FFmpeg vs FFvfw vs Mencoder ? - Page 19

digitall.doc · #**361** 03-26-2004, 10:25 AM

yep, we missed you these days.

But as you can see, little changes. A new encoder (QuEnc) that makes alse use of ffmpeg engine (but I think you already know it, since you said it encodes constant bitrate), but I tried it and with the same final size, gave me worst result...
So here I am with mencoder.
Don't know if you missed a post where I commented my test. But I wondered there if in mencoder news/forums they showed intention to make mencoder accept avs files,... well, and some other doubts.
I also have been too busy these days, but I think this weekend I'll be back to tests

bilu · #**362** 03-26-2004, 01:24 PM

@digitall.doc

http://www.kvcd.net/forum/viewtopic.php?p=69883#69883

The ratecontrol code is being rewritten. Michael Niedermayer is active on both the Mencoder and FFMpeg mailing lists, so libavcodec improvements will get both encoders at the same time

Bilu

digitall.doc · #**363** 03-26-2004, 05:35 PM

bilu,
I followed your link, and read there a little...
Nic say that from the parameters we're using with Mencoder, we don't really understand the parameters...
Is he refering to our parameters?, do you know what he is refering to?. In what way he thinks it can be improved?. Since in my tests, I got better (to me) results.
Have he posted the parameters he uses in QuEnc, apart from that we can adjust in the GUI?.
I think we can learn a lot from each other

.
Don't you think so?.

bilu · #**364** 03-26-2004, 06:44 PM

I think no one should take any sort of conclusions until Michael fixes the ratecontrol code.

So far the rate control stuff looks bad to everybody

Bilu

digitall.doc · #**365** 03-27-2004, 11:19 AM

... smoooooth answer

You're prudent, man

But,... are you thinking in change your last settings?

bilu · #**366** 03-27-2004, 01:51 PM

The only good settings to test right now are those who aren't quantizer-based, i.e. those that make a difference when testing with vqscale=2. Everything else is subject to change.

Even Nic would agree with this one

Bilu

digitall.doc · #**367** 03-27-2004, 02:26 PM

I see....

Freezing tests right now

. Since what now is giving me good results, maybe won't be so good in future mencoder (ffmpeg?) versions, with changed rate control.

And don't want to test other settings, that can be useless in near future.
Even don't see the point to keep testing QuEnc, as it also will have to change...

waiting news...

bilu · #**368** 03-28-2004, 05:04 AM

Screenshots:
http://clientes.netvisao.pt/bilu/bruno/bf_bug.zip

I've posted this in the mailing list now.
It's the B-frame bug I've been telling you guys, it happens whenever you switch between interlaced and progressive or vice-versa, specially in high-action.

Mailing list thread view:
http://sourceforge.net/mailarchive/m...msg_id=7642853

Bilu

digitall.doc · #**369** 03-28-2004, 05:14 AM

bilu,
already downloaded. But I think that just happens on interlaced material, isn't it?.
I hope they'll fix it, since I will try to backup some home made films soon, and as it's DV source I guess it'll be interlaced.

What about rate control, new settings and so on... I know, I know, just 15 hours from my last post. But I have my PC warming up for tests

Is there any new in the mailing lists (advances, problems met,...)

bilu · #**370** 03-28-2004, 05:27 AM

Better relax dude

You can check things happening in libavcodec from here:
http://www1.mplayerhq.hu/cgi-bin/cvs...eg&sortby=date

You'll see Michael has been busy lately

And most of the tests I could do with a fixed quantizer were already here:
http://www.kvcd.net/forum/viewtopic.php?p=65838#65838

Main difference being that I abandoned the denoiser here:
http://www.kvcd.net/forum/viewtopic.php?p=69038#69038

Bilu

digitall.doc · #**371** 03-28-2004, 08:26 AM

Quote:

Originally Posted by bilu

Better relax dude

bilu, believe me if tell you that I'm probably one of the most quiet people you know. I'm almost ever relaxed. It's better for heart health.

In a thread like this, with about 300 posts, I just wanted to create a curiosity/expectancy/whatevercy ambient. Just that.

I thank you for the link, since I lost link to mailing list you posted some thillions of posts earlier.

I'll wait your news with increasing anxiety

... don't make us wait too long

digitall.doc · #**372** 03-29-2004, 10:11 AM

From mailing list: http://mplayerhq.hu/pipermail/mplaye...ch/043280.html
It's a bit old (from March 1st) and maybe subject of change if ffmpeg ratecontrol is changed...

Quote:

I played with cbp. It seems to give a net gain in scenes where there is alot of motion. PSNR values dont go down, but stay the same on static scenes. It seems a win/win to enable, except it does use alot more
processing time.

