FFmpeg vs FFvfw vs Mencoder ? - Page 13

kwag · #**241** 03-10-2004, 07:33 PM

Quote:

Originally Posted by bilu

Quote:

Originally Posted by kwag

Actually, it's

cd /d/mingw/mplayer/

If you look here http://www.mplayerhq.hu/DOCS/HTML/en/windows.html
you'll see it works both ways

I did it like I posted and it worked.

Bilu

Check again

I think the DOCs are wrong

It doesn't work, at least on my machine. As a matter of fact, take a look at the mount points on Windows:

Code:

C:\DOCUME~1\karl\LOCALS~1\Temp on /tmp type user (binmode,noumount)
c:\ActiveState\perl on /perl type user (binmode)
C:\msys\1.0\bin on /usr/bin type user (binmode,cygexec,noumount)
C:\msys\1.0\bin on /bin type user (binmode,cygexec,noumount)
C:\msys\1.0 on / type user (binmode,noumount)
C:\msys\1.0 on /usr type user (binmode,noumount)
c:\mingw on /mingw type user (binmode)
a: on /a type user (binmode,noumount)
c: on /c type user (binmode,noumount)
d: on /d type user (binmode,noumount)
e: on /e type user (binmode,noumount)
f: on /f type user (binmode,noumount)
g: on /g type user (binmode,noumount)
k: on /k type user (binmode,noumount)

So on Windows, you must change to /a or /c etc. not to /a:, /c: etc.

Code:

$ cd /c:
sh: cd: /c:: No such file or directory

But:

karl@HOME ~
$ cd /c

karl@HOME /c
$ 
$ pwd
/c

karl@HOME /c
$

works

-kwag

bilu · #**242** 03-11-2004, 04:59 AM

My /etc/fstab only contains:

d:/mingw/mingw /mingw

And works like this:

Quote:

Administrator@HENRIQUESB /c
$ cd d:

Administrator@HENRIQUESB /d
$ cd c:

Administrator@HENRIQUESB /c

As you can read in \msys\1.0\doc\msys\README.rtf :

Quote:

Msys file system bindings (mounts) are automatic and happens as described in table 1. These automatic file system bindings are not changable by the user. User defined file system bindings can be created by specifying them in the /etc/fstab directory as explained in table 2.

Code:

TABLE 1 - Automatic file system maps:
+ ---------------------------------------------------------------------------- +
| The automatic mounts are relative to where the msys-1.0.dll (DLL) is located |
| such that the following is true:                                             |
|                                                                              |
| / - is the parent directory of the directory containing the DLL              |
| /bin - the direcotry containing the DLL                                      |
| /usr - the parent directory of the directory containing the DLL              |
| /usr/bin - the directory containing the DLL                                  |
| /tmp - the value of the TMP environment variable                             |
| /c - C:\                                                                     |
| /d - D:\                                                                     |
| . . .                                                                        |
| /z - Z:\                                                                     |
|                                                                              |
+ ---------------------------------------------------------------------------- +

It may be related with the translations in this table, because it does work.
But you must use CD C: , not CD /C: as you posted.

EDIT: But your method is more correct and the filename completion works only with that method (just tried the TAB for a directory name completion using both /d/ and d:/ )

Bilu

bilu · #**243** 03-11-2004, 05:34 AM

Found out what was the problem with VMesquita's man page conversion to HTML: just open the man page in a text editor, replace .IPs with .IP and then do the conversion.

I updated the link with a package including an HTML man page with the correct values. Also include the utility for the man page conversion. Just do:

man2htm2 mplayer.1 > mplayer.html

Cheers,
Bilu

bilu · #**244** 03-11-2004, 05:40 AM

I'd like to have your feedback about:

- using Avisynth sources
- "PTS to SCR delay" messages when using -of rawvideo

I never got AVS sources to work even with VMesquita's build, will try again now

Bilu

digitall.doc · #**245** 03-11-2004, 06:03 AM

Hi bilu,
I'm always using avisynth (through makeavis) as a source, making use of vmesquita's method, and always worked fine. Once I even compared to just resize with -vf internal mencoder (or mplayer?) filters (crop, expand,...) and bicubicresize with avisynth, and avisunth encoded faster

.

