Hi -
> From [EMAIL PROTECTED] Mon Jun 2 22:39:54 2003
> > Hmmm, the common problem mentioned has been "splotches of grey" in low
> > light scenes - hadn't heard 'ghosting' mentioned before.
>
> That's the same problem, just differently described. If you look at each
> individual P or B frame, there are "splotches of grey" where, er, things
Similar symptom but slightly different problem. Hmmm, how does
one view standalone B and P frames? I know I frames can be
decoded and viewed by themselves but not the other types.
> Each time something moves it drops a new grey splotch and the previous
> splotch fades a bit, so you get an odd fading trail, or a grey blur,
> from a moving object -- depending on how fast it moves.
Ah, that describes a different problem than the moving blocks of
grey that I have seem in dark scenes. What I was seeing (by looking
at still frames converted to PPM) was that what appeared to be 'black'
to the eye was really many slightly different colors - the U and V
components were just different enough that the encoder thought it saw
moving colored blocks. A fairly effective way of reducing the
effect was to use 'yuvmedianfilter' on the chroma only.
> Just to make sure I have this right, a higher quantization means a
> *lower* degree of information recorded for that macroblock/frame, right?
Correct.
> through my current set of DV sources and see if I can find a sample that
> has this issue. Sadly, I only watched 'Baby Cart in Hades' after I
> deleted the raw DV; it shows this really, er, well.
I see later on that you mentioned the data originated as analog from
broadcast TV of a movie that probably wasn't the highest quality in
the first place.
> > -q 2 is extremely low - a value of 5 or perhaps 4 is about as low as...
>
> OK, that makes sense. I figured that I would need to play with the
> numbers to get -q 2 working sanely, but it actually seemed to be fine,
> so I left it alone.
I think part of the problem you're having is that "-N 1.5" and
"-Q 1.5" are combining to toss out a lot of information and '-q 2'
is then being very careful with whatever quality still remains.
As it turns out the "-N 1.5" setting is more aggressive than the
comment "mild noise reduction" would indicate.
You'd be far better off using "-N 1.0 and -q 4" or similar.
> I have tried, previously, to find a rough guide to what -q value to give
> for fitting a given length of time to a target size, but failed.
When you specify "-q" you're telling the encoder to use VBR
encoding. At that point "-b" sets the MAX bitrate, and "-q" says
how hard to push the encoder UP TO the MAX.
> Is there any rough guideline you can suggest for trying to pick the
> quantizer?
Sure - I've never been known to back away from expressing an
opinion or two ;-)
This assumes that the target is a DVD. VCDs and SVCDs are so
bitrate limited (SVCDs have a max of 2500Kb/s for 480x480/480x576)
that usually you're concentrating on getting the bitrate down
and hopefully not killing the quality completely.
For DVD with good quality sources I use either "-q 5" (for more playtime
on a single disc) or perhaps 4 (if I know that I have lots of space).
-N between 0.5 and 1.0 (higher setttings are useful for low quality
sources or if the destination is VCD).
On one set of DVDs I'm in the process of creating now I only need to
fit about 1hour maximum of video on a DVD - thus the only constraint
I have is to keep the bitrate under the legal DVD maximum. "-q 4
-N 0.6 -b 8500" is working well with the average around 7700 and peaks
up to ~8700.
> From: [EMAIL PROTECTED]
>
> I have experienced the same or very similar effects (mainly on the
> chroma channels), but in my case the ghosting is actually in the
> original DV. I hadn't noticed if mpeg2enc made it any worse, but
> Like Steven, I'm surprised you get such low bitrates using -q 2. I
> can never get lower than -q 4 for DV stuff.
Indeed. The HF stuff was rolled off, the quantizatin for active
blocks increased (-Q) and then -q 2 was used to try and get high
quality out of what was left. NOT saying that's causing the
ghosting (my suspicion is that the original has some Y/C lag similar
to what your camcorder has in low light conditions).
> From: Daniel Pittman <[EMAIL PROTECTED]>
>
> Right, I had a play around with the footage I do have in hand in DV
> format and couldn't reproduce the problem even when I tried very hard to
> make it generate artifacts; it stayed good, if blocky, even at stupid
> So, either this is some sort of noise effect in the source as was
> suggested, and so specific to the source, or I need darker material to
> play with.
So it's specific to the one movie. That really makes it sound
like there is something not 100% right with the original movie
that started the thread.
> Sadly, I don't have the DV source to that movie around any longer; it
> only occurs very occasionally in things and I don't tend to keep the
> source around for long after encoding finishes.
At 12GB/hr (for NTSC at least - PAL's a bit larger) I can understand
not keeping stuff around for a long time. Couple weeks ago I
bought a 200GB drive (with discount and rebate it was only US$149)
and it's 59% full already. Time to start doing some encoding
and reduce the backlog ;)
> While it could conceivably be enhancing noise that already exists in the
> source footage, two things convince me otherwise:
>
> 1. the effect is *not* present in the I-frames.
> 2. the effect is very low quality footage while the rest of the film is
> considerably better, including brightly lit scenes.
Actually #2 would lead me to believe it is in the original - "very low
quality" means there is most likely 'noise' present (not necessarily
visible to the eye). "Noise" in the either the sense of "speckles",
"static" , anyother "corruption" (chroma lag, etc).
> I noticed you discussing that the other day -- my video source is an
> analog to DV bridge box converting footage broadcast via analog cable
Canopus or one of the other similar units?
> television, so it's generally pretty good quality.[1]
Digital TV or analog? In the US, at least where I live, broadcast
(over the airwaves rather than cable) TV is fairly to very poor
quality.
> The film in question was a 1970s Hong Kong picture, though, so it's not
> the cleanest material I have ever seen.
Ah ha! Thought so ;)
> *nod* This does not seem to be the same problem as you experienced with
> your camera, although some of the effect is similar.
There are evidently several problems that have similar looking
visual effects. I think Dan Scholnik's camcorder Y/C lag is
a different problem than the ghosting you're seeing, and the
moving blotches of grey that I see is yet another problem.
Cheers,
Steven Schultz
-------------------------------------------------------
This SF.net email is sponsored by: eBay
Get office equipment for less on eBay!
http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5
_______________________________________________
Mjpeg-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users
