What needs to be done - this is an asynchronous meeting by mailing list.
Tom Van Braeckel
First of all, my apologies for not showing up on the IRC or coding on the project for such a long time. I've been looking for a job the past few weeks.
Secondly, I just wanted to hear everyone's thoughts on what still needs to be done.
As far as I know, the decoder is fully imported into FFmpeg upstream.
The encoder is pretty usable but still has that blocky bug - can we submit this into FFmpeg or should we fix that first ?
Does the code need cleaning up - hacks that need to be figured out ?
Voroshil: I saw you discussing some hacks on the FFmpeg mailing list - are they resolved ?
Voroshil: Do you still have issue #8 ? I can't reproduce it on my player.
Dobrica: Do you think we should close issue #2 (because the quality looks good now) and create a new issue about the "blocks" being visible ? What can you say about this issue and how to fix it ?
Alex: Have you had some time to work on that GUI ? I still like the idea of presenting a nice little package with an installer and a GUI for dummies.
Reynaldo: Maybe you could help us find the cause/solution to that final "visible blocks" problem ? I joined you to the Google Group so you'd get this message but I don't know - did that work ?
Leleobhz: We haven't met, have we ? I'm interested in hearing about you. What brings you here ?
All in all guys, I think the project has made great progress and has gotten to that point where I rather use our code on linux (even with the bug) than boot into windows and use the AMV Conver Tool (which also has bugs). I'd like to thank everyone for the effort they put into our project - I love you all.
Greets,
Tom.
Vladimir Voroshilov
> Hi guys,
>
> First of all, my apologies for not showing up on the IRC or coding on the
> project for such a long time. I've been looking for a job the past few
> weeks.
>
> Secondly, I just wanted to hear everyone's thoughts on what still needs to
> be done.
Pls syncronize both ffmpeg dirs (trunk and branch) with current FFmpeg SVN.
> As far as I know, the decoder is fully imported into FFmpeg upstream.
> The encoder is pretty usable but still has that blocky bug - can we submit
> this into FFmpeg or should we fix that first ?
I think those blocks are happenned due to single quantization matrix
used for both
luma and chroma planes.
Those matrix is fixed and differs from used matrixes in AMV.
I'm waiting for anybody fixing those ugly blocks to start preparing
patch for FFmpeg.
> Does the code need cleaning up - hacks that need to be figured out ?
I'll do all required cleanup myself during commit prepare procedure
taking in mind developers suggestions/patch policy.
> Voroshil: I saw you discussing some hacks on the FFmpeg mailing list - are
> they resolved ?
Only one thing i can remember is unusable ADPCM encoder without AMV muxer,
because i still not clearly undestand
time_base/pts/samplerate/framesize behaviour
(in other words how to proper initialize that values for encoder and decoder).
All issues related to decoder/demuxer are resolved, IMHO, since patch
is accepted and applied to upstream.
> Voroshil: Do you still have issue #8 ? I can't reproduce it on my player.
I can't check right now: havn't player near me.
> All in all guys, I think the project has made great progress and has gotten
> to that point where I rather use our code on linux (even with the bug) than
> boot into windows and use the AMV Conver Tool (which also has bugs). I'd
> like to thank everyone for the effort they put into our project - I love you
> all.
Thanks.
--
Regards,
Vladimir Voroshilov mailto:voro...@gmail.com
JID: voro...@gmail.com, voro...@jabber.ru
ICQ: 95587719
Alexandru Trutia
Thanks.
Tom Van Braeckel
Hi! I can make the GUI pretty fast :). All i need to get it done is a binary distribution of the code. If anyone could compile it for me, please send it in an e-mail and i'll start working. Also, I am not too much into ffmpeg's command line switches. I think I need one of you to help me with the AMV conversion part so I know how to implement the calls.
Thanks.
Great - I'll put a linux i386 binary on our Google Code page so you can get started.
We can then put in the bugfixed binary when that final bug is fixed.
- Who's going to put in the Win32 binary ?
Alex, could you try to keep this gui as general as possible ?
