only 4 bounties ... all those not linked to Apple support.
Improving Rhythmbox (or wkMusik) & making it a GTK2 based crossplatform Xiph codec based perfect Itunes clone would be much more clever ...

Edit1: & the best way to have a perfect/complete Rhythmbox Itunes clone to fight mp4 growing popularity is to join Gnome ... so this closes the logic circle behind all that ...

... as always ... that's all folks ...

This post has been edited by OggZealot: Oct 20 2004, 18:51

[idioteque]

Maybe we should start a fund to bring Monty back to coding Vorbis

Oh, wait, I'm afraid it has already been proven that doesn't work.

Heh...very cyncial, but very true.

[dev0]

OggZealot, you should seriously stop posting about things you don't know shit about.

... but this topic leads me to another wondering I already had before:
Why Xiph doesn't  join Gnome ?

Why should it in the first place?
Xiph works on multimedia tools and codecs and Gnome on a desktop and its applications. Where's the connection? Gnome developers are already using Xiph's stuff. That's the beauty of opensource: Collabration between projects without any need to take each other over. But as you obviously have never even seen a single line of code, you can't understand.

Xiph is actually pure open source ... & if you look at the linux world:
distro:
there is Debian in the one hand & Fedora/Mandrake/Suse & co on the other hand,
desktop:
            Gnome in the one hand & KDE in the other hand ...

Red Hat, KDE are all "semi-commercial" while Debian Gnome is the "true" linux

In what way is KDE 'semi-commercial'? It's released under various opensource licenses and works just like many other open source projects. RedHat is a company and works on several products, one of them (Fedora) in a community process just like Debian.
Why is Gnome 'true opensource'? And why are you talking about Debian Gnome? The last time I checked my Debian box ran KDE beautifully.

specially if it can boost Xiph codecs integration to the desktop ... a simple exemple that comes to my mind is windows explorer vorbisext shell program extension integration in all Gnome based file browser ... so there are potential advantages in joining Gnome ...

You have never even used Gnome, right?

Xiph is the protector of multimedia on the net ...
for me multimedia:
Audio ==> Video      <== Image
Vorbis==> Theora    <== OpenSource JPG (Remember Forgent)
Flac    ==> CorePNG <== PNG

What for? Opensource doesn't work like that.
Also: You need to understand the difference between a standard and its implementation.

finally, 3 bounties on 7 linked to Apple Itunes-Ipod-Quicktime ... this is pathetic ... I don't care at all about Apple software supporting Xiph, this is an abheration & a waste of money, there should be only 4 bounties ... all those not linked to Apple support.
Improving Rhythmbox (or wkMusik) & making it a GTK2 based crossplatform Xiph codec based perfect Itunes clone would be much more clever ...

You can't be serious. The iPod and iTunes are used by millions of people. Proper Quicktime/iTunes integration of Xiph's work would benefit Xiph more than any tuning by QK or AoTuV could.

There's so much more nonsense in your post, it makes me wonder where you get these ideas from.

[rjamorim]

Ok Gnome is a desktop, a gui, & Xiph encoders are command line ... but Gnome is actually more than a desktop ... if Beep (Xmms) & Rythmbox are part of Gnome I don't see no reason why Xiph couldn't be incorporated in Gnome ... in fact in the heart of most linux users Xiph is already un-noticably a part of Gnome.

Well, are the Gnome Foundation guys even interested in absorbing Xiph?

This might not be all that easy. For starters, I would bet Monty isn't interested in having to report to other people - right now, he's the CEO of Xiph, and that gives him lot of maneuver room.

finally, 3 bounties on 7 linked to Apple Itunes-Ipod-Quicktime ... this is pathetic ...

This is realistic. I think Xiph acknowledged Apple's might in the DAP scene. It can only be fruitful to them if people port Vorbis for iTunes and iPod.


The problem I see with you, OggZealot, is that you are very closed-minded. Like so manu GNU zealots, you believe everything open source is good and everything closed source is bad and evil. There's no point in having such extreme manichaeistic view of the world.

