D Richard Felker III | 1 Feb 01:14 2004

Re: lanczos or bicubic spline or..?

On Sat, Jan 31, 2004 at 12:06:59PM -0800, * wrote:
> --- * <afe0108 <at> yahoo.com> wrote:
> > Hello, I scoured the list but couldn't find any opinion about
> > which -sws option is generally considered the best.  Taking into
> > account the quality of the implementation in mencoder if that's
> > an issue.
> > 
> > From what I've read -sws 9 (lanczos) might be the best for sharp
> > details?  Here's a page comparing them:
> > 
> > http://aroundcny.com/technofile/texts/tec042901.html
> > 
> > But I can't personally tell any difference between -sws 2 or 9
> > or 10.  Can anyone else?  Thanks.
> > 
> 
> Hello, no one has an opinion which is better?  If I want to test it
> myself could I assume that (everything else being equal) higher PSNR
> values mean it's better?  Thanks.

No. The PSNR the encoder reports is relative to the input image it
gets, so higher PSNR values just mean the scaler produced data that
compresses better. In general I would recommend -sws 1 or 6 for
encoding. Gauss (7) could also be useful if you're trying to make very
low-res/quality encodes and avoid artefacts, while Lanczos (9) could
be useful if you're using constant-quantizer encoding and want maximum
quality. But IMO, at a fixed bitrate, the scaler that blurs the most
will look best.

Rich
(Continue reading)

Joe Neeman | 1 Feb 01:49 2004
Picon

Re: nvidia/cvidix "Can't find chip" -- NOVIRUS

After changing nvidia_vid.c and recompiling, I have almost the same results 
as Corey.

>Actually, it seems that simply switching to a text console and back is
>all that is needed to make xvidix show something (garbled or otherwise).
>My running mplayer -vo cvidix was just incidental.
>
>After further testing, the results seem to be the same for me with or
>without the patch.
>
>
># mplayer -vo xv
>result: works fine      /*same for me*/
>
># mplayer -vo xvidix
>result: green screen         /*same for me*/
>
># mplayer -nocolorkey -vo xvidix
>result: black screen              /*same for me*/
>
># mplayer -vf scale=640:480 -nocolorkey -vo xvidix
>result: (bigger) black screen              /*same size black screen (but 
>the movie was 640:480 already)*/
>
><switch to text console, then back to X>
>
># mplayer -vo xvidix
>result: garbled video              /*this one works fine for me :)*/
>
># mplayer -nocolorkey -vo xvidix        result: garbled video               
(Continue reading)

Corey Hickey | 1 Feb 06:12 2004

Re: Bug report (mencoder sig-faults)

Mark Williams (MWP) wrote:
> Hi yall ;)
> 
> Ive just tried encoding a movie using todays CVS (31/01/04) of mplayer and ffmpeg (for libavcodec).
> First pass completes, but looks shocking (very boxy).

Are you cropping and scaling appropriately? From your "VOP=" below, I'm
guessing not. Read DOCS/tech/encoding-tips.txt. The basic gist is that
you need to crop out any black bars, _entirely_, and then scale the
image so both axes are multiples of 16.

> On trying to start the second pass, mencoder dies with a sig-fault.
> 
> Im using:
> 
> LAVC="vcodec=mpeg4:mbd=2:trell:mv0:cbp:naq:v4mv:vme=4:vrc_eq=tex:vmax_b_frames=1:vb_strategy=0:vstrict=-1:preme=2:dia=2:predia=2:vqcomp=0.8:lumi_mask=0.05:dark_mask=0.01"
> VOP="hqdn3d,pp=dr/h1/v1"
> 

Are you using the same lavc parameters for both passes? If so, try using
dd to extract the last 1MB or so from the video (use mplayer -dumpstream
first if you're encoding a DVD). Try encoding that 1MB chunk and see if
you get the same results. If so, read:
http://www.mplayerhq.hu/DOCS/HTML/en/bugreports.html
and submit a complete bugreport, including the upload of the 1MB chunk
to the mplayerhq ftp server.

