headers
Dan Dennedy | 1 Jun 2007 05:16
Favicon

Re: mjpeg avi files

On Wednesday 30 May 2007 12:07, Lars Täuber wrote:
> Is it possible to edit mpjeg files with kino losslessly?

No, sorry.
http://www.kinodv.org/help/intro

> The mjpeg coding style and dv-coding style don't differ as much from each
> other as from other video codecs (mpeg2) i thought. Does a (necessary)

True, but the Kino developers just never got around to adding that and have 
moved on to other projects.

> conversion from mjpeg to dv loose any information (e.g. color depth)? What

yes

> about different aspect ratios? (1280x720 defined for dv?)

no, dv is based upon NTSC and PAL and 16:9 variants thereof. There is a 
dvcpro-hd, but that is not supported. A 1280x720 will be converted to a an 
anamorphic widescreen NTSC or PAL resolution similar to your widescreen DVD 
movies (ignoring framerate differences).

> The reason I ask is, I think about buying a canon powershot tx1 that makes
> 720p mjpeg files at 30fps. Will kino handle this files somehow?

No, but you should give Cinelerra a try. Kdenlive may be able to better handle 
this within a year, but not yet. 

I am curious to know... after editing, what do you want to do with the video?
(Continue reading)

Permalink | Reply | 1 comment |
headers
Stefan M. Fendt | 1 Jun 2007 08:15
Picon
Favicon

[Fwd: Re: DV: deinterlacing at the right moment]
Picon Favicon
From: Stefan M. Fendt <stefan <at> sfendt.de>
Subject: Re: [Kino-dev] DV: deinterlacing at the right moment
Date: 2007-06-01 06:15:20 GMT
Am Donnerstag, den 31.05.2007, 23:16 +0200 schrieb Levente Novák:

Just a very short response here (I will if you like to, explain it in
more detail, later, but I am somewhat limited in time this morning...
And I hope it can be understood, like it is.)

> I am very interested how you demonstrate the matematical correctness of
> your Correct Deinterlacing.

Not "my" deinterlacing. It is already used in daily products (eg 100Hz
TV-Sets and LCD-TV-Sets). Bellers, de Haan et al, use and speak of
deinterlacing as a visual enhancement, not a visual reduction, BTW too.
Just only a few hard boiled TV-engineers trying to keep that myth alive
by not wasting a thought on what they say...

OK, here we go...

Basis of the proof:
(Continue reading)

Permalink | Reply | 17 comments |
headers
Dan Streetman | 1 Jun 2007 17:00
Picon

Re: [Fwd: Re: DV: deinterlacing at the right moment]

I think you are downplaying the "advantages" of interlacing.  Granted,
progressive is ALWAYS better if frame/refresh rate and bandwidth are
not issues.  But, they usually are.

Bandwidth is set at 25 Mbps for DV.  Using your "mathematically
lossless" example you would have to record progressive DV at (for PAL)
50 fps.  That is twice the bandwidth of interlaced DV at 25 fps.  So
unless you drop quality by increasing compression it simply won't fit
in the bandwidth limitation.  If you decrease framerate to progressive
DV at 25 fps, then high motion scenes may suffer (note, _may_), and as
you said, converting interlaced DV at 25 fps to progressive DV at 25
fps _is lossy_.

As far as 720p vs 1080i, saying that 720p is always better than 1080i
is misleading, as that is not always true.  I think all agree that
1080 is better than 720 as far as resolution.  A good deinterlacer can
relatively successfully turn 1080i into 1080p, especially in
low-motion scenes, and at double the bandwidth (e.g., a hardware
deinterlacer during playback).  I propose that the human eye can tell
the resolution difference between 720 and 1080 before it would notice
interlacing artifacts from the 1080i->1080p conversion, especially
with a good deinterlacer in low motion scenes.

