How cinematic color grading is actually done
There seems to be an interest, as well as a bit of confusion, as to how color grading is done for cinema (which includes video). I’ve read the various threads on MM and, while there is some great information, there is unfortunately a lack of understanding as to how the process works on larger budget film. This is exacerbated by the plug-in manufactures that provide color tools that do not provide the kind of power necessary to really do the work—Magic Bullet and Colorista come to mind, as they were referenced in the MM forum before. There may be new plug-ins that do what I’m going to describe, but since I don’t use them, I’m only going to comment on what I saw posted on this site.
Don’t get me wrong, these new plug-ins have some nice presets. However, I’m pretty sure that once the actual process is explained, some individuals will find that Photoshop is better for smaller jobs, and that Da Vinci’s Resolve (or Apple Color, which has apparently been discontinued) is preferable for longer works.
Color correction vs. color grading
The reason those aforementioned plug-ins don’t work all that well is because they only offer global changes, and that is only part of the process. In order to understand the entire process, we first need to understand the difference between color correction and color grading. First off, color correction refers to normalizing footage for color balance and setup.
When shooting digitally, color balance isn’t such a big issue because we can generally do a custom white balance prior for each scene—but that wasn’t always the case. With film, for example, you only had daylight-balanced film and tungsten-balanced film. Even if you were shooting daylight-balanced film in daylight, the actual color temps could still fluctuate quite a bit depending on the intensity of the light (just as with still photography). Shooting tungsten didn’t have as much variation, but even different batches of film could have slight differences in color temp. Because of this, we needed to constantly correct all the footage so that we had a consistent white point from which to begin grading. Also, we wanted to normalize our levels: we refer to levels as lift/gamma/gain—lift is our black point, gamma our mid-point and gain our white point. Again, we’re not trying to be creative here, just trying to correct the footage for consistency. These changes are global in nature.
Color grading is where we make creative changes to create a “look” for the film or a particular scene within it. These changes are localized. Because of this, we use several different “rooms” to accomplish any color alteration (today these “rooms” are just tabs in a program, but there was a time when they were actual, physical rooms).
So, let’s say you want to do that teal-in-the-shadow or cream-pink-in-the-highlights look. What I’ve seen discussed on the threads is to simply pull your shadows towards teal or your respective highlights toward pink until you create a balance; however, these are global changes and not how we would really go about doing this.
Instead, we would separate those actions into two of our secondary rooms. The first would use an HSL (hue, saturation, luminance) mask to isolate the shadow tones, and we would grade those tones toward teal, without affecting the rest of the image. Then we would open another room and select the highlight tones we wanted to alter and adjust those toward pink. Next, we would open up a third room to handle our mid-tones or maybe only our skin tones.
After all that is done, we might open up a few more rooms to make changes to certain specific elements, such as the sky, an open doorway, or a person in shadow. We might use ten or more of these secondary rooms to get the look we want. Better systems have the ability to work with nodes, which allows for very complex selections and actions.
After all of our secondary work is done, we go to our “Primary Out” and make necessary tweaks to the overall look. As with the “Primary In,” these changes are global and corrective.
Plug-ins that only allow for global changes are fine for insta-looks, but they lack the necessary power to really dig into an image color-wise.