-Corey
Mark Williams (MWP | 1 Feb 07:59 2004
Picon

Re: Bug report (mencoder sig-faults)

> Mark Williams (MWP) wrote:
> > Hi yall ;)
> > 
> > Ive just tried encoding a movie using todays CVS (31/01/04) of mplayer and ffmpeg (for libavcodec).
> > First pass completes, but looks shocking (very boxy).
> 
> Are you cropping and scaling appropriately? From your "VOP=" below, I'm
> guessing not. Read DOCS/tech/encoding-tips.txt. The basic gist is that
> you need to crop out any black bars, _entirely_, and then scale the
> image so both axes are multiples of 16.

I am cropping/scaling... the settings are just not on those two lines i gave.

> > On trying to start the second pass, mencoder dies with a sig-fault.
> > 
> > Im using:
> > 
> > LAVC="vcodec=mpeg4:mbd=2:trell:mv0:cbp:naq:v4mv:vme=4:vrc_eq=tex:vmax_b_frames=1:vb_strategy=0:vstrict=-1:preme=2:dia=2:predia=2:vqcomp=0.8:lumi_mask=0.05:dark_mask=0.01"
> > VOP="hqdn3d,pp=dr/h1/v1"
> > 
> 
> Are you using the same lavc parameters for both passes?

Of course.

> If so, try using
> dd to extract the last 1MB or so from the video (use mplayer -dumpstream
> first if you're encoding a DVD). Try encoding that 1MB chunk and see if
> you get the same results. If so, read:
> http://www.mplayerhq.hu/DOCS/HTML/en/bugreports.html
(Continue reading)

Corey Hickey | 1 Feb 10:37 2004

Re: Bug report (mencoder sig-faults)

Mark Williams (MWP) wrote:

> Ive done more testing.
> It seems "naq" is to blame.... is it still experimental?
> 

I've used naq lots of times, without problems. Either way, you should
make a complete bugreport so it can be fixed.

-Corey
Sascha Sommer | 1 Feb 11:22 2004
Picon

Re: nvidia/cvidix "Can't find chip" -- NOVIRUS

> Actually,  I also tested them all with a non-640:480 video and the results
> were the same as with the 640:480 video.
>

Are you using the nv X-Server? If not please redo your tests with it. Then
we can find the relevant pieces of
code to fix nvidia_vid.

Sascha
Mark Williams (MWP | 1 Feb 12:01 2004
Picon

Re: Bug report (mencoder sig-faults)

> Mark Williams (MWP) wrote:
> 
> > Ive done more testing.
> > It seems "naq" is to blame.... is it still experimental?
> > 
> 
> I've used naq lots of times, without problems. Either way, you should
> make a complete bugreport so it can be fixed.

Yeh, well i also used it on another 2-pass encoding using those exact same options, and it worked fine.

I will submit a bug report.

Thanks.
Sascha Sommer | 1 Feb 12:42 2004
Picon

Re: nvidia/cvidix "Can't find chip" -- NOVIRUS

> I tried xvidix, and still got a green screen. I tried cvidix, and there
> still wasn't any video, with or without -nocolorkey. Then I decided to
> make some logs in case I was doing something boneheaded. When I tried
> xvidix again, instead of a green screen, the video appeared in strange
> diagonal stripes. I wish I could take a screenshot. The stripes have the
> correct luma/chroma from the video, but it's all jumbled to the point of
> being unrecognizable.
> 

Now it should work without -vf scale 

Sascha
Dave Sorg | 1 Feb 12:47 2004
Picon

Question About installing

I'd like to install on OpenBSD, but I keep getting an error that I need the 
codecs. Which file are they in, and what directory do I install them to? 
Also, once the player is installed, how do I use it?

_________________________________________________________________
Find high-speed ‘net deals — comparison-shop your local providers here. 
https://broadband.msn.com
Matic | 1 Feb 14:07 2004
Picon

mencoder and tv


Hi all.
I have tried to use MEncoder 1.0pre3-3.2.3 to record tv from my tv card,
but i don't succeed to record audio stream.
I have tried whit a lot of combination of audio codec (copy, pcm,
mp3lame, lavc) but i always get in a range of time that goes from 3-4
seconds to 90-120 seconds the message:
not enough audio samples!
followed by a million of "error reading audio: File descriptor in bad
state" in few seconds
using "noaudio" as -tv parameter is the only way to make mencoder to
work, but the resultant output is without audio of course.

PS
The audio channel in my tv card is redirect to Line In of my soundcard
PPS
with audio enabled, mencoder works for few seconds, and the resultant
output has the audio

Has anyone my problem ? or a solution of it ?

Thanks to all
Giacomo

Gmane