Most NLE software has blessed the editor with advanced color correction tools. Both a blessing and a curse, color correction is best used in small doses. For proper color correction, high quality monitoring (video and scopes) is essential. Even experienced editors/colorists are leery of their corrections when they have no calibrated standard to judge their work by. So if you intend to make the most of these tools, you’ll have to invest in decent hardware scopes and accurate monitors. Software scopes can work in a pinch, but most I’ve seen are awful and give you little if any valuable information.
Types of correction
Film always made heavy use of post-production color correction (film timing or color correction during film transfer). Early video production in the studio or in the field tended to rely on trained video engineers (aka camera shaders) to “paint” the camera’s look for the best creative results. As video production has taken a cheaper run-and-gun approach in more recent times, color correction has increasingly become a post-production video function. Fortunately this has coincided with the increased availability of affordable tools placed in the hands of editors and colorists.
Most color correction tools include proc controls (video, gamma, setup, saturation and hue), primary correction (RGB balance) and secondary correction (adjusting the saturation and hue of only a segment of the color spectrum). Many NLEs also include filters for legal broadcast color limiting and monochrome effects, like black-and-white or sepia. Judicious use of these tools will generally improve your video.
To my eye, most of the DV and even HD cameras generate a somewhat neutral image that begs to be color corrected to gain back that certain visual “snap”. This is analogous to the film world’s one-light transfer. It may look good but doesn’t really bring out everything there is in the image. Proper correction can give your video a definite visual style – if you know what you’re doing.
Warm and cool
Most of the time when you are correcting an image, you aren’t trying to produce a look that is faithful to the natural environment, but rather one that the viewer perceives as more pleasing. We tend to like images that have more contrast and are more saturated than what our eyes would really see. We also like an image to have a little more red than is truly natural. Images that are red to golden in hue are considered “warm”, while images with a bluish cast are considered “cool”. Each evokes a certain psychological response.
Often when I correct an image, I will increase the contrast to make sure there is a definite black and definite white object or area of the picture. Obviously you can’t create what wasn’t shot, but giving the home TV set these high and low parameters to lock on will result in an image that translates best from your suite to over-the-air broadcast. If the image is “flat” with peak levels below 60 IRE, you will see on image at home that looks washed out. If you increase the contrast, certain levels at the bottom will become “crushed”. That may be desirable but you will lose visual information in the dark areas of the screen. When you do this, be mindful of the color saturation. Lowering the darker values results in a more saturated-looking image, especially in reds and blues. To compensate, reduce your chroma saturation. Since the eye likes a bit more red, I will frequently adjust hue offsets or curves to increase the amount of red only in the midrange areas – or shift the balance to a little more red. This tends to add just a touch more red into flesh tones and gives you a “warmer” look. Be careful though that reds don’t become over-saturated or tint to the green or purple.
Midranges and gamma
Any color correction with control of midranges and gamma values gives you a powerful tool at your fingertips. Most of the visual information you want is in the middle portion of the lighting values of your video. If you have a person’s face in front of a lamp or window, you often have midrange values (the face) that are darker than desired, because the lighting or camera settings didn’t properly compensate for the bright lamp or window. Increasing the brightness of the midrange shifts these value up (as seen on a waveform) and makes the image of the face brighter and hopefully more pleasing on the screen. You have to be careful when you do this, because you are effectively compressing the range of values between the mids and the highlights. This will result in the highlights on the cheeks of a face to become too flat, creating a cartoon-like effect and leaving these highlights pasty and unnatural. Since color correction is always a dance between luma and chroma values, you can often minimize this effect by reducing the midrange or highlight chroma saturation.
Changing the gamma values changes the point where an image crosses from dark to light. Adjusting gamma up or down alters the linearity of these values as viewed on a waveform. Raising gamma increases the visibility of information in an image, especially in the darker areas. Increasing gamma values improves shadow detail, but also increases the amount of video noise. This can be a pleasing effect, often mimicking the appearance of film grain and texture, but only if used in the least amount. Extreme gamma increases can be helpful if you’re going for that high-gain, “security camera” look.
Most cameras are set with peak video levels that exceed the ideal “legal” value of 100 IRE. Their maximum values are around 105 to 110, with chroma values in bright colors like yellow at around 115 to 120. This is done on purpose, so that you get more midrange information, but the intent is that the excess peak level is going to be clipped off by some type of hardware limiter. Since most NLE don’t use a standard hardware proc amp to keep the video legal, you will have to correct for this or add a “broadcast safe” limiting filter. If the color correction tools offer separate control of highlights, you can often lower the video gain and/or saturation of only the highlights. This might give you far better results than “shearing off” the top values of the image.
In addition, if you have increased the midrange or overall master levels, reducing the highlights (level and/or chroma) will help to minimize the effects of the level compression that occurs between midrange and highlights within the image.
Sampling depth and codecs
Color correction results can only be as good as the video you begin with and the environment in which you work. If the footage was shot on film or Digital Betacam and you are working in a 10-bit uncompressed NLE, then you will have far better latitude than if you dump that same footage down to DV and work in that realm. If you start with DV footage you have even less to work with.
The same is true in HD. For instance, if I shoot with a Panasonic VariCam and post with its native codec inside Final Cut Pro HD (DVCProHD), I will quickly see color correction artifacts, such as banding in the image, which wouldn’t be there if I had loaded the same footage into an NLE that allowed me to work in uncompressed HD. This is especially true if I am adjusting midrange or gamma values, because I have to do this within an 8-bit scale, with only 256 steps that define the values of the image. If I were working in a 10-bit system with 1024 steps, color correction artifacts would only show up at much more extreme ends of the adjustment/correction range.
Lastly, when you color correct, take it slow. Get up and look around and then come back to it. Play through the sequence to see your correction in context. Give your eyes something else to look at. You want to avoid going farther and farther “out into left field” as your brain adjusts to more bizarre corrections, thinking that they look good. If you don’t adhere to these tips, you might look at your work the next day and wonder ‘what was I thinking!’
© 2004 Oliver Peters