Future of Nandub, divx and other stuff...

[aleksander]

Well...
New Year has begun…
Last year has been for me a year full of new adventures. The biggest one was – Divx…
I remember how I have discovered this excellent site and it’s fabulous board…
Congrats Doom9 – this is really a PIECE of work….Hope you can make it run forever :)
My first attempts were with Flask – I can’t forget my excitement when my first movie has been done….Since that a lot has changed – I discovered Nandub and it’s tremendous SBC and later Gknot. What can I say – You guys – Nando (Vv_Nn) and TheWef – are my heroes.
Of course I have to admit that without Doom9 I would still be sitting on my butt in front of computer wondering “What the hack is going on with this divx and all that stuff ???”…
I remember when I wrote Nando once asking such a stupid question (forget it – I won’t mention it here – don’t want you guys to laugh)….
I don’t wanna bother you with my personal (kinda) life, so let me get to my point…
What has happened to Nando??? I know – this has been asked many times. But let me ask you once again – What happened to Nando???
For a long time there is nothing new going on in relation with Nandub. Of course – there have been Nandub-lumafix and Nandub-ecf-file-fix released, but nothing in terms of Nandub and SBC itself.
So I have figured out that maybe I will try to encourage some of you guys to do something with it.
I have to mention here that I don’t have a CLUE about programming and all that stuff.
Why don’t you guys try to get together (not literally) and think about new version of Nandub that will make use of SBC and will incorporate divx 4.xx also?
@ Koepi – what has happened to Maras??? I remember your hacks – could you try to do something with divx 4.xx???
@ DDogg, realSpiderman, rmatei, Steady, Ookami, dividee – guys – you have the knowledge – you know WHAT exactly is going on… Could you try to get together and work with some of the best programmers over here on the new Nandub???
Take a look at PowerDivx Team – they have the potential, will and help of other people. Maybe it’s time to create something like Nandub Team????
I’m sure there will be a lot of people willing to help in testing and providing some necessary info about stuff you need –
@ manono, cofferscuffs, MaTTer – guys – you seem to be very active here – can you say a word and help me to encourage them????
@ BetaBoy – you’re the guy from DivxNetworks – I’m sure (at least I hope) you can provide some useful info and documentation about divx 4.xx.
@ everybody – what do you thing ??? – any of you over there thinking in the same way as I am???
Let’s give them our support – maybe we will be successful !!!!!
I know that divx 4.5 and divx 5 are around the corner (should be) so maybe it is pointless to start working on new version of Nandub which will make use of SBC and divx 4.xx. But maybe we will still be able to make use of SBC with divx 5 ???
Nando has introduced many options allowing us to control the codec wholly. Maybe it is pointless with divx 4.xx or 5???
I’m writing here since I want to know your opinion, guys. I want to know if there is still something that can be done – something that will help us control the quality of our rips – something that we can use to make it even better.
Waiting for your opinions…
Take care
aleksander

p.s. this discussion is supposed to be serious, so please do not post something like : Why don’t you do it yourself if you’re such a smart-ass“...

[diji1]

Why don’t you do it yourself if you’re such a smart-ass :) :) ...lol

[BlackSun]

Interesting post IMHO. Of course this is a good idea to join our forces/knowledges/skills. But a project need leaders who will able to motivate coders.

Actually we are doing something similar with the PowerDivX Team: antp, Matthias Lenk (coder of DeDynamic) and me as the PowerDivX Coder Team, and ChristianHJW and BetaBoy (even if he doesn't show himself) as the leaders.

We have some greats projects on our board: Using MP3PRO, OGG and AAC with AVI and the TMF file format which is an AVI replacement (see the spec in the articles sections at www.powerdivx.com). Tronic (TMF Creator), IngoRalfBlum(Coder of the only Ogg DSF) and Nic (Lot of DSF) are working on these projects and any help will be appreciated.

In our board you'll understand why AVI can handle properly VBR Audio, even Avery Lee himself give us a post.

Of course PowerDivX 4.0 is still under development. Check out www.powerdivx.com for any informations or if you want to participate...

Good luck aleksander

[aleksander]

BlackSun....
I visit your page from time to time. Actually my idea came out when I was looking at your page. You had the power to get all those people together and motivate them to work. Can't wait to see the results of it - especially interesting seems to be developing of new file standard - TMF.
If you need beta-testers or something - let me know. I don't have a clue about programming but I can test something for you. Of course I am time-limited, but will try to help...
take care BlackSun and good luck!!!!!
aleksander

ps. About a project leader - well - only one person comes to my mind - Doom9 !!!

[BlackSun]

Hmmm, I don't think Doom9 is a good idea because he's too busy... I think it will be interesting to start a small website for coder, like coder.doom9.net or coder.powerdivx.com

This website will announce new projects, will host projects, a forum for developpers... Coder who will visit this website will be able to decide and participate to a project...

My forum and website is opened for such projects...

[aleksander]

Good idea....
Let's wait for Doom9 and his opinion on this. Maybe he will start something like that...
Anyways -- good ideas how to make this project run are always welcome...
take care
aleksander

[Doom9]

I have to agree with that... I'm really running in the red zone already.. last week I had no free minute.. I was working on projects for my page all the time.

but to get back to the original topic of this thread.. I don't think Nandub will go anywhere from this point on. DivX 4.5 and above will give external programs a lot more control about the codec so stuff similar to what nandub does for divx3 will be possible and probably be implemented in external programs like morphix and divx4log. furthermore let's not forget that there's a new codec on the horizon: xvid.

and I think vbr encoding in the codecs will go a way that no nandub is needed... however, stuff like robshot did with cce will become more important.. you actually manually make some changes to the bitrate distribution and quantizer settings to obtain a better quality (but that means some manual work).

[BlackSun]

Originally posted by Doom9
but to get back to the original topic of this thread.. I don't think Nandub will go anywhere from this point on. DivX 4.5 and above will give external programs a lot more control about the codec so stuff similar to what nandub does for divx3 will be possible and probably be implemented in external programs like morphix and divx4log.



You are wrong here, in the last discussion with Gej, he told me that there is no way to give more control to externals programs. divx4log and MorphiX are dead... I whish I am wrong.


furthermore let's not forget that there's a new codec on the horizon: xvid.


Of course... but they do not have the time and the money of DivXNetworks...


and I think vbr encoding in the codecs will go a way that no nandub is needed... however, stuff like robshot did with cce will become more important.. you actually manually make some changes to the bitrate distribution and quantizer settings to obtain a better quality (but that means some manual work).

I am sure there is more possibilities with Nandub. I've done some tests with XviD, and doing the first pass not ratecontrolled by using a fixed quantizer give a Perfect result ! Why not doing a small encore which will:
- Do a first pass with Constant Quality (for DivX)
- Store 1st pass information in a .stats file (.stats are better than .log)
- Do the 2nd pass by just reducing the average bitrate...

For more information, see the xiphrom source code (Morphix for XviD) at www.xvid.org

[Teegedeck]

Originally posted by BlackSun


I am sure there is more possibilities with Nandub. I've done some tests with XviD, and doing the first pass not ratecontrolled by using a fixed quantizer give a Perfect result ! Why not doing a small encore which will:
- Do a first pass with Constant Quality (for DivX)
- Store 1st pass information in a .stats file (.stats are better than .log)
- Do the 2nd pass by just reducing the average bitrate...

For more information, see the xiphrom source code (Morphix for XviD) at www.xvid.org

Sounds good. (This is pretty much how GordianKnot works together with Nandub, forcing quantizers/CLs to Nandub via .ecf-files. And for me and others, this gives best results.) Have you tested it extensively?

[BlackSun]

Nope I haven't tested it extensively as:
- My CPU is a Duron 700 :(
- Very busy with coding :(

In my previous post, please replace "encore" by "encoder" :D Anyway I think this sound interesting. Maybe we should take a look at the Vidomi code, if I remember, there is an options called 'DivX 4 hack', dunno what is it.

I expect more people in this thread !! Aleksander ?!

[aleksander]

@ BlackSun
Yes...I expected some more people here...There is just you posting here mostly...:):)
Actually I expected people I mentioned in the post (Koepi, Steady, DDogg, realSpiderman, rmatei,dividee, Ookami, manono, LotionBoy,...) to post here with their opinions, but unfortunately they're not :(:(
How to inform them to post their opinions here????
I also nnoticed something interesting (I think...)
Nandub web site is still online.
As far as I know, you have to pay for keeping the web site on a server, you have to pay for a domain and all that stuff...
The www.nandub.org is still working. I can understand that it was online last year, cause probably Nando has paid for the whole year. But why is it still online, where he is not intending to do anything about Nandub??? Ther are hundreds of pages all over the world (including this one, of course) that have Nandub files (together with source). So...either someone is paying for Nandub page because he is a good person with a lot of money, or Nando is keeping this web site online cause he is planning something in the future....
What do you guys think????
take care
aleksander

[BlackSun]

I don't think this is right, the website is hosted by sourceforge..

[aleksander]

OK....So I was mistaken...
If it's SourceForge than I understand it's for free....
Well...I was hoping that maybe something is going on as there was a lot of rumours about new version of Nandub, that was supposed to be "around the corner"....
take care

[BlackSun]

No way... Even no news from Phreak404... Doom9 can you post a news about this thread ? It's time to see if we can do something like a nandub fo DivX 4

[TheWEF]

i know that nando had done beta-test versions that worked with divx4.
pheak404 aka vimvam aka int21h (on this board) is still around.
http://forum.doom9.org/member.php?action=getinfo&userid=49
you could ask him if he has the source.

wef.

[aleksander]

@ BlackSun
I wanted to ask Doom9 for posting a news about this thread, but you seem to be the more right person to do it....
@ Doom9 - hope you can post such a news...
@ TheWef - so...what's your opinion about the whole idea????
take care
aleksander

[BlackSun]

Phreak404/Vim(on Irc)... Stop changing your nick :p

Let's see... so, maybe we need to start the coder.doom9.net ?

[ProfDrMorph]

About Nandub for DivX4: wasn't there an API document somewhen that described how to use the divx.dll in special applications? I had a backup copy of it on CD but currently I've lend the CD to a friend so I can't look up if it was meant for DivX4 or only OpenDivX. This document could help in the creation of a special 2-pass DivX4 application.

I know many people who use Nandub do it because it's so highly configurable and so creates high quality video but we should also think about an "easy 2-pass mode". The user can decide if he wants the easy way ( a simple wizard that asks you a few simple questions and then does the best I can ) or the advanced way. So we would have an encoder that fits the needs of the average user AND the more advanced ones. IMO the best thing to do atm would be including this encoding features directly into Gordian Knot in some way since it already has an easy to use interface that has a lot of useful options.

[aleksander]

If there is such a document, probably the right person to ask for it would be BetaBoy form DivxNetworks.
About an "easy 2-pass"...
Well...We want to have a Nandub for divx4 because of all the features it has and all the control that you can have over the codec.
I think there is no need for "easy 2-pass", because it exists right now. Divx 4 gives you a certain level of control.
We want to have a "total" control, such as with divx 3.11
You can disagree with me. Maybe I don't quite get your point.
But first I want to encourage some people to start working on it. Later we can think of additional features.
take care
aleksander

[Doom9]

I think the api doc should be somewhere on projectmayo.com

@blacksun... I didn't know that.. I thought you or junto told me abou that.. but the last time I talked to anyone involved in DXn was a while back. I also wouldn't discard xvid just because there's no multi million dollar company behind it. PowerDivX is the best example that it works without that. My site is another..

a lot of the moderators that have been refered to here are actually not very active anymore (or not active at all) so you can wait a long time to get a comment. And after all.. it takes coders not moderators to create something

[Teegedeck]

I gotta say, I think external developers investing work into DivX4-applications/modifications are in danger of wasting their time (please don't feel offended) because they make themselves dependent on a commercial codec that A) can change without notice and make your changes naught (they are already leaving you out in the rain as you can read in one of BlackSun's posts), B) is closed source, so it is difficult AND illegal to change it. IMHO it could be much more fruitful to accept that the evolution here will branch into two different directions:
- DivX4 for companies (legality assured) and users that don't want to think about it a lot
- XviD for open source developers and power-users

Since both are MPEG4-compliant, you don't have compatibility-problems with these two; you can just change the fourcc and for example decode an XviD-stream with the DivX4-decore. I also hope that DivX-Networks will leave XviD alone since Isibaar made it clear to BetaBoy that they won't go official with binaries and hence won't eat into their profits.