But to get back to this thread - converting interlaced DV to
progressive for DVD - that conversion is _not_ lossless, as DVD cannot
be twice the framerate.  In my opinion, it is much, much better to
leave the footage interlaced.  Then, either the DVD player will use
its built-in hardware deinterlacer which will almost certainly produce
better results, or if the TV is interlaced (which _many_ TVs are) then
the interlaced material will just play natively on the TV, which will
(Continue reading)

Permalink | Reply | 16 comments |
headers
Levente Novák | 1 Jun 2007 17:40
Picon

Re: [Fwd: Re: DV: deinterlacing at the right moment]

On Fri, 2007-06-01 at 11:00 -0400, Dan Streetman wrote:
> I think you are downplaying the "advantages" of interlacing.  Granted,
> progressive is ALWAYS better if frame/refresh rate and bandwidth are
> not issues.  But, they usually are.
> 
> Bandwidth is set at 25 Mbps for DV.  Using your "mathematically
> lossless" example you would have to record progressive DV at (for PAL)
> 50 fps.  That is twice the bandwidth of interlaced DV at 25 fps.  So
> unless you drop quality by increasing compression it simply won't fit
> in the bandwidth limitation.  If you decrease framerate to progressive
> DV at 25 fps, then high motion scenes may suffer (note, _may_), and as
> you said, converting interlaced DV at 25 fps to progressive DV at 25
> fps _is lossy_.
> 

Yes, it is the exact same reasoning I gave to Stefan in a private mail.
The source (DV 25) _is_ already interlaced and the best you can do is to
leave it interlaced when converting to mpeg2.

Levente

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
Permalink | Reply | 15 comments |
headers
Lars Täuber | 1 Jun 2007 18:47
Picon

Re: mjpeg avi files

Hi Dan.

> I am curious to know... after editing, what do you want to do with the video? 

Store it for later years. ;-)

> Do you have a HDTV and HD player device with which you will show it?

No, I don't. Neither I own a dvd-player nor a tv set. All video material is shown on a (linux) computer.

I'm a hobby filmer. The videos I shoot are family related. They are a kind of documentation.
The only I thing I'm in need is a standardized snow codec from ffmpeg or any other patent free and effective
codec. For audio I use vorbis in a matroska container.

Do you know someone skilled who likes to earn some money (1200 euros or more) for writing a specification for
snow 1.0 and a working implementation? The codec is nearly finished I think.

http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2007-May/030077.html

Best regards.
Lars

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
Permalink | Reply | 0 comments |
headers
Stefan M. Fendt | 1 Jun 2007 20:47
Picon
Favicon

Re: [Fwd: Re: DV: deinterlacing at the right moment]


Hi all,

hmm, there seems to be some confusion on what and more importantly on
why I wrote it...

So, I try to clarify things :-):

1. If you can, record progressive.

2. If you can, record ...

3. If you must record interlaced, either leave it, as it is (if 
   intended to be displayed on a TV-Set...) or, deinterlace it (if 
   intended to be displayed on a progressive Set -> either  
   ProScan-TV or PC) that the field-rate matches the frame-rate.

4. If you intend to "faz" the material (transform to film) record   
   progressive... (if you can't, don't faz it! Or if unavoidable use 
   mcdeint or yuvdeinterlace... These are better than using Reel Smart
   Fields Kit or something other.)

So, why have I written the former text, then? After dealing years with
it, I really get annoyed, if someone mixes up deinterlacing and
decimating frames. The former being (if no simple methods are used) a
high quality operation is alwas brought in miscredit by the later...

cu
Stefan
(Continue reading)

Permalink | Reply | 14 comments |
headers
Dan Streetman | 2 Jun 2007 02:59
Picon

Re: [Fwd: Re: DV: deinterlacing at the right moment]

On 6/1/07, Stefan M. Fendt <stefan <at> sfendt.de> wrote:
>
> Hi all,
>
> hmm, there seems to be some confusion on what and more importantly on
> why I wrote it...
>
> So, I try to clarify things :-):
>
> 1. If you can, record progressive.