This post has been edited by rjamorim: Oct 20 2004, 17:04

[OggZealot]

dev0:
Objection 1:
I made an edit to my first post if this wasn't clear from start:
-1- Xiph would pay for Itunes-Ipod-QuickTime Support
-2- This is absurd
-3- Developping a Xiph Itunes clone that would replace mpg-4 default codecs by ogg codecs would be much clever.
-4- The best way to do this seems to be using Rythmbox as a starting point & then port it to win32 via GTK2
-5- In order to do that Xiph should join or at last team with some Gnome devs ... so it's better being honest & simply join Gnome ... specially if it can increase incomings ...

-6- I agree that Xiph should have a particular statut in Gnome as it's not the same as what Gnome is usually doing

Objection 2:
-KDE is not free on win32, QT library is proprietary ... that's the reason why GIMP, X-Chat, AbiWord, Gaim are all ported to win32 & not KDE based software, so yes Gnome is more free than KDE.
-Red Hat/Mandrake/Suse are all compagnie, Debian is a community. Fedora is between ... that's I didn't speak of it, you introduced it, not me ...
Edit: Oups, I introduced Fedora ... sorry, my wrong, I was thinking Red Hat & wrote Fedora ...

Indeed it's all open source ...

Objection 3:
- so far I installed Fecora, Mandrake & tested many Live CD Knoppix/Gnoppix/MandrakeMove/Suse ... most of them are KDE (Mandrake & Suse favor KDE) but I did tested Gnome & favor it over KDE
- I am certainly not a linux zealot nor very skilled at it as I favor Win32 so far ...

so Yes I am a Gnome newbie ... that don't change anything ...

Objection 4:
Garf often recalls vorbis fans like me that the current vorbis implementation is far from perfect, so yes I know the difference the specification & the implementation

Objection 5:
If that's nonsense ... than it's very serious nonsense because after reading all your objections I still trust in what I explained ...

Call me a visionnaire, an utopist or a complete idiot
I know very well that it's only my opinion & that it worth nothing if I don't code it myself ... but plz just let me tell it even if you disagree ...

Edit: Red Hat-KDE error

This post has been edited by OggZealot: Oct 20 2004, 18:57

[OggZealot]

after re-reading all, the only real point of argue between us is that you seem to think Apple supporting Xiph is a wonderfull improvement for Xiph ...

I disagree on that for a very simple reason I have serious doubt on Apple porting Itunes to Linux ... so even with Xiph codec support, Apple software are of no use for Linux users which are the basis of Vorbis fan ...

don't get me wrong Apple supporting Xiph would be great, but I disagree on the fact that this seem to be a priority for Xiph ... & worst that they would pay for it ...

for me it's an error in the target user audience,

1-If Xiph wants to please it's linux fan base, it should team (or join Gnome) & make a GTK2 Itunes clone then port it to win32 ...

2-If Xiph wants to attract a new apple/mpeg4 fan basis ... then yes having Apple supporting Xiph can be a good way to steal some users.

But honestly what is more important strenghten the actual basis of fan or seducing potential new users ?

Do you really think people using Itunes would suddenly switch to Vorbis if vorbis were supported ... I have serious doubts ... because if these people didn't realize the difference between open & proprietary codecs before why would they suddenly realize it now ... specially has even if Apple would support Xiph codecs via plugins this codecs support WOULD NOT BE NATIVE ... so you still have to be an advanced user ...

even with Xiph plugins for apple software ignorant people will just keep using Itunes just because they saw a colored publicity on TV ... a plugin won't change that fact ...

... so in the end who will use these plugins ? advanced users like rjamorim, which have already chosen mpeg-4 ... it will be a toy for them & nothing more ...

... on the other hand there is the solution of developping an full itunes clone ... which will please old users & ALSO attract some itunes users

I prefer from far having Xiph making a step toward Gnome than having Xiph making a step toward Apple ...

In the end it's a question a point of view ... seducing apple users is a myth for me ...

[JeanLuc]

It can only be fruitful to them if people port Vorbis for iTunes and iPod.

Is the iPod's hardware capable of decoding vorbis yet ? I remember some statements related to the 3G series ...

[rjamorim]

Is the iPod's hardware capable of decoding vorbis yet ? I remember some statements related to the 3G series ...

The iPodLinux developer reported he managed to get vorbis decoding at 80% real time speed on iPod (probably a 3G one). So, I guess it's only a matter of optimization.