[-h]

> Since both are MPEG4-compliant, you don't have compatibility-problems with these two; you can just change the fourcc and for example decode an XviD-stream with the DivX4-decore.

.. but for how long? There are over 15 MPEG4 profiles, and I very much doubt that DivX4 fully (without mentioning it) supports quarter-pel ME, B-frames, MPEG quantisation matrices and whatever else may creep into something as experimental as XviD over time. There have already been some grey-block issues mentioned in the forums..

For that reason, I can see quite a demand for *full* MP4 spec-compatible players in the future, that play whatever valid stream you throw at it. Perhaps with Ogg support hacked in ;)

Should be fun.

-h

[aleksander]

As some of you have mentioned here, Xvid is the way to go....
Maybe we shouldn't be thinking about Nandub for divx4 and start thinking about Nandub for Xvid???
If I made a terrible mistake saying that - don't blame me. I'm not a coder so all that stuff is kinda hard for me to understand...
take care
aleksander

[cofferscuffs]

@aleksander
I honestly don't know what's going on, but it seems to me atm, lots of devolpers are losing intrest in their projects.

[BlackSun]

Damned computer which close the window when I post my message :(

@Doom9: the problem is a motivation problem. Everybody can't have your motivation, mine or the motivation of other open-source soft. Money can be a motivation, but this motivation is different...

Anyway we need to keep in mind that DXN's objectives are different from backing up DVD. It's different for XviD. Maybe XviD is the way to go...:confused:

[Teegedeck]

Originally posted by -h
There have already been some grey-block issues mentioned in the forums... Did you actually see those? I've never seen that bug though I still use the old DivX4-DShow-filter and latest XviD, so I guess it has been fixed, whatever it was? :)

[LotionBoy]

Since someone asked for my opinion, I'll be glad to throw it out. I'm a film/CompSci major. I know Java and some ASM. I'm currently teaching myself C, C++, and hopefully MMX/3dNOW/SSE. Of course, I am also taking 3 programming classes and producing a movie, so my time is kinda limited. I'm also wading through the Nandub/vDub source trying to figure it out. I'd definetely be willing to work on improving Nandub, but at the moment, I probably don't have the skills to be of much use. On the other hand, if we are going to improve Nandub, we are going to need a clearly defined goal list of what improvements are needed and people to work on them. Setting up a website for this might be a very good idea.

LotionBoy

[Ookami]

First, I wrote this offline and in a hurry so bare with me.

Hello aleksander... I'm honored to be in your list of knowledgable people, but I'm really a Nandub newbie. As Nandub came out I was already somewhat bored in testing new programs for DVD backup (this is the main reason why I lost my motivation to update my freeware list), and I moved more toward capture and NLE... And I rather use CBR MP3 and use a rather high bitrate for my capture, so I use Nandub not very often (and if, I use mostly presets like the ones from Demi9od)...

I saw dozens of those "let's make a perfect backup program" project started and none of them lasted. Ideas a really easy to have, but you need the will and time to make it happen. The german speaking folks at Ultimateboard have shown this especially ;) (vielleicht wird's ja was aus dem Alleskoenner Projekt).

I hope that Nando is still somewhere around (maybe under another nick?), if not all the best for him (and if yes, all the best for him too)...

I like programs like FairUse, great UI, perfect readme, patient hobby programmer and so on.

My thoughts on this stuff:

Powerdivx - WOW,wow,wow! BlackSun when did that happened? Was I longer offline than I thought? Can you read my mind (at last someone is noticing the TMF format!)? Please, keep it up!

Project leader - Well, how about BlackSun? It seems he's motivated, has great amount of knowledge and of course he is a newbie (hehehe)...

So, what's the project? I mean, a new Nandub could be everything, from a new revolutionary way to cook eggs to a freeware GPL combo that uses Xvid...

BTW, "xiphrom" is a nice name. Hehehe.... I wonder if I will ever get mentioned for proposing the name to Xvid (and yes, my ego would love it :D )...

Well, take care y'all and watch out for more "..." and "(" from me.

Cheers,

Ookami.

[Doom9]

blacksun.. well.. I still believe in divx4 but with so much time passing by since I was first told about an upcoming p4 optimized release (along with the statement that an amd optimized version wouldn't be far off).. and since that time I haven't really heard or seen much from the DivX4 team. I submitted a few sample clips and pointed out a few issues that clearly need to be taken care off but atm it just feels like the development has stopped. The rate control is still far away from being perfect (I'm getting almost +-20mb doing 4 TV episodes per CD), the quantization for high frequencies is too low (resulting in ugly looking edges), and .. I just hear of no progress. What I know some new versions of DivX4 should be out by know (can't really say much more in public because I've promised not to tell).. but there's nothing. So.. one might think my fears I had when the first divx4 beta came out that the release was premature and my fears that the team would move on to programming drm solutions rather to try and improve the codec has finally come true.

and there's several open source projects that have proven to be very good.. just look at linux, lame, virtualdub, etc. Some people do have the motivation to carry on without any apparent gain for a really long time... can't say if I will have the energy to go on over such a long period myself.

[Emp3r0r]

I very much agree and hope we can bring together much talented people. In my opinion it isn't so much the coder's or the leaders job to make such a project successful... I believe it is more a combination of:
A) Solid Documentation
ex: Doom9's site
B) Community
ex: Doom9's site
C) "motivation to carry on without any apparent gain for a really long time"

C is the tough one but with B to keep people involved and A to get people involved (such as newbies like blacksun :)) I see no reason why such a project would not have an impact on the topic.

I think Nandub will still be around a while due to it's sheer usefulness and this is what we should strive for... Something new, something useful.

[Teegedeck]

I ask myself: [i]What do we want that DivX4 can't do anyway? Before you start throwing bricks at me, I just wanted to say that we should perhaps figure that out for a motivation - as this is important as pointed out. Of course I personally have some thoughts about what the goals of the*encoding* community could be from here on:

1) We want to influence bitrate-distribution manually for two-pass. Until compression doesn't get better (by about 30% I'd say), it still makes sense to set lower rates for credits and boost sequences that are demanding and/or important.
2) We want a really good 2-pass! (DivX4's isn't good)
3) We want a better compression/quality-ratio... (If that would be reached we could perhaps dump #1 and have an easier life...)
4) We want a codec that doesn't crash...
5) And of course we want all this to get faster.

OK, me personally thinks that

ad 1) GordianKnot with XviD basically already makes that possible.
ad 2) GordianKnot with XviD is close to that; if XviD could read .ecf-files and if TheWEF adjusted GKnot to predict filesizes for XviD's quantizers, it would be exactly like BlackSun proposed. [A really good two-pass will get much more complex though, once this adaptive quantization thing really works - with different quantizers for each MB.] And of course it is debatable whether GKnot is inferior to traditional curve-compression in Nandub (I don't think so).

Points 3) - 5) can only happen inside the codec.

ad 3) This means introducing B-frames, good adaptive quantization etc... With DivX4, we can just sit back and wait till it happens, with XviD some of you talented people could try and help speeding it up. It might take a really long time though - B-frames would also be very demanding on hardware, so I guess right now improving adaptive quantization could be a main task, and also users could contribute to this one. It seems once you get into psychological aspects of encoding, it takes very extensive testing till you get results -- if you judge it by the time it took LAME to arrive at the point where it is now (--alt-presets). So point 1) will probably remain important for a long time.
ad 4) Well... You know what I mean!
ad 5) Takes very specialised coding-skills

Well, I feel pretty much cheered up by the open-source examples doom9 mentioned, so I think there really is no reason to believe commercial codecs should by principle be better than what the community here (and over there at videocoding.de) can put together.

[Teegedeck]

By the way, Ookami apparantly was the very first to come up with the name 'XviD'...

[aleksander]

Well...
It's nice to see you all guys posting your opinions here...
This discussion is becoming really great.
Still hope for some other to post (DDogg, realSpiderman, Koepi - where are you guys??)
Koepi's post should be really interesting since he is the one who has done some Nandub hacks so he knows a lot about it.
I'm also glad to hear that LotionBoy is working on Nan/Vdub source...
I can see that you guys really count on Xvid. Since it's open source it should have a long future. But also remember what happened to OpenDivx after releasing Divx4. This project is not what it was before.
All we need right now is just a good list of goals and some coders (yep.... this can be hard do to...)
My initial idea was to encourage some guys to work with Nandub to make it more universal tool - to have the same level of control over other codecs as it has over divx 3.
But as I can see this can be problematic.
On the other hand....
Originally posted by TheWef

i know that nando had done beta-test versions that worked with divx4.

and
Originally posted by Teegedeck

I think external developers investing work into DivX4-applications/modifications are in danger of wasting their time (please don't feel offended) because they make themselves dependent on a commercial codec that A) can change without notice and make your changes naught (they are already leaving you out in the rain as you can read in one of BlackSun's posts), B) is closed source, so it is difficult AND illegal to change it.
If we could only have this version of Nandub and its documentation. It could be easier to work on a stable version.
Maybe it is hard to implement all the functions of Nandub to work with divx 4. But Nando has done it. Maybe it is not that hard at all or maybe he is such an exceptional coder that noone can be a match fo him. Or maybe both. I'm not saying that other coders are not good - No way! I'm just saying that maybe they need to be encouraged. How about the undying gratitude of thousands of people all over the world??? Isn't that enough???
Well...waiting for your opinions guys....
take care
aleksander

[BlackSun]

Originally posted by Doom9
[B]blacksun.. well.. I still believe in divx4 but with so much time passing by since I was first told about an upcoming p4 optimized release (along with the statement that an amd optimized version wouldn't be far off).. and since that time I haven't really heard or seen much from the DivX4 team. I submitted a few sample clips and pointed out a few issues that clearly need to be taken care off but atm it just feels like the development has stopped. The rate control is still far away from being perfect (I'm getting almost +-20mb doing 4 TV episodes per CD), the quantization for high frequencies is too low (resulting in ugly looking edges), and .. I just hear of no progress. What I know some new versions of DivX4 should be out by know (can't really say much more in public because I've promised not to tell).. but there's nothing. So.. one might think my fears I had when the first divx4 beta came out that the release was premature and my fears that the team would move on to programming drm solutions rather to try and improve the codec has finally come true.


Here the truth !! the quantization of the DivX 4 is simply far from being perfect and it seems they don't really care about this. When I told Sparky about this problem, he told me it is normal... :o
Now when I see that the 2_PASS_VBR code from DivX 4 is pretty strange (new quantizer decision based on a Rand() function), I think they are more interested in the feature list than the quality of the features.


and there's several open source projects that have proven to be very good.. just look at linux, lame, virtualdub, etc. Some people do have the motivation to carry on without any apparent gain for a really long time... can't say if I will have the energy to go on over such a long period myself.

I know, it begin to be hard, but I'm in love with PowerDivX :p

[BlackSun]

Originally posted by Ookami
First, I wrote this offline and in a hurry so bare with me.

Powerdivx - WOW,wow,wow! BlackSun when did that happened? Was I longer offline than I thought? Can you read my mind (at last someone is noticing the TMF format!)? Please, keep it up!


Thanks, if you are interested in giving us your opinion check my forum. Tronic use my board as the official TMF/MCS board so everything will be developped at this adress.


Project leader - Well, how about BlackSun? It seems he's motivated, has great amount of knowledge and of course he is a newbie (hehehe)...


I can't!! I am really busy with PowerDivX 4, TMF and XviD. Too bad I'll be happy to lead this project, but maybe aleksander is the right person for this ?! I am the perpetual newbie, ROTFL


BTW, "xiphrom" is a nice name. Hehehe.... I wonder if I will ever get mentioned for proposing the name to Xvid (and yes, my ego would love it :D )...


xiphrom is morphix's revenge :D

[Bugs Bunny]

Oh hey I see Oookie is here hey ya Oookie and of cohse Weffie baby and some of the other old bloke nicks I recognized. Oh lets PONTIFICATE (heh new big word I learned for 2002) about Da Foocher of da vid scene!!

