In Part One we looked at some of the important exposure settings in a video camera. In this part we’ll look at some more, and also the settings and adjustments that pertain directly to color.
The Response Curve
Whenever you buy ‘gamma’, you get a free graph like this:
There are other names for this graph, but I’m going to use the generic term ‘response curve’. Light hits the sensor, then it’s all a blur, and you finally get an image. In between you have a few knobs to turn around. The response curve basically says: If you change the light in so-and-so fashion using the knobs at your disposal, I’m going to do this to your image. We already saw that gamma is one of the major goons in this jungle. Now let’s meet the little puppies.
The chart includes several important ‘regions’ and ‘points of interest’. Understanding these is not hard, but it takes practice to learn how each specific camera is designed to handle light and color and then apply the knowledge for practical use.
The two regions we are concerned about are:
- Toe or pedestal – the black/shadow side
- Shoulder or Knee – the white/highlight side
Pedestal, Black level or Master Black
The video signal cannot go below 0 IRE (which is pure black). The pedestal (or base) refers to the setting that changes how the black portion (toe-end) behaves. If you ‘raise the pedestal’ (positive number), the shadows lift. If you lower the pedestal (negative number), you will crush the blacks.
The proprietary algorithm of each video camera tries to respect your wishes without touching the mid tones and highlights, though the degree to which these are affected is totally dependent on high up or down you push it.
Newbies often mistake the flat-looking image you get by raising the black level to mean an increase in dynamic range. This is not correct. Once the gamma and color space are set, the dynamic range is pretty much etched in stone. The black level is akin to crushing the blacks or raising the shadows in a color grading application.
If you want to picture it, imagine moving the black dot up the vertical axis (raising the shadows) or down the vertical axis (negative zone, crushing the blacks). Nothing else moves.
Instead of raising or lowering the black levels, one could also adjust the curve only for the toe-region (Gamma over a gamma). It tends to look like this:
This keeps the absolute black level constant, but changes how it ‘rolls-off’.
Knee is opposite to pedestal. It is used to get the highlight response you want.
The setting for knee is either:
- On – you want it
- Off – you don’t want it
- Auto – let the camera decide based on a preset
Typically, cameras offer more options for knee than pedestal. I’m not sure of the exact reason, but my guess is that the dynamic range of a camera is limited, and the most important element in a shot is the skin tone. Once a camera operator exposes ‘correctly’ for skin tones, then the blacks and whites are allowed to fall where they may (as shown in Part One). Our eyes are used to objects crushing to black when the dynamic range of a scene goes too high, but we don’t see blown out highlights naturally. This could be why, along with the fact that digital sensors have poor highlight roll-offs, that the knee setting is given more control options.
The first important setting for knee is the knee point. It is the point beyond which the highlights are affected. Sometimes cameras have their own scale while on others it’s the IRE rating or a percentage.
This is the ‘white gamma’ (Slope is gamma) that allows you to control how the highlights roll off. You could make it ‘drastic’ or ‘smooth’ but obviously there is no one correct setting, and this must be adjusted by eye (or waveform monitor) on location for every shot.
The problem with knee in general is that you need either of these two conditions to exist before it can be useful:
- The camera has good highlight dynamic range.
- You expose for the shadows (underexpose) to keep more highlight detail for knee to work.
Either way, without something to play with, the knee setting will hardly matter. Like the black level, it does not increase dynamic range by itself, but only gives us the impression of doing so.
White level or Clip level or White clip
This is an absolute setting that tells the cameras – nothing beyond this point! Anything that crosses this threshold is clipped. I have found this setting to be useless for artistic work, though its most important use is for ensuring highlights stay within the Rec. 709 or TV broadcast specifications (100 IRE for full swing, 92% for studio swing).
Instead of the white level or clip, I prefer the next setting.
Zebra (1, 2)
Zebra is just an indicator (looks like zebra stripes) on the camera monitor/LCD screen/viewfinder that highlights which regions are above a certain threshold (in IRE). E.g., if you set the Zebra at 90% or 90 IRE, any regions in your scene that go beyond this point are highlighted with ugly-looking but absolutely useful zebra stripes. They are not recorded in-camera (you won’t believe that there are professionals who think it does!!) nor are they passed on via HDMI or SDI.
The main zebra (always displayed in stripes), is Zebra 1. Some cameras allow you a second Zebra for additional display (spots instead of stripes). Why on earth would anyone want a second zebra when the first one is annoying as it is? Sometimes, ‘professionals’ like to expose via waveform/IRE rather than by eye. One tradition is to set the Zebra 1 at 70 IRE, because that is where the ‘limits’ of Caucasian skin tones lie, and use Zebra 2 for 90 IRE or 95 IRE to keep highlights from blowing out.
