Archived from groups: rec.video.desktop (
More info?)
> <kmaltby@sbcglobal.net> wrote:
>>
>> If you really meant "virtualdub" then I guess you are talking
>>about some MPEG4/Dvix/Xvid .avi compression. Surely
>>you don't mean the 5:1 DV-AVI ?
>>
>> I would try having Vdub use a different codec to read your
>>file; and using whatever settings are available for the codec;
>>before trying to find such a filter at any of the filter sites on the
>>web.
>>
>>Luck;
>> Ken
>>
"da_test" <davexnet02NO@SPAMyahoo.com> wrote in message
news:nguo51tkj1ju8giac4gn6bconqkahi2944@4ax.com...
> On Tue, 12 Apr 2005 20:31:23 -0500, "Ken Maltby"
> Ken, as you guessed, it is AVI and virtualdub I'm
> referring to, and they are compressed clips,
> divx I think.
>
> I don't think it's a case of hoping another codec will help;
> the right one is being used.
> I was looking at this as a form of chroma noise reduction.
Perhaps the current codec or a different codec has settings
to modify how any "chroma noise reduction" is applied. This
could work if the effects you are noticing were not a result
of the initial encoding but were a matter of how it is being
decoded/uncompressed.
If the effects that you are noticing were introduced during
the initial compression (although what you are seeing may
not even be an effect of compression, but some other factor)
then any correction would be as a change to the legitimate
image. If this is the case then you would have to indicate to
any "correction" process where to apply the change. Other-
wise it would only be possible to have it respond to every
occurrence that matches the mathematical results of an image
analyzing algorithm.
> I've tried a bunch of filters in virtualdub, but they don't really
> address this problem. It's not really noise.
>
Not all Vdub "Filters" actually filter anything, some can alter
legitimate image properties. In the case of your gradient
example, there is a "RegionRemove.vdf" that could fix what
you describe, but it would take a great deal of interaction to
apply the "filter" at each occurrence.
> The reason I provided the jpg is because it gives an indication
> of the banding involved and the conditions.
> The gradient should be a smooth transition from dark to light
> without any of the banding, which is an artifact of the
> compression.
(It can also be an effect of how your graphic program makes
a gradient. In fact there could be a similar involved.)
Are you saying this because you have seen the image before
it was compressed, and it was without this effect then? You
can't see how it looks while it is compressed, so you can't
determine directly if the effect occurred during the compression
as opposed to when it was rendered. The best you can do is
try different rendering codec, and settings to see if you can
isolate the factors involved. If the effect remains under a
number of different rendering codec, then it suggests that the
problem was introduced during the initial encoding (or before).
> Same thing happens in the compressed avi when the scene conditions
> are right. I'm really interested in this from an academic point of
> view - can something be done? The AVI's I have that I noticed this
> are inconsequential.
> Thanks,
> Dave
>
Perhaps a real "Video Gurus" will provide a detailed
explanation of the cause of the effect you are noticing.
Or they may direct you to any of the many web sites,
archived tech articles, and discussions that address
related issues.
You could spend many years just reading all that is
out there on MPEG compression. Most of it tries to
addresses means of producing the most compression
and/or the least negative effects. The amount of
verbiage and mathematical expression expended, for
the tiniest return, can boggle the senses.
Luck;
Ken