On Wed, 29 Mar 2006, [EMAIL PROTECTED] wrote:
Hello list.
I'm trying to digitize old VHS tapes with a DC30+.
I ran into one clearly reproducible problem:
Capturing at PAL/fullframe (lavrec -q50) apparently works perfectly
with nearly no framedrops or A/V sync problems recognized.
Looking at the resulting avi I can see that sometimes the second field of a
frame is shifted one line down. (there seems to be one line inserted at top).
In order to exclude the possibillity of having a field order problem I looked
a a movie sequence with slow horizontal, but no vertical motion. I converted
the avi into a sequence of pnm-images, one for every field (not frame).
scrolling through the sequence (with xv) it is obvious that the temporal order
of the fields is correct (horizontal motion is in sequence). But it is also
clearly visible that sometimes the 2nd field of a frame is shifted down one
line. The 1st field does not have this problem. This of cause completely ruins
deinterlacing. In order to check my analysis wrote a yuv-filter which shifts
the 2nd field up. After this the bad sequences look good (but the good ones
get bad, of cause).
The phenomenon appears and disappeares at random positions of the avi and stays
for a sequence of frames (sometimes seconds or minutes). When I recapture the
movie the changes (problem on/off) appear at different places, but they appear.
This sounds similar to a problem I have with an LML33 (also zoran-based)
card. Sometimes the association of fields into frames would is wrong in
captured video, and piping through "yuvcorrect -T BOTT_FORWARD |
yuvcorrect -T LINE_SWITCH" would fix it.
But I found that once capture is started, whatever field-frame state the
card started in it would continue solidly for hours. So my workaround is
this:
- modify lavrec to display the captured video on-screen while recording
- just keep killing and restarting lavec until it starts up correctly
The right test pattern makes it easy to detect this problem. I found that
the boxy font in the on-screen display of a Horita VITC generator was
perfect. So I enable the display, restart lavrec until its clean, disable
the display and record away.
I'm always using a stable video source, either directly from a camera or
a surplus industrial TBC in front of an SVHS player.
Steve
The problem seems not to depend on the video source (I captured from VHS
recorder (playback and tuner) and from DVBT). I tried two Computers
(AthlonXP/32, Athlon64). I bought a second DC30+ from ebay, same problem.
Software is SuSE-93/original kernel (2.6.11.4), mjpegtools 1.8.
It seems to me that there may be some trigger problem which leads to the DC30+
to sometimes slightly miss the time it switches from the 1st to the 2nd field
so that the second field sometimes begins one line too early.
But: two DC30+ do the same, so don't believe in a defective card.
Is there anyone who experienced the same problem?
Is there a develloper who knows if there is a switch on the DC30+ which could
be relevant in this context? Could my problem be caused by some
misconfiguration of the card?
Any information could help.
Regards,
Ralf Oehler
--
--------------------------------------------------------------------------
| Ralf Oehler
|
| GDA - Gesellschaft fuer Digitale _/
| Archivierungstechnik mbH & CoKG _/
| Ein Unternehmen der Bechtle AG #/_/_/_/ _/_/_/_/ _/_/_/_/
| _/ _/ _/ _/ _/
| E-Mail: [EMAIL PROTECTED] _/ _/ _/ _/ _/ _/
| Tel.: +49 6182-9271-23 _/_/_/_/ _/_/_/#/ _/_/_/#/
| Fax.: +49 6182-25035 _/
| Mail: GDA, Bensbruchstraße 11, _/_/_/_/
| D-63533 Mainhausen
| HTTP: www.GDAmbH.com
--------------------------------------------------------------------------
time is a funny concept
Bitte senden Sie mir keine Word- oder PowerPoint-Anhänge.
Siehe http://www.fsf.org/philosophy/no-word-attachments.de.html