One thing some of the noobs got to realize is that this is a hobby for most of the blokes here, something ya get into when it's still interesting, and ya get out when things get stale and ya get bored. It's not a "now and forever shall we be wedded" thing, so if ya expect continued developer support for some of the apps like Nandub and FU and whatever, well, don't hold your breath. The blokes have probably moved on, like with a fair number of other (non-programmer but still techie) blokes that used to hang around here and dispense a lot of interesting dialog. They've all done their bit, and we thank them for it, but things change, people move on.

And they move on because it's stale. Around the MPEG2AVI/Flask days (which was like, uh, last year), Divx tools suck, so there was a groundswell of interest to make it better. First we got the MM4 tools (which FU best exemplifies), then we got Nanadub, which heralded the advent of 2-pass VBR. But that's it. The quality are pretty good at this point, probably the limit for MPEG4, and there's not a lot you can do to improve it, so people move onto more interesting things. It's the same with the others. When the tools were still developing and in flux, there was a lot of talk on bugs and getting better quality. Now it's mostly idle chit-chat from clueless noobs who wants to know what interleave means.

I see there are still blokes waiting for the next rev of FU or Nandub. I can't figure out why they bother. FU & NDub are intimately tied to Divx 3.11a, which is a hacked and now orphaned codec. Because it's a hacked MS property, it will never see mainstream support. As soon as the other codecs reach its level of quality, it will disappear. The writing is on the wall.

As for retrofitting NDub for Divx 4, I'll ask, why? Divx 4 has a pretty good (and more friendly) 2-pass VBR already, why is NDub needed? Whatever faults Divx4 has right now is because of the codec and not the interface, and the codec is constantly changing, unlike 3.11's single version. NDub is a hack on a hacked codec. You can't have a codec hack when that codec doesn't stay constant.

Da Big Picture (or "My View from Under a Rock"):

The main thing with Divx/MPEG4 is because of the CD size limit of 700 megs. We want to copy ("backup" if you want to be politically correct) DVD vids onto PCs, and CDs were the only common and cheap medium available. MPEG1 was too low-res, and SVCD/MPEG2 ate up too much space (and still look sucky), so Divx fitted nicely. But all you blokes who follow hardware are aware that DVD recorders (in various flavors) are making their way into the mainstream, and before too long that 700 meg limit will be 4.7 gig limit, which means that we can copy DVDs w/o having to transcode to lower-quality MPEG4. The big thing will be MPEG2 editing/reauthoring tools. Why waste time converting to MPEG4 and losing quality, when DVD platters are (going to be) cheap?

BTW, Doom9 bubba was talking about MPEG4 having its uses for transporting vids online. I can understand his feelings, because he's into the vid scene. But for homies like moi who don't do a lot of trading (OK I download vids every once in a woil), all we're interested is zapping some DVD vids onto, er, backup copies that we get locally. MPEG4 will, I suppose, be de rigueur for the trading blokes, but for the MAINSTREAM peeps, MPEG4 will mainly be used for net streaming (from commercial sources), and it will compete with several other technologies, like Real and WMV.

I read recently that some big honcho companies (like Apple, Intel, Sony, and some others) got together and agreed upon a "standard" specification for MPEG4. Not sure what that means, but it certainly doesn't mean that Divx is the last word on MPEG4. Puter companies are hurting in this recession, and everybody is looking for that killer app to push growth. No surprise, AV fits the bill, so there'll be much more development on this front (along with wireless).

Anyway, be happy that there are decent tools available. Why clamor for Divx 4 tools when the codec itself still isn't that great?

[aleksander]

Originally posted by Bugs Bunny

As for retrofitting NDub for Divx 4, I'll ask, why? Divx 4 has a pretty good (and more friendly) 2-pass VBR already, why is NDub needed?
.....
Why clamor for Divx 4 tools when the codec itself still isn't that great?
Well...The main reason for developing Nandub was to give us a better control over the codec. Divx 3.11 itself sucked (have you seen movies made with divx3.11 Low-Motion???). Thanks to all the features that Nandub has, we are able to create rips of superior quality. Of course I agree that divx 4 2-pass is more user friendly (read: easy) but it doesn't mean that it cannot be improved. This codec has some improvements over divx 3 in terms of quality which, if controled by other programs, can give us even better quality rips. I agree that divx 4 itself isn't that great - but maybe thanks to tools like Nandub it can be better???
You're also taking about the fact, that it is pointless to work on MPEG-4 because DVD-writers are becoming cheaper and more popular. But still for another year or two they will be out of range of average people. And here's the field for making some improvements over divx 4 and its successors. I think there will be some people that, inspite of DVD-recorders being cheap, would like to have few movies on the DVD-discs instead of one (with ac3 sound, menus, maybe without those unnecessary bonuses). That's why we shouldn't give up on developing new codec and tools for them.
@ BlackSun - thanks for mentioning me as a project leader, but I really don't have the skills (I'm not a coder) to be such a person. Of course, if this project (actually idea right now) is going to work, I will participate in it as much as I can.
take care
aleksander

[BlackSun]

@ BlackSun - thanks for mentioning me as a project leader, but I really don't have the skills (I'm not a coder) to be such a person. Of course, if this project (actually idea right now) is going to work, I will participate in it as much as I can.

As long as you can motivate people, it is enough :D

[Teegedeck]

That was interesting reading, Bugs, but I don't agree with the underlying "why bother at all" attitude. Ever wondered why we are still hanging around this forum though DVD-burners will come someday? Well, I guess there's 2 possible answers:
1: I get the impression that most of us (with the exception of traders), but especially the coders, don't have the money to get the latest hardware.
2: It's our hobby...


Originally posted by Bugs Bunny

I read recently that some big honcho companies (like Apple, Intel, Sony, and some others) got together and agreed upon a "standard" specification for MPEG4. Not sure what that means [...]

...I think it means that there _might_ be standalone-players one day that can play what is in our archives, now, so we won't have to re-encode it all.
Different MPEG4-profiles will be used (even RealMedia goes for MPEG-4) but to bridge those differences can't be such a big challenge... I guess all of you heard that Cirrus-Logic is probably already developing a decoder on a chip? It also indicates that in the long run MPEG-2 doesn't have a future since it simply is inferior to MPEG-4. The industry certainly realized that, too.

Anyway, be happy that there are decent tools available. Why clamor for Divx 4 tools when the codec itself still isn't that great? Pretty damn right :) We clamor for a better codec.

Anyway, as for _me_ there is no DVD-burner waiting around the corner :( , I'm happy that XviD can give me the full quality and anamorphous resolution of a DVD on, say, 3 CDs, and I hope that the space needed will shrink down to 2 CDs - or one (well, probably not) -because I simply don't want to use 3 CDs...

[-h]

By the way, Ookami apparantly was the very first to come up with the name 'XviD'...

I've seen 'xvid' a few times on the internet, from sites pre-dating the XviD codec. Perhaps worryingly, these projects are video related..

Re: DivX4 this, DivX4 that..

I can only see DXN paying less and less attention to the ripping community - and why should they? This community will typically not pay for the tools it uses, whereas businesses will be queueing up to throw real money at a DRM-enabled, streaming MPEG4 solution. If my goal was making money, I know what I'd be focussing on.

Re: XviD status..

XviD still has a fair bit of breathing room as far as quality goes - gruel has noted that the current PMVfast ME code has a lot of room for improvement, as do things like inter/intra block choices. I don't think a 10% drop in file size is out of the question, perhaps greater for high-action media like anime, which murder the current intra/inter choice method.

As for 2-pass functionality, there's no reason the Nandub mechanics can't be used - anyone knowledgable in how it chooses its quantisers could port the functionality across in a matter of days. The code's all there - the documentation isn't ;)

Re: MPEG4 itself..

The MPEG4 spec is reaching maturity, and we are definitely going to see a number of MP4-related utilities from companies, no doubt of variable quality and worth. In addition to this, once the H.26L people have finished their work, it will be integrated as another method of video compression in the MPEG4 spec. It reportedly gives bitrates half that of simple profile codecs (i.e. DivX4/MP43/XviD).

Video encoding is definitely not going to sit still.

-h

[aleksander]

Originally posted by -h
....
Re: XviD status..
....
As for 2-pass functionality, there's no reason the Nandub mechanics can't be used - anyone knowledgable in how it chooses its quantisers could port the functionality across in a matter of days. The code's all there - the documentation isn't ;)

YES!! That's what I've been waiting for!
All we need is a coder...Xvid is making its progress and it's open source. So...
Anyone????
Koepi??? You know Nandub very well...
waiting...
take care
aleksander

[-h]

I guess I should've been more specific - there are a number of factors in Nandub that are inherently only useful for the Microsoft MPEG4x implementations - the "few days" figure would cover 90% of the functions that people actually use. A number of the settings seem to be more voodoo than anything else, working sometimes, blowing up others.

Anyway, someone familiar with Nandub's code (not just its usage) will have to see how much they feel is worth carrying over. It'd be easiest to make it an external app, instead of merging with XviD itself (which already saves/reads .stats files).

On the other hand, having the codec's source allows much greater control over encoding than Nandub ever did - entirely new 2-pass methods would most likely prove rewarding.

-h

[Ookami]

Sadly, I cannot post in the original thread. :( AFAIK, the admin is very busy, I'll post a new thread. If it'll work later I will try to merge the postings...

Original posting:

"I will read all those replies later this day, but to answer a few things I read...

@BlackSun

I will take a look at your forum, since I'm no programmer I can only throw in some real newbie stuff...

Keep on your good work. BTW, is Danni somewhere around? He's always been very motivated and helpfull (I will post the first really working step by step MS MPEG 4 manual by him these days, for all those nostalgic people ;) ).

@Bugs Bunny

Are you related to the guy named "BabySmurf" from the old forum? Your humor is similar :D . Ookie... ROTFL.

@Emp3r0r

I agree 100% with you...

@-h

Now to clear this Xvid stuff up... On the first Doom9 board (ezboard) some people asked for a nice name for a new codec that MAYBE will be developed... I came up with Xvid )-; . Steady came up with a few dozens of names etc. Sadly I didn't saved this thread, there were many good and funny names on it... If someone else came up with the same name, that only means he invented the laser, er, I mean... Well... You know. *grin*

But seriously, gdlblade was one of the first Doom9 members who really started pushing http://www.videocoding.de ... People like him should get the respect.

Seriously II: When will the project start?

Cheers,

Ookie."

[Teegedeck]

Again, shortly, or not so shortly, about Nandub for XviD: The gain from tailoring a Nandub for XviD probably would be much smaller than what aleksander might be expecting. The tremendous leap in quality back when Nandub came up was due to the fact that it introduced a two-pass-encoding - but DivX and XviD can do that already (and in contrast to DivX-4, filesize-predictability seems no problem for XviD if you use it in combination with GKnot and VDub) - and that it compensated the weaknesses of the DivX 3.11 codec: That M$ codec was never meant for HQ-video but for extreme compression in streaming media, so consequently Nandub had to force it onto lower compression by influencing its bit-bucket (gauge) and stopped it from compressing especially dark frames way too strongly. Both are weaknesses that modern MPEG-4 codecs don't have. All gain that can come from an external application (GKnot, Nandub) is that it might make a better decision on which frame has to get compressed at quantizer 3 and which frame has to get compressed at quantizer 4 than the decision the two-pass-algo, that by no doubt will get integrated into XviD, makes . And we can already pretty well incluence this decision by manipulating the curve with GKnot. So I'm afraid that ATM you better not expect too much.

IF, however, _adaptive quantization_ gets into the game, two-pass encoding can get really, really, really complex and I guess we'll have a new playing-field for tools that perhaps can interact with the codec so closly that they can incluence the choice of quantizer for every single macroblock. But we'll have to see whether we'd be reaching a considerable quality gain or if it doesn't look worse if we just leave it to the codec.

So what I meant to point at: It's the codec itself that deserves the most attention and work since the most considerable gain in quality can only come from the codec IMHO.

[Ookami]

> It's the codec itself that deserves the most attention and work since the most considerable gain in quality can only come from the codec IMHO.

Teegedeck, a very good point! IMO, if even the developers don't know for sure what feature will come (or even go ;) ), then first we should maybe wait for a stable beta release (like Isibaar once stated) before dreaming our heads off. Hihi.