About your compilation, encoding nice (just two small tests) and I really didn't get any PTS to SCR error, but don't know if was just by chance. I'll do further tetsts. What I already noticed is that with your compilation I get 1 fps less compared to vmesquita's (maybe due they were compiled in different ways?). But I find rawvideo really nice. And the Unknown block type errors also disapeared!.

bilu · #**246** 03-11-2004, 06:18 AM

Quote:

Originally Posted by digitall.doc

About your compilation, encoding nice (just two small tests) and I really didn't get any PTS to SCR error, but don't know if was just by chance. I'll do further tetsts. [...] But I find rawvideo really nice. And the Unknown block type errors also disapeared!.

Not by chance

but I really wanted to know if it's working.

http://article.gmane.org/gmane.comp....yer.user/27198

Quote:

From: Tobias Diedrich <ranma <at> gmx.at>
Subject: Re: Telecine output - needed for NTSC DVD encoding
Newsgroups: gmane.comp.video.mplayer.user
Date: Tue, 09 Mar 2004 20:55:41 +0100

Bruno wrote:

> Any advantage of using -of rawvideo instead of -of mpeg ?
> I usually demux to an m2v file.
> Could it be helpful to get rid of those "PTS to SCR delay" screens?

Right, you can skip the demux step with it and get rid of the "PTS to
SCR delay" warnings.

Bilu

bilu · #**247** 03-11-2004, 07:10 AM

This is fun.

http://article.gmane.org/gmane.comp....yer.user/27233

My settings right now, bold settings just for NTSC:

Quote:

BATCH
=====
mencoder -include settings.ini -vf-pre softpulldown -lavcopts keyint=15 (18 ) movie.vob -o movie.m2v

SETTINGS.INI
=========
of=rawvideo=1
vf=il=d,hqdn3d,il=i,yuvcsp
ovc=lavc=1
nosound=1
noskip=1
lavcopts=vcodec=mpeg2video:ildct=1:ilme=1:vstrict=-1:
vrc_buf_size=1835:vqcomp=0:vqblur=0

reme=2:ildctc mp=2:
precmp=2:vqmax=10:mbqmax=10:vbitrate=300:vrc_minra te=300:
vrc_maxrate=9800:scplx_mask=0.3:autoaspect=1

EDIT: you may notice that I'm now using vf=il=d,hqdn3d,il=i,yuvcsp instead of vf=hqdn3d,yuvcsp . It gives better compressibility on interlaced encodes.

Bilu

bilu · #**248** 03-11-2004, 09:44 AM

Someone willing to test this?

p_mask=<0.0-1.0>
inter MB masking (default: 0.0 (disabled))

Bilu

bilu · #**249** 03-11-2004, 10:59 AM

Notch matrices:

Quote:

Interlaced
======

intra_matrix=8,10,22,27,29,37,37,40,9,12,14,28,29, 37,39,40,9,14,27,31,34,37,40,48,12,22,27,29,34,37, 40,58,26,27,29,34,37,38,48,58,26,27,29,36,38,38,48 ,69,18,27,34,36,38,38,48,69,26,26,34,34,38,40,58,7 9

inter_matrix=16,20,22,26,28,32,32,36,18,20,22,28,2 8,32,34,36,18,22,26,30,30,32,36,38,20,22,26,28,30, 32,36,42,24,26,28,30,32,34,38,40,24,26,28,32,34,34 ,38,42,24,26,30,32,34,34,38,42,24,24,30,30,34,36,4 0,44

Progressive
=======

intra_matrix=8,9,12,22,26,27,29,34,9,10,14,26,27,2 9,34,37,12,14,18,27,29,34,37,38,22,26,27,31,36,37, 38,40,26,27,29,36,39,38,40,48,27,29,34,37,38,40,48 ,58,29,34,37,38,40,48,58,69,34,37,38,40,48,58,69,7 9

inter_matrix=16,18,20,22,24,26,28,30,18,20,22,24,2 6,28,30,32,20,22,24,26,28,30,32,34,22,24,26,30,32, 32,34,36,24,26,28,32,34,34,36,38,26,28,30,32,34,36 ,38,40,28,30,32,34,36,38,42,42,30,32,34,36,38,40,4 2,44

Will try them now

Cheers,
Bilu

bilu · #**250** 03-11-2004, 11:53 AM

As Incredible has posted here:

NOTCH Interlaced Matrix
http://www.kvcd.net/forum/viewtopic.php?t=7985

Quote:

In case of interlaced MPEG2 encodings where the scan order is based on an alternate order and NOT ZigZag, we still can enjoy the "Notch" filtering.

