|
|
 |
 |
 |
 |
 |
How to set up and calibrate a digital camera properly
Accurate, consistent camera alignment is an objective process and advancements in measurement technologies, from precision test charts to vectorscopes and waveform monitors now make this a more easily achievable goal. Ensuring that test equipment is accurate is essential. Every item in the system plays its role from the camera, to the test pattern, to the measuring equipment and picture monitor. While any item in the production train can throw picture quality off track, the test chart is likely to be biggest culprit and waster of engineers efforts.
Camera Calibration Charts
An accurate test pattern provides essential information that is reflective of what the camera sees. An accurate test chart does not negate, but complements an electronic signal generator by including the camera's taking characteristics. By "taking characteristics" I mean the many elements in addition to a camera's electronic circuitry. These include the lens and adapters, the prism block, dichroic and trim filters and the color characteristics of the CCD or other chip set. Taking characteristics will also be affected by changes in the color temperature and spectral distribution of the scene lighting. Accurate evaluation of these characteristics requires two essential elements, a precision test pattern and accurate electronic measuring equipment.
Be sure to use quality test charts and patterns? when aligning a camera. Originally designed for engineering, they are now also used in production as on-the-set references for use in post and special effects. They do this by not only providing signals that fall within their prescribed vectorscope boxes, but also at prescribed luminance levels. This enables accurate setting of a color’s brightness in addition to hue and saturation levels. Individual Vectorscope and Waveform calibration sheets are available showing the virtually infinitesimal differences between similar test patterns. These are available mV and/or IRE levels.
Providing accurate color and grayscale values at the correct luminance is particularly helpful when setting up a camera to a particular colorspace. When aligning a camera to a specitic color char, be aware of the differences in Standard ITU-R BT.709, EBU and SMPTE C. For the purposes of this experiment, we use a preproduction camera set up, an on-set single camera shoot and an on-set multi-camera shoot.
Camera Color Reproduction
Typically the aim is to reproduce the image on a monitor or television set as closely as possible to the original scene. To achieve this, the camera converts data from an accurate test chart via a camera's CCD chip set to digital voltage levels. The characteristics of the grayscale and color steps can then be adjusted using the camera's set up controls and color matrices. Optimizing the image signals largely depends on the adjustment capability of the individual make and model of camera. In addition to an accurate test chart a precision waveform/vectorscope/spectrum analyzer is vital to this operation.
 IMAGE LEFT / IMAGE RIGHT
It is interesting to note that while the industry has been concerned about the relatively small colorimetry differences between television systems, (above left), until recently little has been said about the large differences between various makes of monitors and TV sets (above right).
Grayscale adjustment
While it is a relatively simple procedure to set up correct exposure levels using a grayscale chart, failure to use a truly neutral grayscale can result in serious tracking errors. Typical grayscale charts are more yellow in the light steps, approaching neutral towards black. Use of such charts produces images that are artificially blue or cold-looking in the lighter tones. This is the last thing you need in an image, because viewers prefer warm looking pictures. In using a grayscale test pattern, both dynamic range and progression between steps become important. In the early days of television, the dynamic range of a camera was limited to about 25:1 and grayscale patterns had a correspondingly low dynamic range. Unfortunately such 9 step test charts are still being sold to align modern cameras having a dynamic range of 3000:1 or higher.
The progression rate between typical 9 step charts and a modern 11 step pattern is also different. The 9 step is log in terms of reflectivity while the 11 step is designed to produce steps that are linear in voltage.
The lightest step on a 9 step chart reflects 60% of the light falling on it compared to 90% from the white chip of an 11 step chart. Note how the 11 step pattern has greater dynamic range; the white chip being lighter and the darkest chip being darker than the 9 step. The different densitometric curves between test patterns result in significantly different image reproduction from cameras aligned to a nine or 11 step chart. Aligning to inaccurate grayscales of limited dynamic range can result in poor quality images that are virtually beyond redemption.
While every engineer and DP has his/her own technique for aligning a camera, they will typically use the following procedure:
- With a camera focused and framed on an evenly lit calibration chart? the Iris is set to mid scale (typically between f4 and f8).
- Then the exposure/iris is adjusted to set the white chip of the DSC grayscale to 700mv (100 IRE). Tracking of RGB channels should also be checked to ensure neutral reproduction across the grayscale.
- White and black balances are then set. The true black chip of the grayscale is normally set close to 0 mV - Note: NTSC in North America uses a black level set up of 7.5 IRE above true black. The Gamma or crossover step of the 11 step chart is typically adjusted to produce a straight line; this results in accurate grayscale reproduction.
It is important to note that while a camera may reproduce a grayscale perfectly, it says nothing about its color reproduction which could still be very poor.
Color adjustment using a Vectorscope
When using quality color calibration charts, the saturations are reflecting those found in real life. Consequently when a vectorscope is adjusted for 75% electronic color bars, color signals will fall short of the boxes. However, a simple 2x increase in vectorscope gain will place all primary color signals in their boxes when a camera is reproducing colors accurately. With the advanced multi-matrix settings of modern cameras, it is possible to line up each primary in its box, but inadvertently reduce the overall color gamut.
When aligning a camera's matrix, increasing or decreasing the sensitivity of a particular color can affect the positioning of many other colors – consequently the process requires considerable patience. If patterns were made with different luminance relationships, color matrix adjustment would displace the gamut showing a false primary as true, further distorting the working colorspace. This is exemplified when adjusting a camera's color matrix settings. Typically an individual color cannot be adjusted without affecting others. Adjusting B-Y, R-B, etc. settings tends to move all colors on the vectorscope, some more than others, and it takes both skillful "tweaking" and patience to optimize a camera for accurate reproduction. A fact that should always be taken into consideration is that if one or more colors in the chart are incorrect in any parameter, Hue, Saturation or Luminance, accurate reproduction becomes mission impossible. For this reason test charts should be kept current and replaced in a timely manner. The bottom line is; aligning to an inaccurate grayscale or color chart is worse than using factory presets – much like the common error of white balancing to a sheet of ordinary white paper.
Camera Matching
Camera matching becomes impossible without an accurate test pattern and analytical tools. No two cameras are identical which makes it unreliable to upload the settings from one camera to another. The only way to match cameras effectively is to adjust the matrix settings of each camera using the same precision test chart under the identical lighting condition. Set exposure levels and white balance then align the colors on a vectorscope, finally match the second and additional cameras to a freeze frame of that pattern.
In this example, accessing the Vectorscope buffer through the camera will take a snap shot of the current configuration and create a pale brown overlay when going back into the vector display mode. With the first camera still connected to the vectorscope, the second camera can view the same chart, and be effectively matched to the first camera's image. Ensure that the camera stays connected to the scope, as the wave form monitor does not have an internal signal generator, and will not be able to reload the images from the USB Buffer without computing the format of the original capture.
Occasionally a DP may want to divert from accurate reproduction for a special "look" or effect. However, by first aligning to a calibration test pattern, the working colorspace becomes a wide and accurate baseline from which prescribed deviations may be dialed in to produce the desired "look".
|
|
 |
|
 |
|
|
|
|
