[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Green light




-----Original Message-----
From: JC Soret <jclement at globalnet.co.uk>
To: Mikael Reichel <mikael.reichel at culinaire.se>; telecine at alegria.com
<telecine at alegria.com>; telecine-digest at alegria.com
<telecine-digest at alegria.com>
Date: den 15 mars 1998 20:32
Subject: Re: Green light

In reply to JC Soret

>> Its intensity? Come on Peter!! Intensity IS number one issue how else
>could
>> you cope with density?
>
> Yes but as someone else mentioned it is intensity in relation to how much
>light the capture device needs,

This is a good observation and makes the issue a bit more flavored. Further
down I will explain why this is so. This observation only holds true if the
sensors (PMTīs or CCDīs) are so sensitive that their ouputs have to be
attenuated or ND filters applied prior to light hitting them?

I was told by Philips themselves that the
>reason why the chroma sensors are only half resolution of the luminance
>sensors is that it allows them to be two times bigger in surface in
>consequence get two times more light.

Actually, by making the sensors double the width - four times more light hit
them as we are talking about a surface here. A line is finite and must have
both length and a width in practical life. So this is a very clever
compromise that in turn build on two all-important factors.
1. The eye is far less acute to color detail than to luminance.
The relationship is about 1/3. Not hard to understand if you think about
that when you mix RGB electrically to get white, the intensity goes up three
times. In the Spirit case it is not a matter of "getting away" with
something more than actually understanding and making best use of how we
humans actually operate.
2. Our perception of resolution is surprisingly NOT first and foremost
related to the technical resolution figure (and extremely NOT with
interlace) but much more to the intensity/contrast.

A few years back in the early CCD days, there was a heated debate over
camera resolution versus sensitivity when manufacturers began boasting
higher res. products. Many did not live up to that high res. feeling despite
great "on paper" figures. As a consequence CCD sensors became equipped with
microlensenses and that sort of covered up the issue.

The bottom line has a name, MTF - modulation transfer function. Its is a
measurement how much of full intensity beween black and white changes (i.e
contrast) you can get before resolution falls below a certain point. It is
basically how fast (steep) this fall is, that determines why we may percieve
a device with "on paper" less lines, pixels or dots to have more resolution
(detail) to one with higher number of lines but with worse MTF. The same
logic applies to cameras, monitors, scanners etc.

How many do not have a monitor that has "lost" its resolution? Well it
hasnīt, not in terms of frequency response but to the eye it has, as it in
its older age has lost intensity/contrast. It still scans the same dot pitch
structure. Inversely, why do Trinitrons appear to have more res? They donīt,
Trinitron technology just allows higher light output than other tube
technologies.  A note - contrast ratio of some Plasma screens are claiming
whopping 400:1, hope they have intensity to match it.

Every device in the chain to the viewer will reduce this MTF function
slightly. Analog more than digital. This is why it is so important for the
end result that scanning is always done in as high res. as is possible. This
will actually improve the end results even though this image is eventually
consumed in a lesser resolution domain.

In addition to this, while the eye has an RGB like imaging structure, the
brain (without which the eye would be of little use) interprets these data
in a Y-C like way. We SENSE in hue, saturation and lightness domains. The
sensor arrangement in the Spirit is separated in a luminance and chrominance
part  so that its is possible to get that detail without having to "suck
out" the chroma and film noise. Instead chroma can be read out with four
times higher light output than the double res sensor can. In addition, I may
be wrong, but believe the RGB chroma sensors are actually offset half a
pixel to one another which adds a positive effect on detail registration.

To make the chroma sensors the same
>size of the luminance sensors would have meant in order to keep the same
>signal to noise ratio as it is now, to put such a big bulb in the Spirit
>that the electronics would have melted within minutes :-) I am wrong ? In
>any case Spirit is the telecine that gives you today the best S/N ratio !

If Kodak/Philips felt that it was neccessary to have 2k chroma res. then
what would have stopped them adding a few more to the luma array?

/mike reichel


---
Thanks to Complete Post L.A. for supporting the TIG in 1998..
No product marketing allowed on the main TIG.  Contact rob at alegria.com
955 subscribers in 36 countries on Mon Mar 16 14:53:21 PST 1998 
complete information on the TIG website http://www.alegria.com/tig3/