But as the Mencoder man page says, it's optional:

Quote:

alt
use alternative scantable

All my interlaced encodes say ZigZag in Bitrate Viewer.
I'm not using the alt switch.

Tried both the interlaced and progressive versions of the Notch matrix over an interlaced encode, and the progressive one was very good, the interlaced one was horrible

So for now on my settings.ini file will be:

Quote:

of=rawvideo=1
vf=il=d,hqdn3d,il=i,yuvcsp
ovc=lavc=1
nosound=1
noskip=1
lavcopts=vcodec=mpeg2video:ildct=1:ilme=1:vstrict=-1:vrc_buf_size=1835:vqcomp=0:vqblur=0

reme=2:ildc tcmp=2

recmp=2:vqmax=10:mbqmax=10:vbitrate=300:vr c_minrate=300:vrc_maxrate=9800:scplx_mask=0.3:auto aspect=1:intra_matrix=8,9,12,22,26,27,29,34,9,10,1 4,26,27,29,34,37,12,14,18,27,29,34,37,38,22,26,27, 31,36,37,38,40,26,27,29,36,39,38,40,48,27,29,34,37 ,38,40,48,58,29,34,37,38,40,48,58,69,34,37,38,40,4 8,58,69,79:inter_matrix=16,18,20,22,24,26,28,30,18 ,20,22,24,26,28,30,32,20,22,24,26,28,30,32,34,22,2 4,26,30,32,32,34,36,24,26,28,32,34,34,36,38,26,28, 30,32,34,36,38,40,28,30,32,34,36,38,42,42,30,32,34 ,36,38,40,42,44

Differences in filesizes (very small interlaced PAL encode, dark scene from The Abyss):

Without Notch: 6177 KB
With Notch: 4993 KB

It's very good to see a 20% improvement over such a small scene where I was already using spatial complexity masking and a denoiser

Congrats for the matrix!

Bilu

kwag · #**251** 03-11-2004, 12:35 PM

Hi bilu,

I know this post is going to bring some heavy discussion, but I have to say this:
The matrix has NOTHING to do with the scan order

Each position of the matrix, is focused on a specific frequency domain.
No matter what scan order is used, the "weighting" or frequency spectrum per position, is always the same. Different scan orders are used, depending on the source material, it it's interlaced or progressive.
So it's a misconception that a diffrerent matrix is required if the source is interlaced or progressive.
So the "Notch" matrix is usefull either on progressive or interlaced material.
This pattern:

Code:

intra_matrix=8,10,22,27,29,37,37,40,9,12,14,28,29,37,39,40,9,14,27,31,34,37,40,48,12,22,27,29,34,37,40,58,26,27,29,34,37,38,48,58,26,27,29,36,38,38,48,69,18,27,34,36,38,38,48,69,26,26,34,34,38,40,58,79 

inter_matrix=16,20,22,26,28,32,32,36,18,20,22,28,28,32,34,36,18,22,26,30,30,32,36,38,20,22,26,28,30,32,36,42,24,26,28,30,32,34,38,40,24,26,28,32,34,34,38,42,24,26,30,32,34,34,38,42,24,24,30,30,34,36,40,44

will drastically change the frequency response, and it's not even close to the actual KVCD "Notch" matrix.

References here: http://rnvs.informatik.tu-chemnitz.d...mpeg_tech.html
And in the book: "Video Demystified" specifically pages 234 and 527.

Edit: Sorry Inc., but after reading more on MPEG specifications, the statement above is very clear.

