The film industry has a tremendous need right now for an open standard for communicating color grading information—a Universal Color Metadata format.
There are those who are attempting to standardize a "CDL" (Color Decision List) format, but it would communicate only one primary color correction. There are those trying to standardize 3D LUT formats, but LUTs cannot communicate masked corrections that are the stock in trade of colorists everywhere. There are those tackling color management, but that's a different problem entirely.
Look at the core color grading features of Autodesk Lustre, Assimilate Scratch, Apple Color, and just about any other color grading system. You'll see that they are nearly identical:
• Primary color correction using lift, gamma, gain, and saturation
• RGB curves
• Hue/Sat/Lum curves
• Some number of layered "secondary" corrections that can be masked using simple shapes, spines, and/or an HSL key
Every movie, TV show, and commercial you've ever seen has been corrected with those simple controls (often even fewer, since the popular Da Vinci systems do not offer spline masks). It's safe to say that the industry has decided that this configuration of color control is ideal for the task at hand. While each manufacturer's system has its nuances, unique features, and UI, they all agree on a basic toolset.
And yet there is no standardized way of communicating color grades between these systems.
This sucks, and we need someone to step in and make it not suck. Autodesk, Apple, Assimilate, Iridas; this means you. One of you needs to step up and publish an open standard for communicating and executing the type of color correction that is now standard on all motion media. This standard needs to come from an industry leader, someone with some weight in the industry and a healthy install base. And the others need to support this standard fully.
Currently the film industry is working in a quite stupid way when it comes to the color grading process, especially with regards to visual effects. An effects artist creates a shot, possibly with some rough idea of what color correction the director has in mind for a scene, but often with none. Then the shot is filmed out and approved. Only once it is approved is it then sent to the DI facility, where a colorist proceeds to grade it, possibly making it look entirely unlike anything the effects artist ever imagined.
Certainly it is the effects artist's job to deliver a robust shot that can withstand a variety of color corrections, but working blind like this doesn't benefit anyone. The artist may labor to create a subtle effect in the shadows only to have the final shot get crushed into such a contrasty look that the entire shadow range is now black.
But imagine if early DI work on the sequence had begun sometime during the effects work on this shot. As the DI progresses, a tiny little file containing the color grade for this shot could be published by the DI house. The effects artist would update to the latest grade and instantly see how the shot might look. As work-in-progress versions of the shot are sent from the effects house to the production for review, they would be reviewed under this current color correction. As the colorist responded to the new shots, the updated grade information would be re-published an immediately available to all parties.
Result? The effects artist is no longer working blind. The director and studio get to approve shots as they will actually look in the movie rather than in a vacuum. Everyone gets their work done faster and the results look better. All of this informed by a direct line of communication between the person who originally created the images (the cinematographer) and the person who masters them (the colorist).
Oh man, it would be so great.
I've worked on movies where the DI so radically altered the look of our effects work that I wound up flying to the DI house days before our deadline to scribble down notes about which aspects of which shots should be tweaked to survive the aggressive new look. I've worked on movies that have been graded three times—once as dailies were transfered for the edit, once in HD for a temp screening, and again for the final DI. Please trust me when I say that the current situation is broken. We need an industry leader to step in and save us from our own stupidity.
And this industry leader should do so with their kimono open wide. Opening up a standard will involve giving away some of your secret sauce. Maybe there's something about your system that you think is special, or proprietary. Some order of operations that you feel gives you an advantage. Well, you could "advantage" yourself right into obscurity if your competition beats you to the punch and creates an open standard that everyone else adopts. The company that creates the standard that gets adopted will have a huge commercial advantage. You can learn about the business advantages of "radical transparency" from much more qualified people than myself.
Of course, there will be challenges. Although each grading system has nearly identical features, they probably all implement them differently. It's not obvious how much information should be bundled with a grading metadata file. Should an input LUT be included? A preview LUT? Should transformations be included? Animated parameters? It will take some effort to figure all that out.
But the company that does it will have built the better mousetrap, and they'd better be prepared for the film industry to beat a path to their door. So who's it going to be?
Until you step up, we will keep trudging along, making movies incorrectly and growing prematurely gray because of color.
Thursday, January 3, 2008 at 6:20PM 