Nice to be used with trellis

Quote:

mv0 can only help. It just always tries encoding with (0,0) motion vector (texture only) to see if that's better than spending bits storing a motion vector.

Already advised by bilu, if I remember well

Quote:

...Anyways cmp turns out to be a great option if you have the processing power

And they tested cmp=2 better than cmp=3

This is from mplayer-users. Maybe I will have a look at developers' to see what's hot now.

audioslave · #**373** 03-30-2004, 05:28 PM

Hi guys,
This is my first attempt on MEncoder so please go easy on me. And so far I haven't had any luck with making any encodes with this supposedly great encoder. When I ran the bat file this is what I found in the DOS window:

Quote:

MEncoder dev-CVS-040229-14:06-3.2 (C) 2000-2004 MPlayer Team

CPU: Advanced Micro Devices Athlon 4 /Athlon MP/XP Palomino 1673 MHz (Family: 6,
Stepping: 2)
Detected cache-line size is 64 bytes
CPUflags: Type: 6 MMX: 1 MMX2: 1 3DNow: 1 3DNow2: 1 SSE: 1 SSE2: 0
Compiled with Runtime CPU Detection - WARNING - this is not optimal!
To get best performance, recompile MPlayer with --disable-runtime-cpudetection.
Reading /cygdrive/c/Documents and Settings/Hea/.mplayer/codecs.conf: Can't open
'/cygdrive/c/Documents and Settings/Hea/.mplayer/codecs.conf': No such file or d
irectory
Reading mplayer//codecs.conf: Can't open 'mplayer//codecs.conf': No such file or
directory
Using built-in default codecs.conf.
File not found: 'frameno.avi'
Failed to open frameno.avi
Reading config file /cygdrive/c/Documents and Settings/Hea/.mplayer/mencoder: No
such file or directory
font: can't open file: /cygdrive/c/Documents and Settings/Hea/.mplayer/font/font
.desc
font: can't open file: /usr/local/share/mplayer/font/font.desc
success: format: 0 data: 0x0 - 0x382A00
AVI file format detected.
AVI_NI: No audio stream found -> no sound.
VIDEO: [AVIS] 480x576 16bpp 25.000 fps 0.8 kbps ( 0.1 kbyte/s)
[V] filefmt:3 fourcc:0x53495641 size:480x576 fps:25.00 ftime:=0.0400
Opening video filter: [expand osd=1]
Expand: -1 x -1, -1 ; -1 (-1=autodetect) osd: 1
================================================== ========================
Cannot find codec matching selected -vo and video format 0x53495641.
Read DOCS/HTML/en/codecs.html!
================================================== ========================
Exiting...
Tryck på en valfri tangent för att fortsätta...

What am I missing?

P.S. Never mind the last line. No there's nothing wrong with your screen - it's swedish!

audioslave · #**374** 03-30-2004, 05:58 PM

I downloaded another build and now it works!

I'm very interested in seeing the output quality of this encoder. From what I've read it's something else, eh?

audioslave · #**375** 03-30-2004, 06:09 PM

BTW Is MEncoder using the KVCD Notch Matrix

digitall.doc · #**376** 03-31-2004, 01:58 AM

Hi audioslave,
you'll enjoy this encoder.
Mencoder doesn't use KVCD matrix by default. You have to give it through inter_matrix and intra_matrix, in the ini file or in the command line.
Take a look at the last command posted by bilu, and there you'll see the matrix.
If it doesn't work for you, make me know and I'll post the complete *.ini and *.bat that have worked for me.

Happy mencodings

audioslave · #**377** 03-31-2004, 04:36 AM

@digitall.doc
Yes, please post the *.ini and *.bat

.

@incredible
How do I use the Slicer routine with MEncoder - for prediction

digitall.doc · #**378** 03-31-2004, 10:38 AM

Well, these are my/our last settings. But as bilu stated, everything is susceptible of changing, since they're going to change the rate control in ffmpeg.
mencoder.bat

Code:

set menc=D:\KVCD\Mencoder
set source=D:\Temp\1
set output=D:\Temp\encoded

%menc%\mencoder -include %menc%\mplayer\settings.ini -vf yuvcsp -lavcopts vbitrate=9800:scplx_mask=0.22:vqmin=2:mbqmin=2:vqmax=10:mbqmax=10
%source%.avi -o %output%.m2v

settings.ini

Code:

