[Ffmpeg-devel] Analog video grabbing chipsets

Piero Bugoni crboca32
Sat Apr 14 22:12:18 CEST 2007


--- Alex Beregszaszi <alex at rtfs.hu> wrote:

> Hi,
> 
> I need to digitalize old family videos, which are
> 10-15 years old and
> recorded on VHS-C.
> 
> My question would be what chipset/cards would the
> folks here use for
> this work.
> 
> Michael, I remember you had done some patches
> regarding better chroma
> for BT8x8 chipsets. What was the real issue? Do they
> finally merged it?
> 
> Actually I would prefer an USB2 capable chip, as for
> the most time I am
> using a notebook. But if there are no good USB2
> grabbing cards, a PCI
> one would be ok aswell.
> 
> Any comments and ideas welcome. Even grabbing
> setting tips :)
> 


Alex,

I do a fair amount of this. I use LW-104 cards based
on the BT878 chip that I bought on e-bay for around
$25.00 US. (The cards are also sometimes known as the
Pico2000).

More info on these is on my website:

http://Bugoni.net

I have looked around for USB video capture adapters,
and they do exist. I have seen them from $59.00 US and
up. As far as I know, these units have a BT878 chip
inside them.

However, as far as I was able to discern, there may
not currently be Linux support for these.

If the device is a BT878 attached via a USB "Bus"
instead of PCI, then probably Gerd Knorr's bttv driver
could be adapted.

I use v4lctl (a command line utility that is part of
xawtv to set the input and tvstd on the card).

Note: the cards I mentioned have 4 inputs (see the
BT878 data sheet as to why), but only one active input
at a time. (Also, one input is chroma-only. Again, see
the data sheet).

The company that makes the cards also makes them with
4 BT878 chips on them. Generally speaking, if you want
4 simultaneous inputs, you need 4 chips. Each chip
shows up as /dev/video(x), where x is 0 to 3.

I usually use cif as the size to capture, or, possibly
320x240. Bigger sizes may not work. For ntsc, I always
specify exactly 29.97 F/s (not 30). For PAL, I guess
it would be 25 F/s. The card I mentioned does both.

For quality I use -sameq, or -qscale = .1.

I also use -deinterlace to get rid of the "comb
effect" (Thanks to the person here who pointed that
out). 

Anyone who can point out the applications of
interlaced vs. deinterlaced video would be helpful. I
just know that if you do not use -deinterlace, you
will get the "comb effect", but will supposedly suffer
a loss of quality, when you deinterlace.

I usually save as mpeg, or flv, but I am generally
considering saving both video and audio as separate
"raw" streams for later post processing.

Just recently, Fry's electronics had a special for
400G hard drives at around $99.00 a piece. That price
range is showing up more and more on the net.

Good Luck.

Piero.

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 




More information about the ffmpeg-devel mailing list