Hello Bernhard,
Sorry again that I haven't gotten back to this sooner. It seems that
work severly restricts my other interests.
I have found the source of the image artifacts in my lavrec recordings
that don't appear in my Pinnacle driver based recordings.
>>> The recording command used was the following: lavrec -d 1 -g
>>> 608x464+16+8 -q 40 -f a -i n -B -t 10 -s -U xxx.avi
> The artifacts do not change when you use a higher quality like -q 60 ?
> -q 100 should be avoided, it might cause problems.
I did other experiments and found that the image always seemed to
be the same going from -q 40 up in steps all the way to -q 100.
It didn't look like a compression based problem to me at this point.
I had some time to experiment this Saturday, so I looked around
the forum a bit and found a thread from a while ago that had the
word checkerboard in it. This is what the image problem looked like
to me. That thread was focused on output not input, but it appeared
to have been solved by changing some configuration on the
luminance/chromanance filter. I looked for things of a similar
nature in the saa7110 chip configuration and found that when I changed
the settings for the two bit APER value in control register 0x06,
I got different video appearance. I actually got a the video recorded
with lavrec to look just like the Pinnacle based video. From the
graphs in the saa7110 spec sheet, it looks like changing the APER
value effectively changes the width of the filter notch. The standard
value in the saa7110 driver uses a narrow notch. I changed it to the
widest by using APER value of 0, and the checkerboard disappeared. The
image does appear to be a bit softer though. So it looks like there
is a compromise between a bit sharper image with the annoying
checkerboard edges, or a bit softer image with no edge artifacts.
I imagine the optimum setting for APER would vary depending on the
input signal quality, but I'm happy to have a softer image that is
consistent across my various input signals.
If anyone is interested in seeing the difference between the standard
APER settings and my experiment with APER of 0 I can put up some
samples.
On field switching.
> You should be able to correct interlacing problems with yuvcorrect.
> Please take a look at the mjpeg-howto section 3.4 Notes about "interlace
> field order - what can go wrong and how to fix it"
> (https://sourceforge.net/docman/display_doc.php?docid=3456&group_id=5776#ss3.4)
> What you describe sound like that the card is not able labe the movies
> accorintg to what it is dooing. Does the interlacing change during the
> recording (like when you have dropped/ins. fields) ?
Thank you for these comments. I have used these tools some when I captured
something from a TV broadcast that I couldn't record again. I could get
the video to look proper in those cases. Still it seems odd that this
problem of getting incorrect field order or field sync doesn't
appear to occur using the Pinnacle drivers. I have a couple ideas on this
that I will try when I get the time. For now, I will do my recording with
xawtv running. That seems to make the field order correct when recording
with lavrec.
Thank you very much for your time. I really appreciate the suggestions,
and thank everyone who works on this project for providing these linux
based video tools.
Andrew Piecka
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Mjpeg-users mailing list
Mjpeg-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mjpeg-users