color science

ARRI LogC | Kodak 5219/7219

August 19th, 2019

Arri Alexa SXT shot ProRes 4444 XQ at 3.2K transformed with color science to display the characteristics of Kodak Vision 3 5219, arriscan.

Arri Alexa SXT shot ProRes 4444 XQ at 3.2K transformed with color science to display the characteristics of Kodak Vision 3 5219, arriscan.


While film and digital image acquisition devices do not capture image information equally, carefully acquired digital footage can be processed to display similar characteristics of motion picture film, with a couple of limitations. 

In the following article I will be presenting some color science I have developed for transforming digitally acquired imagery from various ARRI Cameras to display, as close as possible, the characteristics of Kodak Vision 3, 5219/7219.


Below is a film showing side-by-side the ARRI Alexa Mini next to Kodak Vision 3, 7219 - shot in controlled lighting conditions.
Super 16mm was chosen as I own the equipment, but it should serve the purpose of the demo well, through the more visually pronounced filmic flaws. The film stock's response is the same as 5219, and thus scaling of the spacial attributes is the only difference between the color models for 5219 and 7219. Click for a side-by-side framegrab gallery.


Here's the same film with the same color science, but Super 35mm sized qualities; grain, halation and resolution. ARRI Alexa Mini only.
Here's a framegrab gallery for this section.


Now, while these examples may to some degree demonstrate that the differences between film and digital acquisition are negligible with the correct data manipulation, I will note that this is only partially true, especially for shooting circumstances that are not as controlled as this demo. The limits are defined by the technical differences in the ways the two systems capture image information.

Besides speed, grain, rolling shutter, gate weave, moiré and resolution, these are the strongest differences (and biggest issues for data manipulation) in the way film and digital capture devices record and image, or see a scene

Dynamic Range
- Including inertia point(or speed point as kodak calls it), and highlight color reproduction.
(Test: Can you program clipping to take the color of the nearest non-clipping pixel (pre-gamut remapping) + X math to make it brighter)

Translation of Color
- Including captured spectrum/spectral distribution curve, captured gamut, natural color mapping and effects of white balance.

- Including edge contrast (and flaring)

Needless to say, making a 1:1 transformation is not possible if image information is captured so differently that no data set can control it.
This is the biggest issue for color science and can only be addressed with creative color grading, cinematography or by changing or updating current camera technology. 

More details...
- remember white balance on a colored item. E.g. warm and cool light, or magenta and green light on a red ball.
- Remember testing grain in all zones and of all (or many) colors in different zones. The grain may be one of the hardest things.

Other challenges that can make data transformation difficult (test all these before making a statement):

- Washing out colors
- Colored flares e.g. anamorphic flares
- (different way of seeing flares coming in from the side?)

Translucent or transparent objects (NEEDS TESTING)
- Objects in front of the lens, out of focus
- Glass e.g. windows
- See through materials
- Water e.g. the ocean, water splashes, rain
- Smoke, atmospheric haze, blizzard


Whatever the final findings are. Try to keep it open and informal.

Technical specs & notes

Clips in 7219 film aren't graded, except for RGB balance etc.
Use of 85