[Tig] Subject: The Technicolor look

[Richard Kirk]

Hi.

Following from Jeff's post. I wrote the Technicolor plug-ins for 
Baselight, so I can add a bit of background. Plus, it ws fun and I like 
talking about it.

You will not get a plug in that will give you the 'Technicolor look' in 
one go. To get that, you will probably have to balance the camera and 
light the set the way a Technicolor cameraman would have done, with 
Natalie Kalmus looking over your shoulder. The Baselight plug-ins try 
and match the physics. This produces very vivid reds, which may go out 
of gamut, so you may wish to reduce the saturation.

The early Technicolor process 1 & 2 used only two colours.

Process 1 used two filters in the projector and the camera.
This was not that hard to simulate - we could guess the two colours of 
glass they would have used in the projector. Anyhow, hardly anone has 
ever seen it.

Process 2 used two dyes, cemented face to face. There were many choices 
of dyes: the orange-red and blue-green used here are similar to the 
colours used in 'Ben-Hur'. This gives odd-looking skies,  and  
salmon-pink tones for flesh and blonde hair. Orange and green dyes were 
also used - these gave better flesh tones, but could not achieve a 
mid-tone grey. The concentration and mixture of the dyes were varied to 
suit the material - something we cannot do with the modern processes. 
However, we can vary the contrast control to simulate the variations in 
dye density, and the white hue to give a convincing balance for the 
flesh tones. I have not been able to get good spectral information about 
the dyes used in this period, so there is much guesswork. It would be 
possible to simulate other dyes with the tools I have developed for this.

Process 3 printed two dyes on the same substrate. This was a better 
engineering solution, but the dyes and the colours were the same as 
process 2. This is probably a reasonable simulation for 2-colour 
Kodachrome, though the red was more orange.

Process 4 used a 3-strip camera - we have estimated the response of this 
to typical subtractive colours - and 3 dyes. We have chosen 3 azo dyes, 
which probably represents the 1950 reprints of films such as The Wizard 
of Oz, rather than the original prints, which used a mixture of dyes for 
each separation, and varied from one film to another. The process also 
had a light grey exposure of the green channel in the black and white 
film base, which also carried the keylines and the sound track.

Process 5 took the print off the Eastman tripack negative. We have 
simulated the exposure in a typical negative material. Our 'before' 
colours are a simulation of a Vision print, and our 'after' colours are 
the simulated three-colour print. The grey exposure was usually omitted 
in process 5.

As I have been modelling exposures from negatives, the 'look' filters 
ought to do a reasonable job of translating a film-like image into a 
guess of what the same negative exposure might look like if output using 
the various Technicolor processes. The filters are not restricted to log 
images - they should work on HD too - but the image should look 
'film-like'. The spectral modelling is probably fairly good for 
processes 1, 4 and 5, while 2 and 3 have needed some guesswork on my part.

We could take this further: we could add all the Technicolor lab 
controls, and support for random dyes spectra. I wanted to do the 
green-orange 'Wax Museum' look.

Cheers.
Richard

-- 
FilmLight Ltd.  	Tel: +44 (0)20 7292 0400 or 0409 224 (direct)
Artists House,		Fax: +44 (0)20 7292 0401
14-15 Manette Street    
London W1D 4AP