Deflicker - a short question

Support for my VirtualDub filters
Post Reply
SteveMcKlick
Posts: 3
Joined: Sat Mar 03, 2012 8:24 am

Deflicker - a short question

Post by SteveMcKlick » Sat Mar 03, 2012 9:30 am

Hello friends,
i am new to this forum, but I think, my question is not discussed here earlier.
It is a question or a suggestion for designing a deflicker filter for virtual dub.

I am NOT a programer, therefore I just can hope, that this idea will become a seed
or maybe there is already a solution.

OK - here is the problem:
I do have a Highspeed video - made by myself - and here the flickering of a neon bulb
or most of the artificial light caused by the 50 hz flickering of the current line is
pretty visible and annoying.

But: In my video, the flickering is NOT even over the whole picture.
There are regions - whith high reflection - where the flickering is very visible
and other regions (the dark parts, not very reflective)
where the flickering is hardly visible.

If I have understood the "Deflicker" of Donald Craft correctly,
there is the possiblity with the "window" (which I would call "number of frames") which are used to make
an averaging of the flickering in light and dark - over the number of frames.

This is very helpful, if the flickering is more or less even over the whole picture or frame.

BUT - for areas with different flicker-behaviour, this is NOT very helpful.

So, maybe the deflicker can be improved with this:

1. to divide the frames of the video in small tiles (smallest tile would be a pixel)

2. to analyse over a number of frames the "flicker-frequenz"
I think, this could be done automaticely, using a FFT-Analysing program, since the variation of the "light/dark" could be different in amplitude,
but should have the same frequency on nearl ever pixel.

3. Searching for the average "Light/Dark-Value" but NOT for the whole frame, but for the tiles or pixels.

4. Applying the results value to the pixel for getting the result video after filtering.



OK - I honestly do hope, that my description was understandable,
and I am eager to hear your answers...

All the best and thanks for your work

Steve

User avatar
admin
Site Admin
Posts: 4382
Joined: Thu Sep 09, 2010 3:08 pm

Re: Deflicker - a short question

Post by admin » Sat Mar 03, 2012 9:49 am

It's a good idea and the need for something like this has been apparent for a while. The problem for me is I have no time or real personal need to work on it. :(

SteveMcKlick
Posts: 3
Joined: Sat Mar 03, 2012 8:24 am

Re: Deflicker - a short question

Post by SteveMcKlick » Sat Mar 03, 2012 11:27 am

Hello, are u Donald himself?

I am happy, since
a) you answered so fast
b) my problem seems to be understood

Here are my questions:

a) do you have enough knowledgde, to know, that such a deflicker for virtual dub is not already somewhere programmed?
you seem to have deeper contact with avery lee - therefore, you seem to be a person of the core of virtual dub - is that correct?

b) your own deflicker filter - the one you already programmed? Do you think, this one is
- open source
- possible to be changed into the filter I suggested?

c) are u the only one attending this board, being able to do programming or are therer also other guys around

d) in which programming language is your filter programmed?


ok - thats it so far!

Thanks for your answer, your filter and this board!!

Steve

User avatar
admin
Site Admin
Posts: 4382
Joined: Thu Sep 09, 2010 3:08 pm

Re: Deflicker - a short question

Post by admin » Sat Mar 03, 2012 11:44 am

SteveMcKlick wrote:Hello, are u Donald himself?
The one and only. :lol:
a) do you have enough knowledgde, to know, that such a deflicker for virtual dub is not already somewhere programmed?
I am not personally aware of any such filter as you described.
you seem to have deeper contact with avery lee - therefore, you seem to be a person of the core of virtual dub - is that correct?
No, I am not involved with VirtualDub core development. I have worked with Avery in the past when I was interested in VirtualDub filters, i.e., before Avisynth took over. Nowadays I have almost zero time to work on VirtualDub filters.
b) your own deflicker filter - the one you already programmed? Do you think, this one is
- open source
- possible to be changed into the filter I suggested?
Yes, it is open source and is available at my main web site. Sure it can be used as a base for what you propose.
c) are u the only one attending this board, being able to do programming or are therer also other guys around
There are several great developers that attend this board. Getting them interested in this project may be difficult, however.
d) in which programming language is your filter programmed?
Visual C++.

