[Theora] video source

[Ralph Giles]

On Wed, Sep 15, 2004 at 08:40:36AM -0500, John Kintree wrote:

> In order to experiment with the theora encoder, I bought a mini-DV camcorder a 
> couple of weeks ago; a Panasonic PV-GS9 for $250.  I wanted a mini-DV because 
> it is claimed that capturing video from the camcorder to a computer over a 
> firewire cable results in an exact copy of the recorded video.

This is true, though the image quality you get from an affordable video 
camera is generally disappointing. Another option is a firewire webcam; 
the image quality won't be any better, and it won't be useful as a 
camcorder, but it will be a bit cheaper and return uncompressed images.

> That was not my experience.  The video I captured to my computer over the 
> firewire cable, using both MainActor and Kino, was not equal in quality to 
> the video as it displayed on the camcorder's own LCD display, or on my TV 
> when connected to the camcorder with the provided A/V cable.  When I played 
> the captured .dv format files, that were about 225 MBytes per minute, the 
> video was noticeably jerkier, and stair-stepped on diagonal lines, than the 
> video that was stored on the mini-DV tape.

There could be a couple of things going on here. It's quite possible 
the camera is using the pre-compression image in the viewfinder or 
analog out. Obviously you can check this by comparing playback from 
the mini-dv tape instead of using a live feed. What's on the tape will 
always be lossily compressed, and should be bit-for-bit what goes over 
the firewire.

It could also just be a playback issue. That is, the file on the 
computer is the same as what the camera would play back, but the 
software dv codec doesn't do as good a job reconstructing the images,
probably as a speed optimization. How did your computer compare to the 
P4 you tried at the store?

Jerky, in particular, should never happen, and indicates a speed 
problem either with capture or playback on the computer side. Unless 
the camera is defective, of course. Someone's dropping frames.

Note also that things like edge aliasing (stair-stepping) won't be as 
obvious on the camera's LCD, because those aren't full resolution and 
the image there is subsampled. A good TV will show them though.

You might try one of the public dv test clips to try and sort 
camera/capture issues from playback. For example 
http://umn.dl.sourceforge.net/sourceforge/libdv/pond.dv

> Since there is a loss of quality when the video is compressed, I want to start 
> with as good a quality video source as possible.  I know this is sort of 
> off-topic, but I don't know anyone else who has done a firewire transfer with 
> a mini-DV camcorder, and wondered if anyone here has had, or knows someone 
> who has had a similar experience, or was my experience atypical?  

My general experience is that  the results over firewire are completely 
reasonable, with no obvious quality degredation. One can see 
compression artefacts if one looks for them, but they're generally 
dwarfed by limitations of the source image in consumer camcorders.

I have a Panasonic DV851D (which I paid a lot more than $250 for 3 
years ago...wow!) which works fine. I've also used a friends Sony 
Digital-8 camera, several professional cameras, and the output of a 
mini-dv deck digitizing input from and older analog camera, all with 
none of the issues you describe. So either you're unusually sensitive, 
or something was definitely wrong.

Hope that's of some help,
 -r