[Doom9]

don't count out ms yet... check this out: http://www.microsoft.com/windows/windowsmedia/thirdgen/default.asp. I think this is what they demoed at CES. Of course.. we'll see how good the hype is once it's out... wm7 was really nothing new and wm8 was better, but still lacks behind nandub and divx4

[Doom9]

thread merging rocks :)

[Teegedeck]

Originally posted by Doom9
don't count out ms yet... check this out: http://www.microsoft.com/windows/windowsmedia/thirdgen/default.asp. I think this is what they demoed at CES. Of course.. we'll see how good the hype is once it's out... wm7 was really nothing new and wm8 was better, but still lacks behind nandub and divx4
That page apparantly has ceased to be... From what I've heard about it, WM9 will look great at low bitrates, but be restricted to VHS-quality. That's why Panasonic will support it in their DVD-standalones, I guess. It would fit the picture as we could already experience it with WM8: you can set the bitrate as high as you want but you don't get a quality gain. Damn, I'm hanging around the forum the whole day! I'm an Internet-junkie! Maybe I should apply for moderating something, too. ;)

[Doom9]

huh? ms doesn't like external links or something.. I went back one layer to http://www.microsoft.com/windows/windowsmedia/default.asp and from there I was able to access it.

anyways.. this will work: go to www.microsoft.com/windowsmedia (maybe you have to type that in your browser I dunno), then click on the 3rd gen windows media button ( a gree one)).. and there you go

btw.. I'm doing another ripper comparison and it takes so damned long (checking seamless branching titles is a major pain in the ass believe me.. you have to rip with each pbc, then split up the thing on your hd into cell levels using vstrip and compare line by line, byte by byte) so I'm glad I have somebody to talk to while I'm waiting for an operation to finish.

[Bugs Bunny]

>Well...The main reason for developing Nandub was to give us a better control over the codec. Divx 3.11 itself sucked (have you seen movies made with divx3.11 Low-Motion???).

And the reason Nandub gives ya such control is because it was intimately tied to the codec. You don't have that with Divx4, as it's under continued development. The only way it would work is if the Divx company blokes officially sanction and support such a tool. But being that it's their main IP asset, I'd rather doubt that would happen.

Divx the company is a business, with employees to pay and profits to be made. Then, who would they cater too, people who pay money, or people who don't? It's simple, they go after the paying blokes. What do paying blokes want? Streaming! Do they want quantization/size control? Not much, since size control is only because of the CD size limit, and streaming vids don't bother with that. Am I surprised that size control still don't work roit in Divx4? Nope, since that's not their priority. Their priority is to make not only the codec, but vid servers with reliable isochronous transfers (oooh another big word...)

>you're also taking about the fact, that it is pointless to work on MPEG-4 because DVD-writers are becoming cheaper and more popular. But still for another year or two they will be out of range of average people.

Not really. The earlier DVD-RW drives are hitting the $300 mark, and the +RW are at $400, and they're just out of the gate. The magic number is $200, the level considered for mainstream consumption. We'll see that price point this year, probably around fall.

>Ad here's the field for making some improvements over divx 4 and its successors. I think there will be some people that, inspite of DVD-recorders being cheap, would like to have few movies on the DVD-discs instead of one (with ac3 sound, menus, maybe without those unnecessary bonuses).

Why?

Two things:

1) The main reason VCDs and SVCDs are still big (bigger than Divx in fact) is because you can play it on standalone DVD players. Divx doesn't have that important advantage, and never will. Commercial players will incorporate to MS WMV format (some have already committed to WMA), and may be Real, but not to some small company's codec that isn't widely supported. You may have multiple vids on one DVD, but it's no good when you are limited to playing it on the computer.

2) Media cost right now is still relatively expensive (around $5 for a DVD-R). At roll-out, they were about $10/disk. Note that this is still cheaper than the starting cost for CD-R ($20) and CD-RW ($25) when they first appeared. And given the interest and level of corporate commitment, I see the same downward price curve, which means that DVD-R will likely reach the $2 mark, may be even lower, this year. Once it gets below that point, why bother trying to fit multiple vids on one platter?

>That's why we shouldn't give up on developing new codec and tools for them.

Well nobody said we should give up on making new toys. But a codec is a "big deal", beyond what any one bloke can crank out in his spare time. And remember that Divx 3.11 is already pretty good, so a new codec has to be substantially better than that for people to switch (which is why I haven't bothered with Divx4).

I'm not saying the scene is dead. I'm saying that it doesn't exist in a vacuum, and industry developments do have impact on the hobbyist blokes. Commercial interests in digital AV is strong, and changes are afoot. Divx was good (still is good), and it's good enough to keep using until the smoke clears and we see what else comes out.

BTW, the scene never dies, it just rides on top of the tech wave and morphs into whatever is the next cool thing. May be 10-20 years from now, we'd all be playing codecs for holos. Dang, how old will I be then... I hope I can still have sex. Life has no meaning if ya lose the ability to screw, is wot my grandpa said. He's dead.

[Teegedeck]

>>Well nobody said we should give up on making new toys. But a codec is a "big deal", beyond what any one bloke can crank out in his spare time.<<
Well, I guess you're _wrooong_! :D
And that's part of what I like about those 'spare time coders'.

>>And remember that Divx 3.11 is already pretty good, so a new codec has to be substantially better than that for people to switch (which is why I haven't bothered with Divx4). <<
So you didn't try DivX4 or XviD? Well, I was very reluctant to switch over, too, and still used DivX3.11 for my 'HQ'-encodings while I watched XviD's development and tested it. But then I got really annoyed by the fact that you can feed as many bits into DivX3.11 as you like and the damn thing STILL turns 'plain' structures, like a white wall or a desert into masses of twinkling blocks! Reason enough to switch.

[gldblade]

Eh... I'm a little late in joining this discussion, but here goes nothing.

>But seriously, gdlblade was one of the first Doom9 members who really started pushing http://www.videocoding.de ... People like him should get the respect.

Well... I know I was encouraging people here and there and I made quite a few posts at the forum, but it's nothing compared to what people do now. Suxen_drol, the_ether, and several others have taken over project and contributed much more than I ever did. And, of course, Isibaar, who had the idea to create XviD in the first place.

>So you didn't try DivX4 or XviD? Well, I was very reluctant to switch over, too, and still used DivX3.11 for my 'HQ'-encodings while I watched XviD's development and tested it. But then I got really annoyed by the fact that you can feed as many bits into DivX3.11 as you like and the damn thing STILL turns 'plain' structures, like a white wall or a desert into masses of twinkling blocks! Reason enough to switch.

I like the idea of integrating Nandub with XviD. However, the first question that comes to mind is if XviD is similar enough to DivX 3.11 to convert over. For example, XviD is able to change DRFs between macroblocks. That's something DivX simply could not do. This particular example complicates things a bit. Nandub was built to work on a frame-by-frame level. It will have to be changed to work between macroblocks to work with XviD. Or maybe some of XviD's special capabilities will have to be ignored (for now).

BTW, the developers at www.xvid.org seem to dislike motion measurement. Motion measurement was something big with Nandub. Practically every important decision was based on bits and motion. Of course, there were other things such as luma correction and the like, but bits and motion were the most important. One of those may be missing, depending on how you plan to implement this. If you plan to have Nandub be the encoder and call XviD, Nandub may be able to do it's own motion measurements. However, if you plan to actually put Nandub's code into XviD (in the two-pass mode), motion measurement will have to be implemented. Which shouldn't be too hard to do considering XviD is open-source, which is the beauty of it. Just a heads up.

Which is an interesting question. How do you plan to implement this? Put Nandub's code into XviD or have Nandub call XviD? They both have they're benefits. The first choice will give you a moderately easy-to-use codec that will be accessible in any encoder. That would be very nice. But it would probably be harder to accomplish from the programmer's viewpoint. And the codec configuration window would get quite large, though you could probably have a simple and advanced mode. The second choice will be easier to complete. It will not require the implementation of motion measurement within XviD. And Nandub is, of course, a great program. However, XviD would only be accessible in one encoder.

I'd like to see some of Nandub's code in XviD, but I don't think all the features will survive the conversion...

>you're also taking about the fact, that it is pointless to work on MPEG-4 because DVD-writers are becoming cheaper and more popular. But still for another year or two they will be out of range of average people.

You're forgetting that not everyone uses MPEG-4 to rip off your local DVD rental store :devil:. (BTW, when was that smiley introduced? Never noticed that one before...) What about streaming? Or commercial uses of video compression?

>The main reason VCDs and SVCDs are still big (bigger than Divx in fact) is because you can play it on standalone DVD players.

Uh, I don't see how that's true. VCDs and SVCDs aren't big specifically because they play on standalone DVD players. It's just that they use MPEG-1 and MPEG-2, which DVD players just happen to decode just fine.

>BTW, the scene never dies, it just rides on top of the tech wave and morphs into whatever is the next cool thing. May be 10-20 years from now, we'd all be playing codecs for holos. Dang, how old will I be then... I hope I can still have sex. Life has no meaning if ya lose the ability to screw, is wot my grandpa said. He's dead.

:devil:

Ten to twenty years from now seems good to me. I'm only in my mid-teens. I've got all the time in the world... :)

[Teegedeck]

Originally posted by gldblade
Eh... I'm a little late in joining this discussion, but here goes nothing. Nice to see you!


>But seriously, gdlblade was one of the first Doom9 members who really started pushing http://www.videocoding.de ... People like him should get the respect.

Well... I know I was encouraging people here and there and I made quite a few posts at the forum, but it's nothing compared to what people do now. Suxen_drol, the_ether, and several others have taken over project and contributed much more than I ever did. And, of course, Isibaar, who had the idea to create XviD in the first place.
hey, I'm proud to be# 8 on the videcoding-forum joined members list ('Maelstrom')..., you only rank 13! ;D


BTW, the developers at www.xvid.org seem to dislike motion measurement.
Seriously, if you compare encodes with and without motion-influenced bit-redistribution, you'll see that you CAN actually detect that stronger compression in hi-motion scenes, and more importantly there is no arguing the fact that it really leads to problems when only the background is moving but the stronger compression is also affecting the calm center of the scene. I proposed something like 'motion-masking' as an additon to lumi-masking on the xvid-forum but Isibaar convinced me that even if you only compress the macroblocks where there is motion stronger, you still can detect that. So for high-quality encoding he proposed that the macroblocks with much motion in them should only get 'softened' in a future version of XviD. (BTW, even Nando recommended not to use motion-based compression back then.)


Ten to twenty years from now seems good to me. I'm only in my mid-teens. I've got all the time in the world... :)
You... YOUTH!

[int 21h]

Originally posted by gldblade
Uh, I don't see how that's true. VCDs and SVCDs aren't big specifically because they play on standalone DVD players. It's just that they use MPEG-1 and MPEG-2, which DVD players just happen to decode just fine.


That's why many many of those DVD Players are marketed specifically with their ability to play MP3, SVCD, and VCD.

[gldblade]

>hey, I'm proud to be# 8 on the videcoding-forum joined members list ('Maelstrom')..., you only rank 13! ;D

Hey, I would have joined sooner if I didn't loose the thread Isibaar posted on www.projectmayo.com when he started the project. If I remember correctly, I was the first or second person to reply. :)

>Seriously, if you compare encodes with and without motion-influenced bit-redistribution, you'll see that you CAN actually detect that stronger compression in hi-motion scenes, and more importantly there is no arguing the fact that it really leads to problems when only the background is moving but the stronger compression is also affecting the calm center of the scene. I proposed something like 'motion-masking' as an additon to lumi-masking on the xvid-forum but Isibaar convinced me that even if you only compress the macroblocks where there is motion stronger, you still can detect that. So for high-quality encoding he proposed that the macroblocks with much motion in them should only get 'softened' in a future version of XviD. (BTW, even Nando recommended not to use motion-based compression back then.)

Hmm... I hope the "soften" effect won't be too strong like in luma-masking. Luma-masking definately needs a lot of tweaking...

I'd like to see how Isibaar plans on implementing this. For one thing, how do you figure out how much softer you make an image based on the frame's compressibility? This might also need quite a bit of tweaking.