[dev0]

Long story short:

1. MP4/AAC is not a proprietary codec. We had many discussions about this topic and no matter how much you twist and turn it: There are lots of things one could favour Xiph's work over MP4 for, but 'proprietary' is not one of them.

2. GTK2 has serious problems on any platform but X11, why bother writing/porting software that uses it, if you could utilize a stable platform like QuickTime, which would add support for your codec in a ton of apps at once? (illiminable's new DirectShow filter are another good example)

3. KDE is as free as Gnome. QT's native windows version is only available via a commercial license, but you can use the X11 version in combination with cygwin's Xfree86 port. However that's irrelevant regarding the freedom of the code. One could also argue if porting free software to a non-free platform has anything to do with freedom at all.

4. Why would anyone want to work on another mediaplayer? Improving support in existing frameworks (DirectShow, GStreamer, QT) is more important and caters a much wider audience than another audio player.

5. You don't even the balls to use a GNU/Linux system yourself, but think you can judge what long time open source supporters are supposed to do to improve popularity of their products? Come down. I'm pretty sick of hearing your braindead pseudo-revolutionary plans.

6. RedHat has always been a GNOME company ( employing GNOME devs etc.) and is still using GNOME as the default desktop. Stop talking out of your ass.

[Digisurfer]

Ogg Vorbis Bitrate Peeling sounds interesting. I've come across the term on these forums before but never considered it as something likely to be important. Reading the criteria it must meet has rasied my curiosity though. I think I have an idea from reading there how this might work, but I'm hoping someone might be able to provide a good thorough definition, or a link to one. Thanks!

[OggZealot]

My misstake for Red Hat & Gnome ... this error comes from here ==> as Mandrake was originally a fork of Red Hat, & Mandrake favors KDE I was thinking Red Hat was KDE oriented too ... I am not a big Red Hat fan so sorry for the error ... I am actually planning to retry Mandrake (I am French) & Debian (Both with Gnome) ... so I promise I'll try to practice what I preach ... for 1-5 well I am aware ...

[SebastianG]

http://wiki.xiph.org/Bounties

I wonder if it is a good or bad sign that nearly all of these bounties are related to Vorbis development.

I don't want to sound too pessimistic but I seriously doubt that anyone is going to fulfill the requirements so he can claim bounty #3 and #7.

BTW: Isn't #7 a bit contradictionary ? I guess I don't fully understand the meaning of the next to last requirement.


SebastianG

[idioteque]

I don't want to sound too pessimistic but I seriously doubt that anyone is going to fulfill the requirements so he can claim bounty #3 and #7.

BTW: Isn't #7 a bit contradictionary ? I guess I don't fully understand the meaning of the next to last requirement.

You're not being pessimistic, peeling is a pipe dream and the bounty for #7 is ridiculous. No one has been able to demonstrate that peeling is even remotely possible. Segher's peeler sounded horrible and I've never seen Monty outline how peeling is suppose to be implemented. This is most likely because it can't be done without adding a lot of unnecessary data while encoding to help the peeler along. It's not worth it.

[SebastianG]

You're not being pessimistic, peeling is a pipe dream and the bounty for #7 is ridiculous. No one has been able to demonstrate that peeling is even remotely possible.  Segher's peeler sounded horrible and I've never seen Monty outline how peeling is suppose to be implemented.  This is most likely because it can't be done without adding a lot of unnecessary data while encoding to help the peeler along. It's not worth it.

The requirement of making old Vorbis streams somehow peelable is really tough!
A new/better encoder could - besides from optimizing the codebooks / encoder strategy for peeling - include some meta informations in the comment packet like what quality level you get after a certain VQ coding stage. I could imagine an encoder/peeler pair which can be used like this:

oggenc -q2/6 song.wav
> endoging at quality: 6.00 and 2.00 (peeling combo mode)
> ...
=> producing a q6 stream which can be easily peeled down to q2. Most likely the bitrate of the peeled (q2) stream will be higher than 96 kbps average since it's quite hard to get the same quality at the same rate compared a 'standalone' q2 stream. So, maybe resulting in 110 kbps at agerage... So The encoder could write a meta tag like "PEELING=p:3 b:110 q:2" which actually means: The peeler can chop of all VQ coding passes after the 3rd which results in a 110 kbps stream with a similar quality to the q2 mode. (support for multiple modes possible like -q1/4/8)

