Auteur: Georges Giralt
Date: 14-04-2006 19:26
Bonsoir !
Bon, mon Américain a répondu. En Chinois, enfin, pour moi ...
Heureusement, il a compris le Français, enfin pour ce qui cncerne les références !
Il est heureux que les gens d'Orsay rédigent en Anglais !
Bon week end !
=============================================
>Concernant l'effet de l'éclairage sur la granularité (effet Callier
>> généralisé), je ne connais que le travail de Pierre Chavel et Serge
>> Lowenthal.
>>
>> Il s'agit de savoir, au-delà de la densité moyenne, l'influence du jeu
>> de condenseurs et de l'éclairage (ampoule, ses dimensions, sa position)
>> sur les fluctuations de densité = bruit de la granularité résultant dans
>> l'image.
[...]
>> Première partie qui explique l'effet Callier sur les densités :
>> Pierre Chavel, and Serge Lowenthal Noise and coherence in optical image
>> processing. I. The Callier effect and its influence on image contrast
>>
>> Journal: Journal of the Optical Society of America, vol. 68, issue 5, p.559
>> Publication Date: 05/1978
>>
>> Deuxième partie qui explique l'effet Callier sur les fluctuations de
>> densité (bruit de granularité)
>> Pierre Chavel, and Serge Lowenthal Noise and coherence in optical image
>> processing. 2: Noise fluctuation JOSA 68, 721-732,(1978)
So I got these two and the other paper, and read through them
quickly. None of their finding explained what I observed, and I feel
that my experience is related to something that falls outside of their
assumption and analysis technique.
They modeled emulsion noise as stationary wideband Gaussian source
added to a low bandwidth signal. They did not assume anything more
about the grain noise besides that the Gaussian noise is of
stationary, wideband nature. This bandwidth creates a third kind of
"aperture" to be considered in terms of coherence in the enlarger
system. The two are usual "source size" and "pupil size." They
analyzed for the mean of the output signal in their part 1, and
variance part 2. I wish I had some real sense of numbers to use for
these variables, but I suspect my enlarger setting is partially
coherent system in terms of source size and pupil size (which is what
matters for the mean value or average density/image contrast), but for
the granularity issue, the noise bandwidth and the resolution
bandwidth become also important. Their analysis is based on limit
values at coherent and incoherent cases and they also give approximate
signal-to-noise ratios for intermediate values.
Their appendix raises some questions about the validity of Gaussian
assumption and approximation used for incoherent cases.
Another question is how wide is the grain bandwidth. With today's fine
grain films developed in fine grain developers, the grain shape is not
much larger than the wavelength so wideband noise assumption may not
hold too well.
So, my current thinking is that, by changing condensers, enlarging
lens and magnification ratio of the system, I was probably changing
the apparent source size and pupil size in relation to the negative's
grain bandwidth. It's still partially coherent, but there might be
some fluctuation not captured by the approximated limit analysis
presented in the paper.
Perhaps what's more interesting to learn from these papers is the kind
of considerations given between coherent and incoherent enlarging
systems in the past.
=================================================
|
|