This is simply NOT TRUE and very misleading.  Recording progressive is
NOT always the best thing to do.  It depends on many things.  To
further clarify, if your final intention is to display the footage on
an interlaced TV, then it is BETTER to record interlaced!

>
> 2. If you can, record ...

What does this mean?

> 3. If you must record interlaced, either leave it, as it is (if
>    intended to be displayed on a TV-Set...) or, deinterlace it (if
>    intended to be displayed on a progressive Set -> either
>    ProScan-TV or PC) that the field-rate matches the frame-rate.

Well we've already gone over this one and how it's not an option for
most people to double their framerate, certainly not if creating a
DVD.
(Continue reading)

Permalink | Reply | 13 comments |
headers
Stefan M. Fendt | 2 Jun 2007 10:00
Picon
Favicon

Re: [Fwd: Re: DV: deinterlacing at the right moment]

Hi Dan,

Am Freitag, den 01.06.2007, 20:59 -0400 schrieb Dan Streetman:

> > 1. If you can, record progressive.
> 
> This is simply NOT TRUE and very misleading.  Recording progressive is
> NOT always the best thing to do.  It depends on many things.  To
> further clarify, if your final intention is to display the footage on
> an interlaced TV, then it is BETTER to record interlaced!

What YOU write here is simply NOT TRUE. There is only one exception were
interlaced material due to higher temporal resolution _can_ (better:
"might sometimes") be better: Sports.

In every other context the higher temporal resolution does not matter
and the higher spatial resolution of progressive material is a strong
benefit on modern setups. Interlaced displays are dying slowly. All
modern displays leaving our company are progressive (Management decided
interlacing to be a 1920s bug!). There is not a single interlaced
display left in our pipeline, here...

Despite that, before you further recommend recording interlaced, you
might want to read this: 

http://www.atd.net/HDTV_faq.html . 

Even with a lot less nominal spatial resolution 720p is just a lot
better (perceived to be sharper, clearer even in moving parts of the
image) than 1080i.
(Continue reading)

Permalink | Reply | 9 comments |
headers
Raphaël Jacquot | 2 Jun 2007 10:04

Re: [Fwd: Re: DV: deinterlacing at the right moment]

Dan Streetman wrote:
> This is simply NOT TRUE and very misleading.  Recording progressive is
> NOT always the best thing to do.  It depends on many things.  To
> further clarify, if your final intention is to display the footage on
> an interlaced TV, then it is BETTER to record interlaced!

this argument is pretty bogus, considering most tvs nowadays consist of 
either plasma or LCD flat panels, which are inherently built for 
progressive scan.

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
Permalink | Reply | 2 comments |
headers
Richard Neill | 2 Jun 2007 10:40
Picon
Picon
Favicon

Re: [Fwd: Re: DV: deinterlacing at the right moment]

Dear Stefan,

> This is valid even for those sub-optimal frame-modes, 
 > some cameras (XL-1, NV-GS 400/500) do
> offer.
> 

I'm interested in this point, having just acquired a 2nd-hand PAL XL1 
for filming. At the moment, I'm capturing from the camera, and using 
ffmpeg to deinterlace, then convert to mpeg4 for storage/streaming. 
Would the "frame" movie mode be better for this? What exactly does that 
do - I'm afraid I never got the manual when we bought it.

Am I correct in understanding that:

  50 half-frames/sec  gives the best temporal resolution, and the best
     flicker-reduction on an interlaced display (but suffers from chronic
     "mice-teeth")

  25 full-frames/sec is the best overall, (especially if  we can shoot in
     25p, and haven't had to derive this from a 25i signal)

  50 full-frames-sec can be derived from 25i with perfect deinterlacing,
    but at a cost of a doubled frame rate.

  25 full frames/sec, derived in software from 25i is a rather ugly 
hack,
     necessitating some data loss.

Thanks,
(Continue reading)

Permalink | Reply | 2 comments |

Gmane