All Things Color for Film and Digital Cinema
Subscribe | Log in

Densitometry for Lookup Tables Pt.1

A densitometer is a device that measures the optical density of an image on any printed medium, be that celluloid print film, paper, photographic film negative or any other medium. If the image is printed on paper we could read its optical reflectance, if printed on 35mm print film we would be reading its optical transmission qualities.

As human vision is measured in logarithmic values so too are the results of the densitometer. When taking readings of a Motion Picture print over a densitomter we are reading the prints transmissive qualities of light. In Status A densitometry (more on that in the next post) the lighter the patch the more it transmits light. The denser the image the less transmission of light. Density is defined as  

Optical Density = log10(1/transmittance)

I have a X-Rite310 densitometer which is common to most Film Labs around the world. It reads, for any given reference, the Red, Green and Blue optical transmission values which is imperative in creating RGB Color profiles or 3D lookup tables (3D LUT’s) for film.

Fig:A a 4x4x4 example target

Fig:A 4x4x4 example unity target or Identity LUT

 

Printing a set of specific color and grey scale patches to the desired medium, like the ones in figure A, an operator can read and translate the optical transmission characteristics of each individual patch and create a ‘lookup table’ . This profile can then be used to visualize on a monitor what the image will look like when printed to film. It is also profiling the characteristics of the imaging device which recorded the patch to film. In our case an Arri Laser film recorder. The more colors sampled the more accurate the representation will be. (The example used is extremely ‘low res’ and is only used for demonstration purposes.) To achieve an accurate representation we tend to use a ‘matrix’ of 4096 color patches!

 Lets look at those color patches from Fig:A mapped into an RGB cube to see just how these work. (See Fig:B)

Fig:B Representation of the 4x4x4 Color Cube

Fig:B Representation of the 4x4x4 Color Cube

 

If you compare the image in Fig:A and the Color Lattice to the left you can see each of the 64 patches are represented in a 3 dimensional cube which we can call the RGB Color Cube. The extreme corners represent %100 code values for all the primary colors of the cube, and Black. If you were to only look at the points down the diagonal axis from Black to White you would be looking at a grey scale.   From this would could create an identity LUT which would read something like Fig:C in ASCII text. Being an identity LUt it would have no optical effect to the original image.

 

 

 

ASCII 3DLut 4x4x4 identity matrix

ASCII LUT

         Starting at Black and reading the 4 points across to red expressing them as RGB Triplets. (I am assuming a an 8 bit scale which doesnt divide very well in 3 from 256 code values so excuse the rounded math)

 

        We then move up one row towards Green and read out the next row of 4 points.

 

 

         Until we read the top at which point we move in one row towards Blue and then begin reading the rows on that plane in the same way. Eventually after 64 entries of RGB triplets we have successfully mapped the cube.

 

 

 

In the photographic desktop publishing world images like the one in FIg.D are used. These are called Targets. The most used one is the Kodak IT8 Target.

kodak_IT8_target

Fig:D Kodak IT8 Target

 

 

 

 

 

 

 

Fig:B 4x4x4 identity LUT

Fig:E 4x4x4 identity LUT

Due to the different optical characteristics of different papers, films and imaging devices the original image will never look the same when printed, unless, through software we mimic those qualities on the monitor we are referencing the digital image on.

When printed to the desired medium and each of the patches are measured on a densitometer we remake the cube and will notice that the points are slightly deformed taking on and showing us the optical properties of the medium.We then transpose those results to represent the films qualities in to a new ASCii LUT

Fig:C transformed LUT

Fig:F transformed LUT

 

We then transpose those results to represent the films optical qualities in to a new ASCii LUT that can be fed into the monitoring device.

 

 

What I have just described is the most rudimentary basics of color management and 3D color Lookup’s. The Following Posts will go deeper into real world color calibration also touching on, using Marcie,monitor characteristics, viewing environments, film projection characteristics, different recording and printing stocks.

 

Adrian.

 

Image References:

Nvidia.com – Graphics GPU technology examples

RPIImaging.com  - IT8 printing Targets.

cube-lattice

One Response to “Densitometry for Lookup Tables Pt.1”

  1. natacha347 says:

    This post is just what I consider an easy to follow posting. Very clear and to the point. I will have to certainly watch out for more content like this.

Leave a Reply