[aleksander]

Originally posted by TheWEF
i know that nando had done beta-test versions that worked with divx4.
pheak404 aka vimvam aka int21h (on this board) is still around.
you could ask him if he has the source.
wef.

@ int21h - I'm glad you have joined our discussion. Well..What about this beta-version of Nandub that worked with Divx4??? That was my initial idea, to have it working that way. But many posts have been posted (which is very cool - that's how it supposed to be) and this idea has changed (which is also cool, cause maybe there will be something going on).We're thinking about using Nandub for Xvid. Check this out:
Originally posted by gldblade
...
Which is an interesting question. How do you plan to implement this? Put Nandub's code into XviD or have Nandub call XviD? They both have they're benefits. The first choice will give you a moderately easy-to-use codec that will be accessible in any encoder. That would be very nice. But it would probably be harder to accomplish from the programmer's viewpoint. And the codec configuration window would get quite large, though you could probably have a simple and advanced mode. The second choice will be easier to complete. It will not require the implementation of motion measurement within XviD. And Nandub is, of course, a great program. However, XviD would only be accessible in one encoder.

I like this idea very much. To be more precise - the very first one mostly. We will not be limited to one tool - Nandub. But as you have pointed out - this may be hard to acheive. On the other hand, if we will be able to make Nandub to use Xvid, there will be no contraindication to make it to use other codecs as well (it should be easier at least).

@ everybody posting here - I just love the way, how this discussion is being carried out... Keep it that way!!!

take care
aleksander

[Teegedeck]

Originally posted by gldblade

Hey, I would have joined sooner if I didn't loose the thread Isibaar posted on www.projectmayo.com when he started the project. If I remember correctly, I was the first or second person to reply. :)
*sigh* You've won. -_-


Hmm... I hope the "soften" effect won't be too strong like in luma-masking. Luma-masking definately needs a lot of tweaking...

I'd like to see how Isibaar plans on implementing this. For one thing, how do you figure out how much softer you make an image based on the frame's compressibility? This might also need quite a bit of tweaking.
Adaptive quantization seems the main task for our heroes over there, once they got a stable beta release. For users, at least Isibaar planned an intuitive front-end to it, where a slider will influence different threshholds for adaptive quantization, maybe there also will be 'expert' settings.

[-h]

Re: Microsoft's "Corona" - considering they said WMV8 was 20% better than WMV7, I assume Corona's additional 20% improvement is just an (even more) savage prefilter, certainly not any algorithm improvements.

As for Nandub integration with XviD, the easiest (and perhaps most workable) way would be to create an external program which edits XviD's .stats output, or whatever extended format gets created. We wouldn't be frantically updating the vfw CVS tree, and the functionality would be identical (could still be used by any VfW-capable program). We also wouldn't be limited to whatever 2-pass code gets built into XviD, but would allow any number of external tools to edit the 2-pass process. Forcing such functionality into the codec could also trigger forking..

In short, the intimate tying of a Nandub-like processor to a codec is only workable if the codec is not in development - otherwise, any small change in the codec (nevermind the huge changes happening at the moment) require re-tweaking and re-coding to reflect the changes. Abstracting the internal changes with something like a .stats file makes maintaining the code much much easier, and gives people more time to actually find useful settings.

The "psychovisual" properties of XviD, which will mostly be motion masking, temporal masking and luminance masking, are all going to require extensive subjective tests to find agreeable settings as well. Also of interest will be whether smoothing with a low quantiser is more visually acceptable than no smoothing with a high quantiser (current lumi masking system), or a combination thereof.

-h

[Teegedeck]

Originally posted by -h


The "psychovisual" properties of XviD, which will mostly be motion masking, temporal masking and luminance masking, are all going to require extensive subjective tests to find agreeable settings as well.


I'm looking forward to years of testing... ^_-
No, seriously, what makes me so happy about this new codec is not only that it's better but that there is something to test and to talk about with the developers - and the psychovisual thing really is something that can be tweaked to hell and back. XviD WILL be the LAME of video-encoding. :) Right?

[gldblade]

>*sigh* You've won. -_-

Great! So what did I win? :) Anyway, back to the main discussion...

>The "psychovisual" properties of XviD, which will mostly be motion masking, temporal masking and luminance masking, are all going to require extensive subjective tests to find agreeable settings as well. Also of interest will be whether smoothing with a low quantiser is more visually acceptable than no smoothing with a high quantiser (current lumi masking system), or a combination thereof.
>No, seriously, what makes me so happy about this new codec is not only that it's better but that there is something to test and to talk about with the developers - and the psychovisual thing really is something that can be tweaked to hell and back.

This process might take a LONG time... The problem with this is that who is going to visually test the codec? Obviously the developers can, but there's only so much they can do at once. If we have the general public visually test the codec, the reports they send back will have to be validated, which may become a big annoyance. Maybe we could set up some sort of feedback page on the XviD website.

Of course, this is off in the future, so nothing really to worry about for now.

[Teegedeck]

This process might take a LONG time... The problem with this is that who is going to visually test the codec? Obviously the developers can, but there's only so much they can do at once. If we have the general public visually test the codec, the reports they send back will have to be validated, which may become a big annoyance. Maybe we could set up some sort of feedback page on the XviD website..

Yes, I think this might be a good idea. We need lots of discussions among users with good vision 8) like LAME had a lot of discussion among those with trained ears... But I think the discussions shouldn't take place at videocoding .de. It could soon get frustrating if the developers at videocoding.de had to validate contradictory opinions of users - no, there has to be some pre-filtering of opinions on the users' side before they ask for changes. I think we might even get competing 'presets' for the 'psychovisual engine' of the codec (like the LAME-presets) and testers will have to decide which presets they like best. ...I think I've had enough vine for today, I'm starting to have visions... *_-

[-h]

The tweaking part could be a pain, depending on how it's implemented. As long as the major masking parameters are configurable at runtime, people can try whatever they feel like to find useful settings, without bugging the development process. It would also allow people to compare results just by posting short clips over here, which can be evaluated by other users with the same goal.

Just looking at LAME - it took this many years until someone sat back and actually coded up some tweaked presets, despite the huge amount of talent actively coding. Developers usually like developing, not tuning from huge testing sessions. That's where the users come in :)

-h

[Ookami]

And we should not forget that "good quality" is relative...

Just look at the bashings and the politics in the MP3 scene...

So, always have in mind. It should make fun, if not somethings wrong :) .

Cheers...

Originally posted by -h
The tweaking part could be a pain, depending on how it's implemented. As long as the major masking parameters are configurable at runtime, people can try whatever they feel like to find useful settings, without bugging the development process. It would also allow people to compare results just by posting short clips over here, which can be evaluated by other users with the same goal.

Just looking at LAME - it took this many years until someone sat back and actually coded up some tweaked presets, despite the huge amount of talent actively coding. Developers usually like developing, not tuning from huge testing sessions. That's where the users come in :)

-h

[dragoman]

Hi,

I would just like to add my 2 cents about what the future of the Mpeg4 standard is. Remember when mp3 was first becoming mainstream in it's use on the internet?

I was a regular on the mp3.com bboard back then and everyone was full of speculations about the future of the format. But one thing that was almost universal was the fact that the major labels (the big 5) would NEVER support mp3.

It was (at that time) absolutely out of the question. Never happen.

But it did.

Now you have half-a-dozen mp3-capable portable cd-players. Hell, I remember when the mere rumor of such a player would cause hordes of people to search for the details and pre-order such a machine.

Now almost every single dvd-player manufacturer is scrambling to put out dvd players that play mp3-encoded cd's.

Car decks are available, one from Sony itself!!! that play mp3 discs (I have one myself, the Aiwa CDC-Mp3).

Judging by this precedent, I would say the idea of a standalone dvd player supporting divx discs is not too far-fetched.

Companies like Apex Digital Inc. are pioneers in dvd-player development. They were the first to introduce mp3-capable dvd players, and I predict they will be the first to introduce a divx-capable one as well.

Certain companies cater to the small, tech-oriented group of people who will pay more than the average consumer for a new technology. I know I count myself as part of this group (I purchased a Diamond Rio 300, 32MB version for 300 bucks....now you can get it for 30).

Audio encoding has moved beyond mp3, no question. There are much better quality ways to compress digital audio. But most companies are moving forward on mp3-compatibility. Why? Because it's popular.

DivX is popular as well, although not as popular as mp3. But it will get there. I am sure we will see more lawsuits from the MPAA in the future as the movie studios try to control their stranglehold on the market. Once they fail, like the RIAA and record labels, they will probably embrace DivX and related technology. It's almost a guarantee.

dragoman

[Bugs Bunny]

>I was a regular on the mp3.com bboard back then and everyone was full of speculations about the future of the format. But one thing that was almost universal was the fact that the major labels (the big 5) would NEVER support mp3.

>Now you have half-a-dozen mp3-capable portable cd-players. Hell, I remember when the mere rumor of such a player would cause hordes of people to search for the details and pre-order such a machine.

It's not a valid comparison. MP3 files are small. (Full length) movies are large. An average movie is roughly 200 times larger than an average MP3. Transferring an MP3 via Internet is feasible on a dial-up. Transferring a movie is only feasible if you have broadband--and that's only for download. Uploading is usually capped, so it takes a very dedicated uploader (or one with more than the consumer-type cable/DSL connection) to upload movies.

Moreover, because of the history of Napster, the MPAA has been vastly more aggressive than the RIAA in going after uploaders, to the extent of pursuing individuals by threatening their ISP access (it works). All of the P2P Napster derivatives are now under legal pressure. While I'm well aware that one can get movies from the net from various sources, the movies available are almost invariably of low quality, and the selection limited. While MP3 is good enough to qualify as CD-quality, the pirated movies from the net can in no way compare to DVD releases.

>Judging by this precedent, I would say the idea of a standalone dvd player supporting divx discs is not too far-fetched.

This will never happen, if only because of one very important reason: Divx is Microsoft's property, and they will never relent to license a hack, when that hack competes with their official solution, WMV.

That said, as has been posted by another, MPEG4 will likely become a de jure standard for net videos, simply because it is an open standard, takes less bandwidth than MPEG2, and as we can judge from Divx, is good enough for a "lo-fi" solution. And as MPEG4 becomes more common, then we'll see more DVD players incorporating it (as had already from Sigma). This will happen when streaming video becomes viable, which won't be for a long while yet. (Divx 3.x movies will not be compatible, although as Doom9 has commented, future Divx codecs may be rewritten to comply with the MPEG4 standard.)

>Companies like Apex Digital Inc. are pioneers in dvd-player development. They were the first to introduce mp3-capable dvd players, and I predict they will be the first to introduce a divx-capable one as well.

MP3 can be licensed from Fraunhofer. Divx is a MS hack, and cannot be used legally. Therefore, it will never appear in a commercial product.

>Audio encoding has moved beyond mp3, no question. There are much better quality ways to compress digital audio. But most companies are moving forward on mp3-compatibility. Why? Because it's popular.

Only MP3 Pro is backward compatible with MP3, and judging from initial acceptance rate, it doesn't fare any better than the other umpteenth audio codecs out there. Perhaps if FH drop their royalty rate, they may have a chance.

There are two separate markets to be considered here: online video (which will likely gravitate toward MPEG4), and in-the-home digital AV (which will be based on MPEG2, because of the availability of DVD players and recorders). The first (online video) won't be mainstream for a number of years yet. The second will be here later this year, and next.

[-h]

DivX is popular as well, although not as popular as mp3. But it will get there. I am sure we will see more lawsuits from the MPAA in the future as the movie studios try to control their stranglehold on the market. Once they fail, like the RIAA and record labels, they will probably embrace DivX and related technology. It's almost a guarantee.

MPEG4 already has sizable industry backing - a lot of companies are just waiting for the spec/licensing to be finalised. Have a look over www.m4if.org (http://www.m4if.org/) for some supporters.

Most of the products will actually want MPEG4 standard input - that is, MP4 streams, instead of AVI files. You can somewhat convert an avi to a standard MP4 stream with mpeg4ip, but I'm hoping to see some more convenient tools in the future. There'll certainly be demand for it.

-h

[dragoman]

Hi,

I was not trying to make a comparison between mp3 and divx based on their filesizes.....not sure where this came in.