- There aren't any easy to find/install/use linux gui's for FFmpeg so I suggest you create a general ffmpeg gui (convert any video/audio format into any other video/audio format) that can then later easily be "restricted" so it becomes an AMV converter (any video into amv). I believe this approach will might benefit us in the future.
- Also, as ffmpeg has many many options and switches, we could decide to give the gui a "basic" and an "advanced" mode or something like that, so I suggest you make it very easy to add/remove options/switches and group them together in named profiles (such as "basic", "advanced", "purely AMV converter",..)
I think for the gui it's important that there's an auto installation package - (for linux I suggest .deb/.rpm/.sh files so Ubuntu users can just click on a download link in firefox, choose "Install package with dpkg" and have a fully working ffmpeg gui (/ AMV converter) installed)
What do you think about this Alex ?
Tom Van Braeckel
2007/10/27, Tom Van Braeckel <tomvanb...@gmail.com>:
> Hi guys,
>
> First of all, my apologies for not showing up on the IRC or coding on the
> project for such a long time. I've been looking for a job the past few
> weeks.
>
> Secondly, I just wanted to hear everyone's thoughts on what still needs to
> be done.
Pls syncronize both ffmpeg dirs (trunk and branch) with current FFmpeg SVN.
Done.
> As far as I know, the decoder is fully imported into FFmpeg upstream.
> The encoder is pretty usable but still has that blocky bug - can we submit
> this into FFmpeg or should we fix that first ?
I think those blocks are happenned due to single quantization matrix
used for both
luma and chroma planes.
Those matrix is fixed and differs from used matrixes in AMV.
I'm waiting for anybody fixing those ugly blocks to start preparing
patch for FFmpeg.
Ok - I'll be discussing it with Dobrica.
> Does the code need cleaning up - hacks that need to be figured out ?
I'll do all required cleanup myself during commit prepare procedure
taking in mind developers suggestions/patch policy.
Ok, thx.
> Voroshil: I saw you discussing some hacks on the FFmpeg mailing list - are
> they resolved ?
Only one thing i can remember is unusable ADPCM encoder without AMV muxer,
because i still not clearly undestand
time_base/pts/samplerate/framesize behaviour
(in other words how to proper initialize that values for encoder and decoder).
I don't understand this very well eighter.
All issues related to decoder/demuxer are resolved, IMHO, since patch
is accepted and applied to upstream.
Nice, great !
> Voroshil: Do you still have issue #8 ? I can't reproduce it on my player.
I can't check right now: havn't player near me.
It happened to me yesterday - but that was because I killed ffmpeg when the video was not full converted yet.
Dobrica Pavlinusic
> 2007/10/27, Vladimir Voroshilov <voro...@gmail.com>:
> > As far as I know, the decoder is fully imported into FFmpeg upstream.
> > > The encoder is pretty usable but still has that blocky bug - can we submit
> > > this into FFmpeg or should we fix that first ?
> >
> > I think those blocks are happenned due to single quantization matrix
> > used for both luma and chroma planes. Those matrix is fixed and differs
> > from used matrixes in AMV.
I think that AMV uses fixed matrices as opposed to ffmpeg which
generate them during encoding.
> > I'm waiting for anybody fixing those ugly blocks to start preparing
> > patch for FFmpeg.
>
> Ok - I'll be discussing it with Dobrica.
Sorry for late reply, but I'm afraid that I won't be able to fix it.
I tried last time around, but I just don't understand encoding code
enough to modify it to use fixed matrixes.
Does anybody has an idea who to ask? upstream FFmpeg perhaps?
I found some jpeg encoding papers which describe encoding using fixed
matrices, so it's quite possible to do it. However, I didn't really wanted
to implement whole new jpeg encoder, but since I don't understand how to
force existing one to have fixed matrixes this might be the only way.