I was dumb enough to follow this advice before knowing that the IRE level of 70 was intended for Caucasian skin, while my skin is Indian (it’s a new color, I’m going to patent it). What do you do in a scene with non-Caucasian skin-types?
But there’s also a problem with holding on to 70 IRE for Caucasian skin. There’s a large range of tones that fall under the ‘Caucasian’ paradigm. It changes on whether you’re in the sun or shade. It changes if you’re dead or blushing or tanned or have makeup on. It changes if you turn into the hulk.
For these reasons and more, I have long abandoned placing skin tones at IRE levels, and recommend you use your eyes instead. Also, since skin tones are no longer an ‘IRE’ problem, I restrict myself to Zebra 1 to check for blown out highlights and that’s it. If you’re not delivering to broadcast legal limits at all, then you really don’t even need the Zebra.
In a war between zebra and the histogram, whom should I believe? In other words, if the zebra clips but the histogram doesn’t, what should I do? Here are some differences:
- Zebra is calibrated to IRE, which is normalized luma (Y’ in Y’CbCr) and only looks at the luma information as a whole.
- Histogram is calibrated to RGB. If your camera does not offer separate RGB histograms, it is impossible to tell (without scientific testing) whether the histogram clips because of one channel or both or all.
Remember, the zebra is calibrated to Rec. 709 specifications, while the histogram is calibrated to the sensor RAW feed (you can call it the sensor’s gamut). If you’re shooting RAW or log, then screw Rec. 709 (even if you’re delivering to broadcast).
Bottom line, you only need to use one. The type of project you do will dictate which one you’ll pick.
So much for the exposure settings. Let’s add a little bit of color.
Hue and Saturation
Hue changes the overall hue of the scene (like adding a tint or wearing colored glasses). Saturation raises or lowers the saturation of individual colors. Sometimes, you have saturation controls for the toe and shoulder regions of the response curve individually.
Sometimes, you can change the saturation using picture profiles or presets as well, and the use of it is entirely subjective.
As far as hue is concerned, it’s a rare setting, because you can change the hue better and more accurately with white balance and the color matrix.
The color matrix is an overall LUT-like matrix (LUTs are matrices as well, I’m just playing with words here) that transforms the colors on your image. Presets, picture profiles, scene files, creative styles, etc. are all matrices that affect the colors in an image in a ‘specified pre-determined’ way. It’s baked-in, more or less.
On professional and prosumer video cameras, you can get finer control with the color matrix setting, which typically offers these adjustments: R-G, R-B, G-R, G-B, B-R, B-G. What should you note? The fact that these are not R, G and B separately, but in pairs. This means, by changing one channel, you also change another channel. How do they relate? Here’s a cool rule of thumb I learnt from Andy Shipsides of Abelcine:
When adjusting any one of these combinations, the first color (R, G or B) will be affected in saturation and the second color (R, G or B) will be affected in both hue and saturation.
Color depth (sometimes ‘matrix’ or another name)
Sometimes you can change individual color channels as well. Usually they are:
- CMY (Cyan, Magenta, Yellow)
The combination of hue and saturation is called phase. In the rule of thumb, we have seen that the first color will change saturation while the second will change both hue and saturation. This together is called the phase.
One thing you’ll notice when you play around with these settings is that they are often very subtle, and don’t allow you to go ‘all out’ like color grading applications do. Camera manufacturers are mindful that most operators will be using these features in tandem with a waveform (for exposure) and vectorscope (for color) while on the field, under pressure.
Therefore, the key goal is to get the shot in a limited amount of time. Settings that give you greater control also will suck time while you get there. It’s a fine line, and most of these settings are more intended to ‘correct’ a shot than ‘creatively color grade’ it. There’s more than one way to crunch the math, and tweaking a setting for one camera will not yield an equivalent corresponding result in another.
The important takeaway is that there is no one correct value or parameter for any of these settings. If there were, the manufacturer wouldn’t give you the option to change it. Think of it as tuning a guitar. Some do it with generic equipment, others do with with specialized equipment, and yet others do it by ear. The goal is to deliver an acceptable Rec. 709 image. Within this playpen, anything goes.
I hope this simple primer will help you deal with these settings in your own video cameras. You will also find many proprietary names and controls, so this is far from a definitive list of any sort.