I was simply pointing out that at least to me, there is a precedent involving a format for digital media that was popular on the net and not with "the establishment."

The fact that these companies who were originally against any form of mp3 at all (and now are using the technology instead of fighting it).

This is all I was trying to say will happen with DivX. Perhaps you are right that the original "DivX", the hacked codec from Microsoft will not be implemented in a commerical DVD player.

But support for other, open source codecs is legal and I think, probable. Thanks to the advent of upgradeable firmware, DVD players will become more versatile and people will be able to upload different codecs to their players.

My point is I don't think the format is "dead" and worth giving up. It will be used for some time to come.

dragoman

[-h]

Unless Microsoft has patents on the parts of MP43 which differ from the MPEG4 spec, they can't stop anyone from creating their own player and integrating it into a hardware device.

The players that have been created so far (ffmpeg, divx4mac, xvid, etc.) are the result of reverse engineering, and are perfectly legal (as long as you pay the MPEG4 patent fees). Microsoft can't object to this at all - they wouldn't have a legal leg to stand on.

-h

[Bugs Bunny]

>I was not trying to make a comparison between mp3 and divx based on their filesizes.....not sure where this came in.

You're saying that MP3 got mainstreamed despite non-support from the industry, and Divx will be the same. I'm saying it's not the same, because the vastly larger bandwidth requirement excludes the possibility of Divx ever attaining critical mass that MP3 had. Remember that MP3 was alive and thriving in the "underground" for a long while, yet it took Napster to bring it into the mainstream consciousness. Napster is now dead, and all of its kinds are under scrutiny and legal pressure. There is no mainstream vehicle to reach the masses (read: casual users). Even if there were, there is no pipe fat enough to enable movie P2P sharing for the masses.

>I was simply pointing out that at least to me, there is a precedent involving a format for digital media that was popular on the net and not with "the establishment."

Sure there is, but it's not applicable in this case, as above.

>But support for other, open source codecs is legal and I think, probable. Thanks to the advent of upgradeable firmware, DVD players will become more versatile and people will be able to upload different codecs to their players.

Because something is possible, doesn't mean it will be done. Codec support only happens when it makes business sense, i.e. if it sells more players, not because the codec is available. Look at Ogg Vorbis or AAC or MP3 Pro. They are available, so why aren't they implemented?

>My point is I don't think the format is "dead" and worth giving up. It will be used for some time to come.

Divx isn't dead, but then it was never "alive" as far as the mass is concerned.

>MPEG4 already has sizable industry backing - a lot of companies are just waiting for the spec/licensing to be finalised. Have a look over www.m4if.org for some supporters.

MPEG4 has an uphill battle. The other two entrenched proprietary solutions, WMV and Real, are already ahead of it in implementation, mindshare, and support, and are on par or better in terms of quality. They are also under continued development. The reason that Apple endorsed MPEG4 over their QuickTime is that they recognized they have no chance in hell of ever pushing QT against the leaders. The "support" shown for MPEG4 is from the sideline players who want a piece of the pie w/o having to pay royalties to Real or MS. It's a three-way race, and my observation is that MS has the inside track.

>The players that have been created so far (ffmpeg, divx4mac, xvid, etc.) are the result of reverse engineering, and are perfectly legal (as long as you pay the MPEG4 patent fees). Microsoft can't object to this at all - they wouldn't have a legal leg to stand on.

Reverse engineering aren't "perfectly legal," if a specific product was used in the process. You have an interesting, and dare I say, unique view of the legality involved.

Regardless, what's legal or what's not is determined by how much money you have to pay the lawyers. If MS comes after you with its billions, it wouldn't matter if God was on your side, unless he has a lot of moolah to loan to you.

[Teegedeck]

RealNetworks is going for MEPG-4. Look it up on their homepage or somewhere on this forum.

MPEG-4 already is an industrial standard, not an outsider format. This is getting tedious...MPEG-4, for example, will be used for coming video-cellulars in Japan and surely will also come to Europe for UMTS mobiles.

[Doom9]

wmv is based on mpeg4 though it's not really specs compatible... and why the content industry might like it because if it's DRM (which has been cracked already afaik) it is in no way as good as nandub and divx4 (you can read about my comparision on my page.. wm8 was slightly superior at the time it was released which was before nandub and divx4).. from what I'm hearing (and I'm soon going to verify that) xvid is better as well. It will be a long time until any format truly makes it to mainstream... only the future can tell which format is eventually going to win.

but I'd bet it will be a true mpeg-4 implementation.. this is an open stardard (there are royalties but sadly enough that's a given for many formats these days) and many companies are working on solutions for it. mobile phones and settop boxes are only 2 very interesting areas where mpeg4 will soon show up.

one more thing.. divx4 and xvid reverse engineering product? don't think so... maybe the divx3 playback part is but DXn is a US based company and I'm sure they checked the legal situation before releasing anything that could get them into trouble... as for xvid.. it's an open source codec (or should I say.. an mpeg4 implementation for academical purposes).. history has shown that you don't even have to pay royalties (hint: lame) as long as you don't offer binaries.. and nowhere on the xvid page will you ever be able to find such a thing

[Ookami]

According to sources like c't, MS MPEG 4 has very little in common with the "real" MPEG4. Also, I remember when "Rob from Ligos" wrote on Oleg's UBB board that he wonders that MS has released this codec as in that time the MPEG4 specs wasn't even near completion...

And isn't it true that MPEG4 and MPEG4 don't have to be the same thing?

If I'm wrong, please, correct me.

If I were a MS exec I would tell my programmers to make a Divx ;-) player hack for the X-box and release it like a new MoRE release (hehehe...), so I the X-box would finally wipe out all the consoles out there ;) .

Cheers,

Ookami.

Originally posted by -h
Unless Microsoft has patents on the parts of MP43 which differ from the MPEG4 spec, they can't stop anyone from creating their own player and integrating it into a hardware device.

-h

[-h]

Because something is possible, doesn't mean it will be done. Codec support only happens when it makes business sense, i.e. if it sells more players, not because the codec is available. Look at Ogg Vorbis or AAC or MP3 Pro. They are available, so why aren't they implemented?

Ogg Vorbis hasn't been implemented due to the stigma attached with supporting a non-final-release codec (not to mention so reliant on floating point math) in hardware - we're actually starting to see the hardware manufacturers (Ria Volt) support it in firmware upgrades. AAC is not an end-user format - just ask Mr. Dimkovic if you don't believe me. MP3 Pro is quite a poor format, rarely used by anyone, and has licensing fees to worry about.

There's a difference between implementing any old codec they feel like, and implementing ones that are both feasible and cheap.

MPEG4 has an uphill battle. The other two entrenched proprietary solutions, WMV and Real, are already ahead of it in implementation, mindshare, and support, and are on par or better in terms of quality.

There are already a number of hardware en/decoding chips for MPEG4 - can't say the same for WMV/Real. As mentioned, Real have thrown their support behind MPEG4 as well. Then you have the fact that there are actual standards behind MPEG4, and I can see why people are half keen on supporting it. The quality of the competitors is far far below what H.26L is capable of, and this is what the implementation people are going to focus on.

Reverse engineering aren't "perfectly legal," if a specific product was used in the process. You have an interesting, and dare I say, unique view of the legality involved.

I don't understand this point. The file format has been reverse engineered - Project Sparky, DivX4, ffmpeg and XviD (not public yet) all have legal bitstream decoders. Microsoft can not claim trade secrets, copyright violations or patent infringements, because they don't hold any over the file format spec that people outside microsoft have written themselves. The only hinderance is that MSMPEG4 uses patents that will become part of the MPEG4-LA group's pool, so anyone using the open source decoders will have to sort out licensing agreements with them. But not microsoft. It'd be like microsoft suing the WINE group.

-h

[gldblade]

>The file format has been reverse engineered

It has?! Or do you mean "The file format hasn't been reverse engineered."

[-h]

It has?! Or do you mean "The file format hasn't been reverse engineered."

The various open source decoders (and DivX4 as well) aren't using any microsoft code - they figured it out themselves. Heck ffmpeg can even encode to MP43.

The main differences between MPEG4 and MSMPEG4 turned out to be VLCs and MB headers, if I recall correctly.

-h

[gldblade]

Oops, I thought you meant something else completely...

[cofferscuffs]

Back to a post on page 1 of this thread:
DivX 4 (aka OpenDivX/Project Mayo etc) support is IN Nandub... open up it's source code - it was just never finished and implemented. BUT, even if Nando did do tests with DivX 4, he would of been dealing with a very old version of the codec... it has changed fast in a short time.

What I would love to see is XviD support in Nandub, but alas all good things (like Nandub) must end....

EDIT: Damn!! I just realised that I just posted what 10 ppls had just said.... I really need some sleep :o.

[Bugs Bunny]

Before I start replying I'd just want to say this: The online video industry hasn't happened, just like the online audio industry hasn't happened. Nobody knows how it's going to go, so what we're doing here is only guessing. That means there is no right and wrong, that I can be wrong or the blokes I'm arguing against can be wrong, or we can all be wrong. Realize this before we get too vehement about the issue.

Re: Divx (or MPEG4) in standalones -- Here, I think some are thinking more like techies (probably because that's what they are) rather than like a business. Tech guys do things just because it's possible to do it. Business guys do things only if it makes them money. Formats get implemented into players not because they're available or good, but only if the companies can sell more boxes--that's to say, if there's demand. The first mainstream MP3 portable (the Diamond Rio) only gets into the market after the Napster craze was well under way, raising the potential demand to millions of users. Why do DVD players have VCD & SVCD format included? Because it's in demand in the Asian markets, and it doesn't cost anything to leave it in for the US/Euro markets. Is there a similar demand for Divx or MPEG4 format, other than the few thousands or tens of thousands movie traders on the net? No, there isn't. Then, is there any sign that Divx or MPEG4 will generate such demand in the foreseeable future? Does the man on the street even know what Divx/MPEG4 is, let alone enthuse about it? No, and no. Until we can honestly say yes to these questions, then any argument about legality or quality or whatever is simply moot. No demand = no hardware product.

Re: Quality -- Doom9 has commented, and I'm sure we'd all agree, that the Divx codec is superior to Real & WMV in visual quality. However, that only comprises a small aspect of the overall picture. Divx is better when doing CD-sized movies, but is it scalable to still look good at lower bitrates? How is its streaming? How reliable is its server software? How is its DRM protection? What kind of technical support does the company offer? These are all the kinds of questions a business would ask, whereas a techie would only care about how the picture looks. In all of these areas Divx/MPEG4 is still largely an unknown, whereas Real & WMV are already in implementation mode. By "quality," I'm referring to quality of service & support, and not just quality of the picture. At this point in time, MS is the frontrunner by virtue of its stranglehold on the viewing platform and its established infrastructure. Divx 4.x is simply another codec from a startup in the sea of startups. Standardized MPEG4 is still a format on paper, with no concrete implementation.

Re: Industry standard -- Anybody who follows the tech industry long enough, will soon learn to disregard "announcements" (hype by another name) until he can see the actual products on the shelf. It's nice to hear Real announcing support for standard MPEG4, but given their entire of products is based on their closed codecs, and they've been doing pretty well up to now, it's hard to imagine they would relinquish their edge unless they are being driven to that point by MS' domination. And MS hasn't dominated yet. My interpretation is that it's just the usual lip service about "embracing open standard" that every company mouths. Again, until I can see a Real stream being played by a standard MPEG4 player, then it's all just more hype. Heck, there isn't even a standard MPEG4 player yet.

Re: Legality of Divx players -- I'm not a lawyer, and from what I can see, none of us here are practicing IP rights lawyers either, so getting into the finer points of legality would be talking out of our ass, so I won't. What I do know is that MS has successfully prevented other companies from using their codecs without their consent (read: license). From all the instances I've read, none has gone so far as to reach the courts. All that happened is that MS lawyers send a cease-and-desist, and the target party backed off. Nobody is dumb enough to up against Microsoft unless they absolutely have to. Knowing what we know about MS' tendency to litigate to protect their codecs, what do you think the chances are of MS letting some hardware company to market Divx-capable players without litigation, especially if such players threaten MS' official products? Let's put this in a wider context: MS is now suing a company called Lindows simply because it rhymes with Windows.

