[Tig] Primary vs. Secondary! The confusion rages on.

Martin Euredjian ecinema
Thu Apr 4 01:16:51 BST 2002

From: "Philip Mendelson" <pmendelson at livewire.TV>

> I think Stan Chayka offered the most accurate definition that I have seen.

I don't think so.  Not today.

Accurate with reference to what?  If the reference is a color corrector
created somewhere back in 1978 as well as industry practices at the time,
perhaps you are right.  However, things are much different today --with the
Sunburst not being even close to a reference for current mainstream
practices-- and, if you include computer-based color correctors, the terms
"primary" and "secondary" --as they might have been accurately defined for
1978-- may fail to show relevance today.  Today you an put the processing
power of the '80's era color correctors in a single FPGA and you can
interconnect several of these in a matrix for incredible flexibility.  What
relevance does a signal-flow-based definition have?

For example, take daVinci's Power Tiers ( http://www.davsys.com/2k.htm )
facility.  Is it "primary" or "secondary"?  Well, Stan's definition, when
refering to "primaries" states that:

"The purpose of making these adjustments is to establish in the picture, the
'correct' color and grayscale content. 'Correct' is not a technical
condition but rather a desired result within technical limitations. As an
example, a snow scene is very seldom balanced as neutral white/gray/black
but artistically a 'cool' or bluish cast is created."

Power-Tiers are used, among other things, to establish a picture from which
to base corrections in the next Power Tier, all the way up to the output.
You setup "layers" of corrections.  In the context of the above definition,
then, "Primary" now depends on the intended utilization rather than the
technical design of the box or the signal flow through the circuitry.

How about the use of LUT's prior to the traditional "primary" correction?

A further look at what is possible today, from a signal-flow perspective
might reveal even more of a flaw in the old definition.
In signal-flow order:

RGB Light control  (Sony)
Sensor control  (PEC, Photodiode or line/area array CCD gain, etc.)
TK Primaries (to use traditional terminology)
TK Secondaries (again, for lack of a better description)
Color-corrector input LUT
daVinci 2K Primaries  (inside and outside window)
daVinci 2K Power Tier 1  (inside and outside window)
daVinci 2K Power Tier 2  (inside and outside window)
daVinci 2K Power Tier 3  (inside and outside window)
daVinci 2K Power Tier 4  (inside and outside window)
Pandora's Mega Layers and Mega Gamma layer..

Given the above rough signal-flow, you could use the old definition of
primaries just about anywhere in the path.  Anything before the point where
you reach a "desired result" in preparation for the more selective work
would be "primaries" and what follows in processing can be considered to be
"secondaries".  It is a common technique to setup a baseline image from
which more selective control can be had.  That doesn't have to happen in the
traditional "primaries" section any more.  And this image does not have to
look anything like the desired output.

The confusion of the situation is exacerbated if you consider sofware-based.
To use a still-photography example, in Photoshop, you can apply such things
as "Auto Balance" repeatedly after applying all sorts of other "secondary"
color corrections.  The program builds a sequential list of the processing
applied.  From that list one may selectively disable/enable one or more of
the processing steps previously applied to the image.  What is "primary" and
"secondary" then?

My point is that a signal-flow view of this concept of "primary" and
"secondary" color correction hasn't been applicable for a long time and
that, any definition that is based on this perspective --though historically
correct-- has little relevance to "primary" and "secondary" color correction
as idioms used in industry.

It's a complicated definition to nail down, that's a given.  As I've stated
before, it is my opinion that the definition of these terms, today, has very
little to do with signal-flow or processing as much as usage within the
context of the results sought.  In this fashion the definitions might go
something like this:

What is done in preparation for fine (selective) image manipulation.
In Dailies, it refers to the utilization of basic scanner and color
corrector controls to achieve a viewable image with basic qualities
indicative of what was photographed and without undue distortions.

Fine image manipulation.  This may include selective color, texture, focus,
keying and other enhancements.
In Dailies, secondary color correction is never used.  This phase in image
manipulation is generally reserved for post-dailes processing in the
post-production chain, for example:  final color timing of weekly-episodic
shows, motion-picture mastering for video release, music video final color
styling, etc.

Maybe the most accurate time-invariant definition is that "secondary color
correction" is what you pay extra for.

Martin Euredjian
eCinema Systems, Inc.
(661) 305-9320
ecinema at pacbell.net

More information about the Tig mailing list