On Thu, 12 May 2005, Dik Takken wrote:
> a new trick while doing it: Use MPlayer to turn down the contrast setting
> of the XVideo subsystem and set the saturation real high. The 'yuvplay'
> utility does not touch XV settings so it will have the same display
> settings as MPlayer has. Now all that is left to see is pure chroma data.
Cute trick.
> Next, I tried to get rid of all the noise in the chroma data. The CCD
I think that is far too ambitious of a goal. Only way to get rid
of _all_ the chroma noise is to create a black&white video. A more
reasonable goal would be to reduce the noise to a less objectionable
level.
> induced chroma noise is quite heavy in the low-light scene I use for
> testing. First I tried spatial filtering only by using y4mspatialfilter
> and yuvmedianfilter. This approach failed, because the groups of
Hmmm, the report yesterday or so mentioned yuvmedianfilter as being
effective.
> adjacent pixels that change color from one frame to the next are too
> large. Removing the noise would require using a filtering radius of > 15
> pixels or so, which blurs the chroma channel too much.
Did you use the normal/default mode of yuvmedianfilter or the '-f'
averaging mode? Something like "-t 0 -f -R 4" might do a reasonable
job.
But wow - 15 pixel groups of noise? Sounds like it's time for a
camcorder with a better CCD - either that or the CCD (or electronics)
have developed a fault of some kind.
> smil2yuv file.dv | y4mdenoise -t 0 -z 0 -T 12 -Z 12 | yuvplay
>
> The downside of using y4mdenoise is that it takes 5 seconds to process a
> single frame on my machine. Even higher error thresholds do not make it
Use of PentiumIII-550 cpus is discouraged for video encoding :-)
Even on a PentiumIII-800 I see about 1/2 frame/sec with y4mdenoise.
Well, it's a dual cpu system so the other processing isn't competing
with the denoising.
Ah, but "-t 0 -z 0" doesn't disable the luma processing in y4mdenoise,
does it? I think those values cause the most exhaustive time
consuming processing y4mdenoise is capable of! The way to speed
up y4mdenoise is to use something like "-t 2 -z 1" or "-t 3 -z 2".
Or get a faster system <grin>
> smil2yuv file.dv | yuvdenoise -Y 1 -U 1 -V 1 -y 1 -u 10 -v 10 | yuvplay
>
> But this also has a downside: It touches the luma data too because, AFAIK,
But with very mild settings. I think you might actually want that
to help reduce the 'sandpaper' or "boiling sand" effect - something
I have seen with lowlight footage shot with a Digital8 camcorder
(1/6" CCD).
> Could yuvdenoise be fixed to ignore the luma channel when the threshold is
> zero? It seems to me that it would be the ideal chroma noise filter.
Probably be fairly easy - perhaps Stefan will see the request and
put that in along with the new algorithm he mentioned the other day.
Cheers,
Steven Schultz
-------------------------------------------------------
This SF.Net email is sponsored by Oracle Space Sweepstakes
Want to be the first software developer in space?
Enter now for the Oracle Space Sweepstakes!
http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click
_______________________________________________
Mjpeg-users mailing list
Mjpeg-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mjpeg-users