SteveMcKlick
Posts: 3
Joined: Sat Mar 03, 2012 8:24 am

Re: Deflicker - a short question

Post by SteveMcKlick » Sat Mar 03, 2012 1:46 pm

Hello Donald,

I have a question which is out of topic.
I have looked at your linke "quantum musing" and was very astonished, since that
paper looks like a real scientific paper, talking about deep insight of quantum things.
Did u really wrote this paper or is this just a coincidence of names?

I am astonished to read this, since a this very day I made some notes on my computer,
more questions about "quantum things".
I am NOT a person with deep knowledge in this issue,
but I am interested and I was reading a book about quantumtheorie from John Gribbin: In search of Schroedingers Cat.

It has been considered as "pretty good" - by some scientific magazines- but it makes me nervous at certain pages.
Of course this book just had to stay at the surface, but I really would like to read this book,
making notes and have a person who can answers some questions.
I dont know, whether I am just not getting the point,
but sometimes it seems, that the physicist gave up some basic philosphical fundaments,
which could be said as a consequence in thinking out of the results of some basic experiments,
but - I dont know, wether they are serious or not - I think, if the consequences are correct,
than a lot of other experiments wouldnt make sense anymore.
The core experiment of quantumtheory - according to Feynman - is the how light - or electrons -
do behave at the double-slit-experiment.
And the core-strange result is
a) that the interference pattern could be also be seen, even if the rate of photons is so low,
that it is only one a time - so nothing to interfere in "old thinking"

b) that the photon "seems to know" that we "want to watch" and therefore "behaves different"
if one slit is closed or not.

For me - the expression of "the photon seems to know" is very difficult and seductive and not clear.


BUT - i think I stop here, to wait for your answer. Maybe you are not interested in quantumtheory at all.

And forgive me - I am just interested - unfortunately I do not have any deep knowledge.

Cu
steve

User avatar
admin
Site Admin
Posts: 4382
Joined: Thu Sep 09, 2010 3:08 pm

Re: Deflicker - a short question

Post by admin » Sat Mar 03, 2012 7:56 pm

SteveMcKlick wrote:I have a question which is out of topic.
I have looked at your linke "quantum musing" and was very astonished, since that
paper looks like a real scientific paper, talking about deep insight of quantum things.
Did u really wrote this paper or is this just a coincidence of names?
Yes, I wrote it and yes it is a "real" scientific paper. I'm thinking of linking all my scientific publications. I have about
7-8 papers like that, in several fields. Also some pure philosophy papers.
I am astonished to read this, since a this very day I made some notes on my computer,
more questions about "quantum things".
I am NOT a person with deep knowledge in this issue,
but I am interested and I was reading a book about quantumtheorie from John Gribbin: In search of Schroedingers Cat.
You need to be very careful with these popular accounts. They usually just want to peddle quantum mysteries (the quantum magic show)
and do not give a balanced treatment of the classical and semi-classical alternatives.
The core experiment of quantumtheory - according to Feynman - is the how light - or electrons -
do behave at the double-slit-experiment.
Not to diss Feynman, but the situation is much clearer these days. The
"single particle interference" can be explained classically if you have photons as extended wave packets
that can split and interfere just like normal waves. The discrete "click" is an artifact of the behavior of the
detector. So light is a classical EM field and the apparent discreteness results from quantised emission and absorption
by matter. The real core experiment is the EPRB experiment, where correlations greater than can be
accounted for classically are claimed to be observed. I'm writing a new paper that provides a classical account of the Weihs
experiment. There are some very surprising things going on that haven't previously been noted.
For me - the expression of "the photon seems to know" is very difficult and seductive and not clear.
For me it is just plain silly.

User avatar
admin
Site Admin
Posts: 4382
Joined: Thu Sep 09, 2010 3:08 pm

Re: Deflicker - a short question

Post by admin » Fri Apr 13, 2012 10:53 pm

Just for fun I linked some of my papers under the topic My Papers in the main site menu.

http://rationalqm.us/papers/Papers.html

Post Reply