The award-winning definitive authority on all things visual effects in the world of film, TV, gaming, virtual reality, commercials, theme parks, and other new media.
Winner of three prestigious Folio Awards for excellence in publishing.
By JOSEPH GOLDSTONE
Edited for this publication by Jeffrey A. Okun, VES
Abstracted from The VES Handbook of Visual Effects – 3rd Edition
Edited by Jeffrey A. Okun, VES and Susan Zwerman, VES
Any visual effects professional whose personal contribution is judged on color should read the three guidelines below. Anyone building a workflow from scratch or integrating existing workflows should read (or at least skim) this entire section.
The Three Guidelines
1. Make critical color decisions in the screening room, not in an artist’s cubicle.
2. At the desktop, use an open-source or commercial color management system if possible, but do not expect too much.
3. Understand and document the color image encodings produced by a production’s digital motion picture (DMP) cameras, renderers, film scanners and the encodings consumed by displays, film recorders and digital projectors that deliver production output to clients.
Turning to camera raw formats for digital still and motion picture cameras, Figure 6.10 shows those color image encodings are quite different from that of Adobe RGB.
Like Adobe RGB, the camera raw colors represent physically measurable colors. Unlike Adobe RGB, there is no mandated breakdown into particular image channels; some raw formats used by single-sensor cameras, for example, provide for two green channels if the camera sensor contains twice the number of green-filtered pixels as it does red-filtered and blue-filtered ones. The encoded colors leaving the camera are similarly unconstrained, but typically the encoding provides for three or four channels and the bit depth (not necessarily the same for all channels) ranges between 10 and 16 bits.
With all this uncertainty, it is reasonable to ask why anyone would want camera raw encodings. The answer is that the image data are not color rendered; in color management terms, they are scene referred. If the vendor provides some post-processing tool that converts the image data to a more standard RGB encoding without losing the raw nature of the image data (that is, without color rendering it for some display device), then this scene-referred data may play very well with CGI renderers and compositing applications. For digital still cameras, Adobe’s DNG Converter and Dave Coffin’s open-source dcraw program12 can be used to perform this type of extraction. For digital motion picture cameras, dedicated tools for this type of extraction (e.g., RED’s REDCINE-X PRO, or ARRI’s ARRIRAWConverter.app) are usually freely downloadable from the vendor; and some free versions of color correction software (notably Blackmagic Design’s Resolve Lite) support several digital cinema camera raw formats.
With raw encodings, color rendering will need to be done downstream before the completed shot is shown in a theater. Color rendering is more easily done by the visual effects team, however, than it can be undone if the camera manufacturer has baked-in color rendering in a non-raw format. The color image encoding for camera raw implicitly has as its viewing environment the original scene itself.
Color Management at the Desktop
With the above as background, the responsibilities of any color management system are thus:
• Interpret image data according to the image’s associated color image encoding.
• Provide a path to and from some common color space in which the correctly interpreted captured image data can be combined (a “working space”).
• Display original or combined image data as an accurate reproduction of what the final consumer (e.g., moviegoer) will see.
(Editor’s Note: The version of the article printed here is a brief overview of the in-depth version in the book. Various sections have been kludged together to give the reader an idea of what they can find in the book.)