On Wed, 27 Oct 2004, Derek Fountain wrote:

> :) No, definitely not a troll! I was considering writing a piece for Newsforge

        Ah, ok.

> The thing is, I only have a DC10+ card, which are increasingly difficult to 
> get hold of these days. I have been blissfully unaware to date of what people 

        Indeed.  They haven't been made in quite some time - the only way to
        get one is eBay or similar.

> might do with the toolset without such a card, hence the question.

        Add more tools to the toolbox of course :)

        Kino and dvgrab are the two main tools.  An ancilliary part of kino
        called "smilutils" is also used to manipulate the DV files after
        capture and editing.

        Kino/Dvgrab/smilutils can all be found at/around:

                http://kino.schirmacher.de/

> /me goes off to google to read about these things. What software do you use to
> capture from these cards, and what format do they capture in? Assuming they 

        The Canopus (http://www.canopus.com) units are not a "card" (well,
        there is one model that is a card but most aren't) but an external
        box (the -100 and -300 are the size of a small paperback book while the
        'pro' equipment like the -500 and -1000 are rackmount,etc) that
        accepts the analog source video and audio (VCR) using S-Video/composite
        cables and then attaches to the computer by means of a IEEE1394 cable 
        ("Firewire").   The analog data is converted to DV data which is 
        transmitted (for consumer devices) at 25Mb/s rate (well, ok - add 
        1.5Mb/s for the audio ;)).

        The only thing the computer needs is a cheap "Firewire" card (which 
        can also be used to attach external discs).

        I use 'dvgrab' to capture the data (12GB/hr) and then 'kino' to 
        edit (cut out the commercials, splice new scenes in, whatever).  Kino
        can capture of course but I prefer a commandline/batch method of
        bulk capture.

        Or if you have a MAC around you can use FCE (Final Cut Express) or
        FCP (Final Cut Pro) to do the capture and editing.  The DV files that
        can be exported from the Apple tools are acceptable as input to
        'smilutils' and so on.  You could of course encode using Apple's
        Compressor program but I've found that mpeg2enc and friends to work
        quite well on OS/X.

        With "raw" DV files you can even use 'dd' as a (very) crude editor
        since the records are fixed length (120000 bytes for 525line video
        systems, 144000 bytes for 625line systems).

> don't capture to MJPEG, you presumably have to convert to MJPEG (in a 
> non-lossy way) in order to use the tools...?

        Ah, slight misunderstanding there - I think I can see why you might
        be on a quest for more info ;)

        Once the data is captured using the lavrec (or other mjpegtools 
        recording/capture tools) the OUTPUT of those is a "YUV" (IYUV, aka
        4:2:0 planar, 12bit/pixel) variant (it's just 4:2:0planar with a 
        header and FRAME markers) called YUV4MPEG2.
        
        The ONLY format that the encoder understands is YUV4MPEG2 (4:2:0p with
        header and frame markers).  The encoder does NOT understand MJPEG! 
        And neither do any of the filters.

> >     Exactly.  The code for capturing in mjpegtools is a relatively small
> >     portion of the total - the bulk of the code resides in the (wide)...
> 
> Yep, OK, but I assumed these tools only worked on MJPEG files, and you only 
> got MJPEG files from a DC10+ card or similar. Shows what I know... :o}

        Nope - almost NONE of the programs understand MJPEG.  The capture 
        program and the decoder program (which turns MJPEG into YUV) are the
        about the only things which really know about MJPEG.

        'smilutils' decodes DV data to YUV4MPEG2 (and 'ffmpeg' and 'mplayer'
        can also produce YUV4MPEG2 formatted data).  Once that's done as the
        first step then the rest of the programs in mjpegtools enter the scene.

        Some of the tools (the filters) can work on other chroma samplings
        such 4:1:1 or 4:2:2 but before the data enters the encoder it MUST
        be scaled/converted to 4:2:0 since that's the only chroma space 
        supported (it's also the only one you can put on DVDs/SVCDs/etc ;)).

        More clear now? :)

        Cheers,
        Steven Schultz



-------------------------------------------------------
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
_______________________________________________
Mjpeg-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users

Reply via email to