--
Dobrica Pavlinusic 2share!2flame dpa...@rot13.org
Unix addict. Internet consultant. http://www.rot13.org/~dpavlin
Tom Van Braeckel
On Sun, Oct 28, 2007 at 03:45:09PM +0100, Tom Van Braeckel wrote:
> 2007/10/27, Vladimir Voroshilov <voro...@gmail.com>:
> > As far as I know, the decoder is fully imported into FFmpeg upstream.
> > > The encoder is pretty usable but still has that blocky bug - can we submit
> > > this into FFmpeg or should we fix that first ?
> >
> > I think those blocks are happenned due to single quantization matrix
> > used for both luma and chroma planes. Those matrix is fixed and differs
> > from used matrixes in AMV.
I think that AMV uses fixed matrices as opposed to ffmpeg which
generate them during encoding.
> > I'm waiting for anybody fixing those ugly blocks to start preparing
> > patch for FFmpeg.
>
> Ok - I'll be discussing it with Dobrica.
Sorry for late reply, but I'm afraid that I won't be able to fix it.
I tried last time around, but I just don't understand encoding code
enough to modify it to use fixed matrixes.
Does anybody has an idea who to ask? upstream FFmpeg perhaps?
Ok, I'll ask the ffmpeg-devel mailinglist upstream.
I found some jpeg encoding papers which describe encoding using fixed
matrices, so it's quite possible to do it. However, I didn't really wanted
to implement whole new jpeg encoder, but since I don't understand how to
force existing one to have fixed matrixes this might be the only way.
Could you pass through these papers ? They may come in handy.
Thx,
Tom.
Vladimir Voroshilov
>
> On Sun, Oct 28, 2007 at 03:45:09PM +0100, Tom Van Braeckel wrote:
> > 2007/10/27, Vladimir Voroshilov <voro...@gmail.com>:
> > > As far as I know, the decoder is fully imported into FFmpeg upstream.
> > > > The encoder is pretty usable but still has that blocky bug - can we submit
> > > > this into FFmpeg or should we fix that first ?
> > >
> > > I think those blocks are happenned due to single quantization matrix
> > > used for both luma and chroma planes. Those matrix is fixed and differs
> > > from used matrixes in AMV.
>
> I think that AMV uses fixed matrices as opposed to ffmpeg which
> generate them during encoding.
>
ffmpeg uses fixed table too, but single matrix for both luma and chroma planes.
And thus matrix differs from AMV's (there are two separate matrixes)
and results in
blocky picture.
We should either ask ffmpeg devs for adding two tables support or do
it ourselves.
Vladimir Voroshilov
>
>
> 2007/10/27, Vladimir Voroshilov <voro...@gmail.com>:
> >
> > 2007/10/27, Tom Van Braeckel <tomvanb...@gmail.com>:
> > > Hi guys,
> > >
> > > First of all, my apologies for not showing up on the IRC or coding on
> the
> > > project for such a long time. I've been looking for a job the past few
> > > weeks.
> > >
> > > Secondly, I just wanted to hear everyone's thoughts on what still needs
> to
> > > be done.
> >
> > Pls syncronize both ffmpeg dirs (trunk and branch) with current FFmpeg
> SVN.
>
> Done.
It is not.
Several files are missing (did you forget svn add for new files ?)
Building is not possible right now.
threethirty
Gutsy 32bit, I don't really know that much about using FFmpeg via
commandline, and I have a cheap Chinese MP3 player that uses amv (1GB
Innovage Shift3). Or if there is anything else I could do just let me
know. I've been looking for a reason to learn dpackage so I may try
to package this for ubuntu (if I can find some time)
Just let me know,
threethirty
three [at] threethirty [dot] us
threethirty.us
Alexandru Trutia
The GUI is ready for both Windows and Linux platforms. I am testing it these days and it will be online in a few days (3-5) as I am buying a domain to host my programs. So far it works great. I'll let you all know where to get it soon.
Regards,
Alex.
Tom Van Braeckel
I can't wait to check it out.
Alexandru Trutia
The program is ready to be downloaded. All flavors :).
Visit my page here: www.bytessence.com.
You can find it in the products page.
Regards,
Alex.