of=rawvideo=1 
ovc=lavc=1 
noskip=1 
nosound=1 
lavcopts=vcodec=mpeg2video:intra_matrix=8,9,12,22,26,27,29,34,9,10,14,26,27,
29,34,37,12,14,18,27,29,34,37,38,22,26,27,31,36,37,38,40,26,27,29,36,39,38,40,
48,27,29,34,37,38,40,48,58,29,34,37,38,40,48,58,69,34,37,38,40,48,58,69,79:
inter_matrix=16,18,20,22,24,26,28,30,18,20,22,24,26,28,30,32,20,22,24,26,28,30,
32,34,22,24,26,30,32,32,34,36,24,26,28,32,34,34,36,38,26,28,30,32,34,36,38,40,28,
30,32,34,36,38,42,42,30,32,34,36,38,40,42,44:mbd=2:vrc_minrate=300:
vrc_maxrate=9800:vrc_buf_size=1835:vmax_b_frames=2:preme=2:precmp=2:
trell=1:cbp=1:mv0=1:naq=0:p_mask=1:vqsquish=0:vqblur=0:vrc_eq=tex:keyint=15:
aspect=16/9

Some comments:
- In the directory tree, we have mencoder.exe and mencoder.bat in a folder, and the mplayer folder (where you have settings.ini) is a folder in the same mencoder.exe folder (don't know if I explained well

)
- I left vbitrate=9800 and scplx_mask=0.22 in the bat file, since they are the parameters I tweak to adjust final file size (increasing scplx_mask value or lowering vbitrate, lowers file size) I now usually change scplx_mask (not above 0.3)
- I use vrc_eq=tex, but is the same as vqcomp=1
- I also used (and have seen some people using it) cmp=2:subcmp=2 (I think is SATD instead of SAD), and predia=-2:dia=-2 (faster as man_page says)
- I also use p_mask=1 but still don't really know how it works (advised by bilu), didn't make tests. If anyone is willing to test...
- These settings: trell=1:cbp=1:mv0=1 are for quality, but slow down... If you want you may try without them and see: 1 speed gain, 2 quality loss. And then choose...

There are some other settings I saw people using and didn't test:
- vratetol=1835 A setting Amenophis uses. It refers to rate control tolerance. Default is 8000, and lower values make bitrate changes more smooth (don't know if it is of interest; what do you think bilu?)
- I saw people using vqmin=1:mbqmin=1:lmin=1, in order to get higher bitrates. It isn't advised in man_page, and I didn't test. But I already tested vqmin=3:mbqmin=3:lmin=2.5, and helped to lower quantizers... If you want to do tests.

Hope this will be of a help, but I insist, is subjected to be change now, and in the near future

EDIT: the file 1 in the Temp folder, is a fake avi done from avs with makeavis
EDIT2: this settings are for PAL progressive sources. In interlaced sources, I recommend you to try bilu settings (ilme, ildct and so)

bilu · #**379** 03-31-2004, 12:12 PM

Quote:

FFMPEG
=====
`-b bitrate'
set the video bitrate in kbit/s (default = 200 kb/s)
`-bt tolerance'
set video bitrate tolerance (in kbit/s)
`-maxrate bitrate'
set max video bitrate tolerance (in kbit/s)
`-minrate bitrate'
set min video bitrate tolerance (in kbit/s)
`-bufsize size'
set ratecontrol buffere size (in kbit)

Mencoder
======
vratetol=<value>
approximated filesize tolerance in kBit. 1000?100000 is a sane range. (warning: 1kBit = 1000 Bits) (default: 8000)

vrc_buf_size=<value>
buffer size in kbit (pass 1/2). For MPEG1/2 this also sets the vbv buffer size, use 327 for VCD, 917 for SVCD and 1835 for DVD. NOTE: vratetol should not be too large during the second pass or there might be problems if vrc_(min|max)rate is used

As usual, libavcodec documentation sucks

By crossing docs between FFMPEG and Mencoder I got to the conclusion that vratetol,much more than the vqcomp or vrc_eq settings, is the parameter that will let enforce the average bitrate behaviour.

Lowering vratetol will make the filesize much more predictable, but it will still be VBR encoding, not CBR as if vrc_minrate=vrc_maxrate.

About vratetol=vrc_buf_size, I'm not really sure it makes sense.
I think VBV specifies a maximum bitrate variation, it could be less than VBV and still be compliant. If someone knows this subject better, please provide feedback.

I'd like to see how a vratetol=0 behaves

@digitall.doc
Thanks for bringing this parameter into our attention.
I was so obsessed with the rate control fixing that I forgot to look at it

Bilu

digitall.doc · #**380** 03-31-2004, 04:23 PM

bilu,
nothing to thank. Almost all I've learnt from mencoding and parameters, and my interest with mencoder, is due to you and your advise and help. I'm the one to thank a lot

.

Don't know if the vratetol value=1835 was intended to be equal vrc_buf_size, but is a value to begin testing with (and, of course, also 0)