Does it matter if you're in the right, when you don't have money to pay the lawyers?

As a business, you think in terms of risk and reward. What are the risks that I'm taking, and what rewards do I get for taking those risks? The risk is pretty obvious: MS sitting on my head for the next 5-10 years as they keep appealing through the court system (assuming I win, of course), the entire movie industry treating me as a pariah for placing a pirated (and having no DRM) codec into my products. So what's the reward? A few thousand hardcore video traders who like to watch their Divx movies. Not exactly a compelling reason.

Lastly, Teegedeck posted a good point that I'll reply to:

>That was interesting reading, Bugs, but I don't agree with the underlying "why bother at all" attitude. Ever wondered why we are still hanging around this forum though DVD-burners will come someday? Well, I guess there's 2 possible answers:
1: I get the impression that most of us (with the exception of traders), but especially the coders, don't have the money to get the latest hardware.
2: It's our hobby...

I agree that when it's a hobby, then it doesn't matter about all the whyfors and whatnots. You just do it because you think it's cool and you enjoy it. That's what a hobby is for, and as techs, it's fun to tinker with things. But it's a different thing from "this is cool, and I like doing it" to "it'll be in wide use as long as it's better". No, I haven't tried Xvid; when it's stable, I'm sure I'll check it out some time. But even if it's head and shoulders better than Divx 3.x/Nandub, chances are remote I'll adopt it. If my reckoning is correct for DVD recordables mainstreaming this year, then I don't see any reason to switch to yet another codec for only a few months, before moving on to MPEG2. The scene may switch to it, and if I'll probably have the codec installed if I want to download any movie encoded with it, but my needs aren't the same as the video trader; I don't have to move my videos through the net. MPEG2 is a better solution hands down for blokes like me.

BTW, by virtue of calling this your hobby, then by definition you are likely to have the latest in hardware, as that's what people spend most of their money on--their hobby.

[-h]

Tech guys do things just because it's possible to do it.

The other vital factor is that it's fun - there's a reason techies hate writing documentation so much ;)

Business guys do things only if it makes them money. Formats get implemented into players not because they're available or good, but only if the companies can sell more boxes--that's to say, if there's demand.

I dare say there is more demand for MPEG4 playback than Ogg Vorbis, but OV is already getting implementations reaching availability. While there was a sizable delay between MP3 usage and eventual hardware support, businesses are now more likely to look at emerging formats so they can get onto the bandwagon earlier. I honestly believe that this is the case with MPEG4.

Re: Quality -- Doom9 has commented, and I'm sure we'd all agree, that the Divx codec is superior to Real & WMV in visual quality. However, that only comprises a small aspect of the overall picture.

You mentioned implementation and support separately to quality, so that's what I was responding to. There is a large amount of work to be done on the server side, but then that's what companies such as Philips, Psytel and others are focussing on. ffmpeg is coming along nicely too.

It's nice to hear Real announcing support for standard MPEG4, but given their entire of products is based on their closed codecs, and they've been doing pretty well up to now, it's hard to imagine they would relinquish their edge unless they are being driven to that point by MS' domination.

It is an interesting turnaround though - Real was bleating not long ago that the entire MPEG4 video spec should be trashed because their own codecs were so much better. To make an announcement like this, means they must feel that MPEG4 has something going for it.

Re: Legality of Divx players -- I'm not a lawyer, and from what I can see, none of us here are practicing IP rights lawyers either, so getting into the finer points of legality would be talking out of our ass, so I won't.

I'm just going by similar situations - in this case, WINE is identical to the reverse engineered MP43 codecs. Both were reverse engineered, lack any microsoft code, avoid infringing on any trademarks (not the case with Lindows - that was just asking for trouble) and unencumbered by patents (the last resort).

There would, however, be cease and desist letters anyway, I agree. I doubt there'll be any MP43 hardware players on the market before MPEG4 is firmly entrenched in hardware, by which time no one would want MSMPEG anyway.

I agree that when it's a hobby, then it doesn't matter about all the whyfors and whatnots. You just do it because you think it's cool and you enjoy it. That's what a hobby is for, and as techs, it's fun to tinker with things.

This is true - but there are a number of profitable industries growing up around 'hobby' professions, like overclocking or home DJs or practically the whole firewire market. There is a growing (and in my opinion encouraging) move to give people with money to burn on strange hobbies (i.e. people like me) what they want in a meatspace device.

-h

[Bugs Bunny]

Hah here I was wondering where the heck all dis MPEG4 beeswax wuz at, then Doom9 bloke went and posted a bunch of MPEG4 player links. Don't ya hate it when people do that to ya...

>While there was a sizable delay between MP3 usage and eventual hardware support, businesses are now more likely to look at emerging formats so they can get onto the bandwagon earlier. I honestly believe that this is the case with MPEG4.

There's an important concept in the MP3/Divx parallel which is worth mentioning. Succinctly, then: Content -> demand. People will go to where the content is, thereby generating demand.

Why does the average bloke install a codec? He installs it because some song or movie is available in that format, and he wants to play it. It has little to do with the codec's technical merits itself. Again, content begats demand, not the codec. He who controls the content, controls the demand. And the one in control of the most popular video content--movies--is the movie industry.

So what was different with MP3? Why did it take off w/o the industry's consent? Because the content, in MP3's case, was user generated. More accurately, it was user converted (from CDs), but the conversion process was so easy and quick, that a large amount of content was built up, and Napster came along and took it over the critical mass threshold. Industry support wasn't needed, because the content was already generated by the users. Distribution was almost as easy, as the MP3 files are small enough for P2P sharing through dial-ups.

[Here's food for thought: Think of what would've happened if the RIAA went after Napster from the outset, before demand hits critical mass, rather than sitting around for as long as they did.]

User-generated & distributed content won't happen with video. It is difficult for users to generate (convert) movies into alternate formats. And whatever content generated can't be easily shared (recall that even for broadband, upload speed is usually only 128Kbits/s). Demand for video content thus can only happen with the blessing of the movie industry. There won't be any grassroots revolution as was with MP3.

So, it's not very relevant to talk of codecs being accepted without talking about the content. Whatever codecs will be popular will be whatever codecs the movie industry uses to put out its content. If I were a movie exec in charge of online distribution, the first thing on my mind would be security (read: copy protection); that would rank even above visual quality. For open-source and other "free" codecs, DRM is something to be hated, let alone to be even considered, and most certainly it won't have as high a priority as other technical aspects. So, a "good" codec for you and me isn't necessarily (or even likely) to be "good" for the movie industry blokes. Industry standard doesn't mean squat in this case, since it's not the (tech) industry who controls the content. An industry standard codec will only prevail if it can offer the movie industry better options than the closed codecs. The make-or-break factor will be DRM, no question.

Now that I've gone with the doom-and-gloom bit, probably is a good time for some cheery. The one thing about the Internet is that it democratizes the playing field, and enables "revolutions" like what we've seen with MP3. It lets normal blokes like us to have a chance at upending the natural order. It just took a single bloke to hack out Divx and improving the vid scene. And if a goofy-looking geek like Shawn Fanning can become a millionaire for a crummy file-sharing proggie, then anything is possible. Online video has yet to exist, other than for some postcard-sized movie trailers, so there are lots of unknowns. May be another bloke will figure an end-around to the movie industry and be the next millionaire, or may be MS will have its way and lord its WMV over us. You can never tell. But as a gambling man would say, you can only bet on the known, not on the unknown.

>This is true - but there are a number of profitable industries growing up around 'hobby' professions, like overclocking or home DJs or practically the whole firewire market. There is a growing (and in my opinion encouraging) move to give people with money to burn on strange hobbies (i.e. people like me) what they want in a meatspace device.

Oh yah and for sure! A tech hobby is the best hobby there is, cause ya can easily make money out of it! Actually that applies to a lot of other hobbies too.. I once met a guy at a bus stop and we started talking, and it turns out that he's a professional card player (gambler), and he's going from casino to casino to make a living. Ask how he got to do that for a career, and he said it started out as something fun, and just sort of grew into a full-time deal. He wouldn't give me any tips though, the tightwad. All he said is that you should read books about card games, and if you know them well, then you're good enough to beat most of the wankers in casinos.

And don't even ask me the number of "computer consultants" I know, who are nothing more than homegrown geeks with a business card. Infested, is what it is.

[-h]

There's an important concept in the MP3/Divx parallel which is worth mentioning. Succinctly, then: Content -> demand. People will go to where the content is, thereby generating demand.

True - I still believe the number of people actively using MPEG4 technologies is enough demand for novelty products. Which is certainly how the market will begin.

For open-source and other "free" codecs, DRM is something to be hated, let alone to be even considered, and most certainly it won't have as high a priority as other technical aspects.

The MPEG4 spec integrates DRM if I recall correctly. A compliant player supports both protected and unprotected streams. Hence the industry attraction.

Oh yah and for sure! A tech hobby is the best hobby there is, cause ya can easily make money out of it!

Well it's no secret that the first thing people try to hack into game consoles is MP3/DivX playback. Perhaps the number of workarounds the hobby community has already produced to aid playback, are a reason no one feels like selling a hardware solution

-h

[Doom9]

time to give you a preview about the future outlook I will have in my upcoming codec comparision (still encoding btw.. some codecs are really slooooow).
I think we have firmly established now that DivX3 is not conforming to the mpeg4 specs. Hence the chance of seeing a player having dedicated DivX3 support are relatively small. Throw the fact that the format is actually sort of illegal and that MS doesn't like it at all and wants to promote their proprietary wma/wmv solutions (and they do have the money and power to get this implemented into hardware). On the other side, there's a demand for standardized mpeg4 solutions. One application is certainly video on demand. To reach DVD-like quality mpeg2 requires at least 2mbit bitrate... throw in packet loss and network overhead into the mix and then look at how many people actually could receive such a stream for reasonable costs (in most european countries 2+mbit lines are out of financial reach for the large majority of people). So, many CE companies but also software companies are working on streaming solutions based on mpeg4, most of which behind closed doors. And they're pouring a considerable amount of money into these projects. The first mpeg4 capable decoding chips should be available by now or be so soon. These chips won't play AVIs as we use them but standard mpeg4 streams (maybe you checked my news.. that prog accepts a lot of input formats to create an iso compatible mpeg4 stream which can be played on such hardware). since xvid conforms to the specs it should be possible to play it in such devices. divx4 is also very close to the specs and will probably be fully compatible in a future version. philips also has a solution available already. mpeg4 is interesting everywhere where you want good quality at low bitrates.. as I said streaming video, vod, but also mobile applications.

So.. besides Microsoft I don't think any company can push their own format thru and chances of an open and stardardized format being adopted and making it to the masses seems the most likely future to me. So... once again, the chance that we can plug all our DivX CDs (most of which are still DivX3) are little, but eventually there will be a stream format as todays mpeg1 and mpeg2 stadards that can be played on both computer and standalones.

[Bugs Bunny]

I think all the "which format will rule" talk is just taking shots in the dark. It's all guesses. Some have talked about MPEG4's technical merits and the wide support, but it won't be on technical merits that codecs get chosen, but more on business politics. For instance, AOL chose Real to be its default player not because it's better than MS' WMV, but because MS is a competitor, so "any enemy of my enemy is my friend". And also, if the precursor to the "officially blessed" online audio scene is any indication, there won't be a single dominant format, but multiple Balkanized formats that are incompatible with each other (for a while at least).

-h brought up an example which is illustrative of the foibles of in-the-box thinking. In this thread, people have voiced their concern about standalone players being able to play Divx. The in-the-box thinking is that we need to have a standalone to play Divx to the TV, and it's because we're conditioned to think that the PC is a separate thing, and the TV & home stereo is a separate thing. Yet, if all the talk from Apple and MS and Sony are to be believed, then the concept of the PC as a digital hub will be realized before too long. Then, if the PC is to be the hub, why do we need a standalone to play movies to the TV? -h mentioned that people are already trying to hack the X-box to play Divx, and I can't imagine why anybody would bother (unless they already bought an X-box). For those following the grapevine, MS reportedly will introduce a "home digital hub" this fall that are based on the X-box architecture.