-kwag

bilu · #**252** 03-11-2004, 12:39 PM

Well, my friend, that's fine by me

You know this subject much better than me and I'm pretty happy with the results anyway

EDIT: And although I could check if you're right by trying the alt parameter, I don't really think it's worth it since I'm getting good results on both progressive and interlaced this way.

Cheers,
Bilu

kwag · #**253** 03-11-2004, 01:05 PM

Quote:

Originally Posted by bilu

Well, my friend, that's fine by me

I just wanted to expose a fact, so I should have addressed it to @ALL. It was not intended as a specific reply to you. Sorry if I posted incorrectly

Greets

-kwag

incredible · #**254** 03-11-2004, 01:10 PM

Well then I don't know why many very knowledged persons for example at doom9.de/Gleitz do recommend using aalternate scan ordered matrix in case of interlaced encodings.

mb1 explaind that once very well, but his page is down now

bilu · #**255** 03-11-2004, 01:17 PM

@incredible

I haven't tried it yet, but I've been pretty happy encoding interlaced sources without using alt.

Bilu

bilu · #**256** 03-11-2004, 01:38 PM

@incredible

http://bmrc.berkeley.edu/research/mp...v38.html#tag15

Search for alternate

I'll try it tomorrow.

EDIT: two more links from Doom9, haven't seen them yet:

http://www.physics.utu.fi/ett/kurss...eemu_tiivis.pdf
http://www.ece.cmu.edu/~ee899/lecture7.pdf

Cheers,
Bilu

kwag · #**257** 03-11-2004, 01:52 PM

Quote:

Originally Posted by incredible

Well then I don't know why many very knowledged persons for example at doom9.de/Gleitz do recommend using aalternate scan ordered matrix in case of interlaced encodings.

mb1 explaind that once very well, but his page is down now

I think they were confused with encoder issues. Some encoders give a different result, depending on the scan order.
The way the matrix is designed (as per specifications), is that every "slot" or position that is, always has a frequency domain range.
Like this (scanned from a book) :

So every position is fixed, or works on a specific frequency range, regardless of the scan order used.

-kwag

digitall.doc · #**258** 03-12-2004, 05:38 AM

Sorry if I jump in here too late, and a bit off topic now,

Quote:

Originally Posted by bilu

Differences in filesizes (very small interlaced PAL encode, dark scene from The Abyss):

Without Notch: 6177 KB
With Notch: 4993 KB

It's very good to see a 20% improvement over such a small scene where I was already using spatial complexity masking and a denoiser

Congrats for the matrix!

Bilu

Bilu, happy to see you tested it, and liked it... I won't say I advised you

What about avisynth?, I'm sure you have already worked with it, but did you tested it with mencoder?. I would only drop avisynth if after testing internal mplayer/mencoder filters they give similar results as avisynth, and faster. As I told you I just tried them once to resize, and was slower

.

BTW bilu, I did several encodes with your settings and command (without denoiser and using avisynth) and when I play it on PowerDVD it plays with little "jumps", as the image stopped from time to time... don't know why.

bilu · #**259** 03-12-2004, 05:53 AM

Quote:

Originally Posted by digitall.doc

What about avisynth?, I'm sure you have already worked with it, but did you tested it with mencoder?. I would only drop avisynth if after testing internal mplayer/mencoder filters they give similar results as avisynth, and faster. As I told you I just tried them once to resize, and was slower

.little "jumps", as the image stopped from time to time... don't know why.

I already had my time with Avisynth (at least 70% of my thousands of post at Doom9 were Avisynth-related

) but here I've got cross-platform options and can got to Linux/FreeBSD/QNX anytime

Quote:

BTW bilu, I did several encodes with your settings and command (without denoiser and using avisynth) and when I play it on PowerDVD it plays with little "jumps", as the image stopped from time to time... don't know why.

This will be a part of my next phase: learn authoring and testing on standalones

BTW, did you test with or without the Notch matrix?

Bilu

bilu · #**260** 03-12-2004, 05:56 AM

For the first time I found a stream where BBDMUX didn't work. But:

mencoder -of rawvideo -noskip -ovc copy movie.vob -o movie.m2v

worked just fine

Bilu