vorbispeel --info song.ogg
> Peeling Meta Tag(s) found.
> * 110 kbps (3 VQ coding passes, similar to q2 quality)

vorbispeel -b128 -osong-peeled.ogg song.ogg
> Peeling Meta Tag(s) found.
> selecting best matching level:
> * 110 kbps (3 VQ coding passes, similar to q2 quality)
=> producing a peeled stream with an average bitrate equal to or lower than 128 kbps by checking the Vorbis comment packet and chopping off some coding passes of each audio packet.

This is IMHO an approach which makes most sense - At least to start with. The peeler does not have to be that intelligent and can do its work pretty fast (faster than decoding) which is what we actually want.
But be warned: The quality/bitrate ratio of a peeled stream will be worse (you need a higher bitrate for the same quality) because of several reasons... for example: the stereo modes aren't that scalable...

Still, the task of updating the encoder to make this possible with a reasonable good quality is tough.


SebastianG

This post has been edited by SebastianG: Oct 20 2004, 22:34

[idioteque]

But be warned: The quality/bitrate ratio of a peeled stream will be worse (need mor bitrate for the same quality) because of several reasons... for example: the stereo modes aren't that scalable...

Not only will the quality/bitrate ratio suffer, but you will only be able to do so much peeling anyways. With changes in quality levels come changes in the psychoacoustic section of the encoder as well as the quantization. It's not just multi-tier quantization we're talking about here. Even if it was just a matter of adding and subtracting precision, the encoder isn't tuned for outputting those streams anyways.

In the end you have to re-tool the encoder, making it less bit efficient for a feature that everyone has been able to live without up to this point.

Can you tell I've frustratingly pursued this topic in the past?

Edit:

Still, the task of updating the encoder to make this possible with a reasonable good quality is tough.

Ah, I see we agree.

This post has been edited by idioteque: Oct 20 2004, 22:40

[rjamorim]

I disagree on that for a very simple reason I have serious doubt on Apple porting Itunes to Linux ... so even with Xiph codec support, Apple software are of no use for Linux users which are the basis of Vorbis fan ...

Oh, for God's sake! Vorbis already has plenty of support in Linux. They don't need more support in that platform. But it's badly supported in Windows and has nearly no support in MacOS. So, yes, adding Vorbis to a big audio platform like iTunes is a great idea.

But honestly what is more important strenghten the actual basis of fan or seducing potential new users ?

They don't need to seduce the Linux users, you halfwit. They already love Vorbis for its openness and whatever

... so in the end who will use these plugins ? advanced users like rjamorim, which have already chosen mpeg-4 ... it will be a toy for them & nothing more ...

LOL. Of course I wouldn't use it. What's the use for me of adding Vorbis to iTunes?

The great idea of adding Vorbis to iTunes is that it would make integration with iPodLinux much easier if it ever starts supporting Vorbis playback.


BTW: you claim I already chosen MPEG4, but you gotta keep in mind that, unlike you, I'm not a zealot (I used to be, granted...), and I use the best format, for my needs, for each situation. My Expanium won't play AACs, so I have been ripping most of my albums using Lame. When I send albums to friends, I will often encode to whatever format they want, even if it is (OMG!) Vorbis.

You badly need to learn to become flexible, dude. Zealotry only makes you look like a moron. Just look at Stallman...

I prefer from far having Xiph making a step toward Gnome than having Xiph making a step toward Apple ...

I prefer Xiph making a step in both directions, but that's just me.

[QuantumKnot]

Personally, I believe optimising lowmem tremor to run on the iPod should be at the top of the priorities. The iPod has the lion share of portable players (something like 92% of HD-based players?) so it would be a tremendous boost to Xiph and Vorbis if we can first get it running under Linux on iPod....then Apple may (a very remote 'may') look into it.

IIRC, the iPod processor is able to decode Vorbis but I think the problem is that tremor currently uses more memory than is available on the chip, so they are forced to use the DRAM, which is much slower to access, hence the 80% decoding. More optimisation needs to be done on tremor.

Plus I got an iPod and lots of Vorbis files so obviously I'm all in favour of it