Some of us already have a cheap computer hooked up to the TV set to play movies. This sounds impractical before, as a computer used to cost a lot of money (relative to a standalone player), but it's not true any more. I'm sure all of us have heard about Wi-Fi (802.11b). Right now it's not fast enough to stream DVD movies, but with Wi-Fi5 (802.11a), with a sustained throughput of 22+Mbit/s, sending DVD streams won't be a problem. Picture a central PC with receivers throughout the house, hooked up to your regular hi-fi or TV sets.

So, the idea of a standalone movie player isn't a big deal at all. I can already play Divx on my TV (using the S port on my laptop). For anybody who wants to have a cheap yet compact PC used for just this purpose, there is the Shuttle SV24 that was reviewed on Anandtech and THG, that has S video & Firewire. The whole thing is a bit larger than a toaster, and the barebone (case & motherboard) is $250. If you have some spare parts lying around, you can easily have a functional "set top PC" for around $400. And it's available today, without a need for any hack.

[Bugs Bunny]

BTW, here's a good comparison between MPEG4, Real, and WMV. Included is also a history of Divx.

http://www.extremetech.com/print_article/0,3428,a%253D3780,00.asp

[ChristianHJW]

Hi to all ... sorry for joining that late, but i wasn't invited ;) ( truth is : i'm not reading the 'General' section as often as i should - in fact i came here to report about the car vcd player ;) ) ..

My comments :

1. Nandub was designed around the DivX3 codec. There is no sense in porting it to other codecs. Also nando is not around anymore, so why call it nandub any longer ? Make a new tool based on Vdub/Nandub code, specifically for XviD or others. DivX4 is locked here ( BlackSun pointed this out many times, and he should know ), so there is no need looking at this. Simply use Vdub ( GKnot ) like many do.

2. Hardware support for DivX 4 is sure to come. This is one of DivX N's major bussiness basics and mainly the reason why they give away the codec for free personal use .... they want to have obvious arguments when talking to hardware companies. The number of downloads they have when launching a new version is a very good argument indeed ( must be 500,000 + each time ; Dan could give more details here i guess ; more news to come soon ) .....

3. About the use of MPEG 4 : DVD-R is standing outside the door .... so is H-DTV, or Hi-Res TV in general. Loved to see 1200 * xxx res DivX movies, bitrates > 3000 kbps, they will fit perfectly on one DVD-R with 2 or 3 AC3 sound streams, thanks to TMF/MCS. Even the good old PAL anamorphic DVDs ( 720 * 576 ) should be good for 848 * xxx Hi-Res DivX movies ..... dont laugh about it ..... use your brains first !!

4. The future of DivX/Video Encoding is GPL ( maybe with the exception of DivX4 ... they need us ) !! Anybody cares that Midas is not showing up anymore ? He mentioned a few times that Dolby was making him hard time about Azid and AC3 ..... i just hope i'm wrong and he is allright but only busy with other things ....

5. TMF/MCS, new A/V format : Tronic and ingo ( hopefully with the help of other coders ) will make this possible. This board here will be even more vital than now because it will offer so many new possibilities ( just browse through the threads ) that coders from anywhere will have a lot of new attractions raising their interest ...... right now, and i agree here with a comment from somebody else, there is not much to do. Be honest, DSPguru's Besweet is the most exciting new soft to come out for quite some time .... or did i miss something ?

Regards Christian

[Teegedeck]

Hi Christian,

nice of you to drop by. Can you tell me a bit more about the TMF format? From what I understand, it allows for much more features in terms of videostreams and audiostreams, menues etc than old AVI, but I simply was too lazy to read much about it... ;) Can you perhaps explain a little?

I hope and think that GPL software will rule in video-encoding, so can TMF be the coming 'container' for XviD-encoded MPEG-4-streams?

[int 21h]

Originally posted by ChristianHJW

1. Nandub was designed around the DivX3 codec. There is no sense in porting it to other codecs. Also nando is not around anymore, so why call it nandub any longer ? Make a new tool based on Vdub/Nandub code, specifically for XviD or others. DivX4 is locked here ( BlackSun pointed this out many times, and he should know ), so there is no need looking at this. Simply use Vdub ( GKnot ) like many do.


He's around, you just have to know where to look.


2. Hardware support for DivX 4 is sure to come. This is one of DivX N's major bussiness basics and mainly the reason why they give away the codec for free personal use .... they want to have obvious arguments when talking to hardware companies. The number of downloads they have when launching a new version is a very good argument indeed ( must be 500,000 + each time ; Dan could give more details here i guess ; more news to come soon ) .....


Ironic that Microsoft beat them to the punch isn't it? WMV already has hardware support, not too mention a secure content protection container. Most of this is a moot point, most users in the US are just now getting to regular DVD players (most with VCD/SVCD/XVCD/DVD-R capabilities). Average turn around time on consumer level hardware of this nature is 5 years. 5 years before Joe Schmoe makes the next upgrade. You know any companies besides SonicBlue that cater to Doom9Geek instead of Joe Schmoe? I didn't think so.


3. About the use of MPEG 4 : DVD-R is standing outside the door .... so is H-DTV, or Hi-Res TV in general. Loved to see 1200 * xxx res DivX movies, bitrates > 3000 kbps, they will fit perfectly on one DVD-R with 2 or 3 AC3 sound streams, thanks to TMF/MCS. Even the good old PAL anamorphic DVDs ( 720 * 576 ) should be good for 848 * xxx Hi-Res DivX movies ..... dont laugh about it ..... use your brains first !!


Why would you do this? Mpeg4 compression in general is of a lower quality than Mpeg2 simply because of the compression and DCT profiles used. Not too mention that HD DVD is already defined in the DVD specs with the High Profile of Mpeg2 encoding. (i.e. 1200 * xxx + bitrate up to 15mbps) If you think Mpeg4, in a frame based format, will somehow become the standard for HDTV and HD DVD, which obviously needs a field based format, you're mistaken.


4. The future of DivX/Video Encoding is GPL ( maybe with the exception of DivX4 ... they need us ) !! Anybody cares that Midas is not showing up anymore ? He mentioned a few times that Dolby was making him hard time about Azid and AC3 ..... i just hope i'm wrong and he is allright but only busy with other things ....


Perhaps. More and more software is becoming non-GPL. BeSweet and DVD2SVCD to only name a few.


5. TMF/MCS, new A/V format : Tronic and ingo ( hopefully with the help of other coders ) will make this possible. This board here will be even more vital than now because it will offer so many new possibilities ( just browse through the threads ) that coders from anywhere will have a lot of new attractions raising their interest ...... right now, and i agree here with a comment from somebody else, there is not much to do. Be honest, DSPguru's Besweet is the most exciting new soft to come out for quite some time .... or did i miss something ?

Regards Christian

For a new A/V format, obviously there is a LOT of work that would need to be done. A Directshow interface has to be written so that WMP and other Directshow players can play it back, decisions about field based or frame based have to be made, etc. Not only that, no one has discussed one of the biggest issues (besides VBR audio streams) with a new format, and that would be, FIELD based or FRAME based? If you simply continue with another framebased format you'll end up with nothing except the same issues with interlacing, etc. And you'll limit what you can do on the software playback side. Frankly, as much as I'd like to see something like this take off, it doesn't have much chance. You said it yourself, DVD-R (DVD-5 right now, who knows what will be next?) is right around the corner, perhaps by Q3 it will be at a very reasonable price. So why would anyone not use the VOB file format with Mpeg2? Considering a good majority of them already have dvd players at home (see above point) and that those dvd players will playback that video. Not too mention, Joe Schmoe will be able to rent DVDs, strip out the video stream and audio stream(s) he wants and make his own DVD without even transcoding.

All of these niche ideas for Mpeg-4 playback via TV are neat and all, but you all need to look towards industry. Industry has very little interest in this. Industry wants to use Mpeg-4 in the online streaming and the telecommunications (cell phones) markets. Obviously because those are the markets Mpeg-4 specs were written for.

You need to think in the mindframe of an enterprising business trying to make money. GPL code can be great, but for companies like Dolby to make money, obviously they need to protect their intellectual properties and when entitites are no longer licensing decoding writes from Dolby, Dolby is no longer making money. The GPL wasn't invented to protect programmers from obeying patents and licenses. And its a shame that many people seem to associate the GPL with these types of activities. You wonder why there isn't much industry support for this, its because of the rogue operations of it all. Mpeg-2 decoding without the proper licensing, Dolby AC3 decoding without the proper licensing, MP3 encoding without the proper licensing, the list goes on and on. You can't spit in the face of the industry in one action and then shine their shoes in the next, hoping for them to lift the banner of personal video enthusiasts everywhere.

You play by their games, or not at all. (I'd prefer not at all, industry doesn't have anything we need)

The only possible use of Mpeg-4 I could see on the horizion out of the above mentioned fields, is PVR (personal video recorders), but even then, Mpeg-4 isn't field based, with no support for interlaced video, so even its uses their are limited (that is the main reason Mpeg-2 is used in PVRs now (Tivo, Sonicblue devices, etc).

[Antti]

MPEG-4 does support interlacing. Not the simple profile, though.

[-h]

Why would you do this? Mpeg4 compression in general is of a lower quality than Mpeg2 simply because of the compression and DCT profiles used. Not too mention that HD DVD is already defined in the DVD specs with the High Profile of Mpeg2 encoding. (i.e. 1200 * xxx + bitrate up to 15mbps) If you think Mpeg4, in a frame based format, will somehow become the standard for HDTV and HD DVD, which obviously needs a field based format, you're mistaken.

No, a certain profile of MPEG4 video has the limitations you've mentioned (and even in that case, it's just an implementation issue). The spec caters for fields. It's evolutionary, after all.

I don't understand the DCT point - the DCT in MPEG4 is of equal or higher precision than MPEG2, and the quantisation matrix can be set to the same as MPEG2 if you feel like it. Anything MPEG2 can do, MPEG4 can do with less space. This will only be magnified when H.26L is finalised - the spec would be a failure if it wasn't the case.

-h

[trbarry]

As long as you can do a decent real time deinterlace/IVTC I'm not sure why you would care how well Mpeg4 does interlace. I thought things would compress better with progressive material.

- Tom

[b0b0b0b]

I think the problem is that some sources come interlaced. Where the top field of a frame and the bottom field of the same frame are temporally different. In this case, if you try to construct a progressive frame, you've lost information.

What I wonder is how many bits are wasted on single pixel deltas due to interlacing in a field based stream.

[int 21h]

Originally posted by -h
No, a certain profile of MPEG4 video has the limitations you've mentioned (and even in that case, it's just an implementation issue). The spec caters for fields. It's evolutionary, after all.

I don't understand the DCT point - the DCT in MPEG4 is of equal or higher precision than MPEG2, and the quantisation matrix can be set to the same as MPEG2 if you feel like it. Anything MPEG2 can do, MPEG4 can do with less space. This will only be magnified when H.26L is finalised - the spec would be a failure if it wasn't the case.
-h [/B]

Perhaps, in every representation I have looked at (of course I haven't directly read the ISO standards, I doubt any of us has), it directly correlates Mpeg-4 with content based encoding and management for online practicalities.

For instance: http://mpeg.telecomitalialab.com/standards/mpeg-4/mpeg-4.htm

[ChristianHJW]

Originally posted by int 21h about nando : he's around, you just have to know where to look ..

Is he still active in video encoding ? The two coders of TMF/MCS might ask him some short questions if he dont mind .... pls. mail or PM me where to find him ...

[NeVeRLiFt]

Thanks! I really enjoyed reading this :D

I think Mpeg4 has a very good future ahead of it.
You have several Mpeg4 codec's to choose from...
DivX5.02 Pro
XviD
And maybe even a hack so DivX3.11a works?
There's also Corona aka WMV9(though I think this will not be Mpeg4 ISO compliant, while DivX5 and XviD are)
The Movie Industry, Cable TV, DVD Unit and Decoder card makers and Cell-Phones and much more are all behind Mpeg4. RealVideo9 and Corona will just have to do there thing. I dont see them messing up Mpeg4 or being able to compete with it.


Just my .02


Thank you for this very informative post. I think alot people fail to see the business side of Mpeg4 and just keep looking at it for a way to rip DVD's or a way to encode there VideoCaptures.... thats why you see so many people asking about Nandub or Fairuse, they love the quality and want to see the same with XviD or DivX5.







:D