Technology Comparison of Digital Systems
A.Bayer Matrix Sensor
Bayer Color Filter Matrix (1)
Almost all digital camera systems use a monochrome sensor covered by a Bayer color filter matrix with alternating rows of green-blue and red-green filters. Each photosite (pixel) of the sensor can only capture one primary color and discards about 2/3 of the incoming light.
The luminance perception of the human eye is more sensitive to green light, so the Bayer pattern includes 50% green-sensitivity pixels and 25% each for red and blue. The double number of green pixels improves the definition of fine detail and the brightness levels, and the red and blue contribute primarily to the color information.
Each digital camera system using the Bayer matrix performs a “demosaicing” process to translate this array of primary colors into a final image with full color information for each pixel. (A rare three-layer sensor by Foveon® (2) creates images where each layer of silicon acts as a color filter).
|Enlarged view of a small section of a painting.||Raw image from a Bayer sensor (schematically) before “demosaicing” processing.|
- Achievable Image Quality of the Bayer Sensor
The collected data from the Bayer-filtered sensor must now be calculated or developed in the demosaicing process to produce the color image. Since each pixel is only filtered for one primary color the other two colors are interpolated (or “guessed”) based on the surrounding pixels. Although most camera systems do an excellent job of representing the general image, it is likely that very fine details or color in the image may not be accurately reproduced.
Modern algorithms attempt to draw conclusions about the color in a non-linear procedure that accommodates typical and generic conditions, but struggle with situations that have unique or sensitive characteristics. For example, the diagonal arrangement of the green pixels allows for a good representation of horizontal and vertical structures, but diagonal lines and details at an angle can contribute to additional digital artifacts.
Often camera systems combine the Bayer-interpolation simultaneously with error correction (defective pixels and columns on the sensor). This is further complicated if color management and image enhancements, such as sharpening, contrast and color saturation are done at the same time.
Typical examples of image artifacts due to the Bayer-interpolation are shown with images of the Color Resolution Chart. The images were captured using one of the best medium-format camera systems under ideal conditions. Use the Color Resolution Chart to make your own comparisons and conclusions.
|The Bayer-interpolated result is influenced by color, contrast and structure of the image content|
The background on this test chart patch was medium green. In the image, the color is too vibrant and the color between the grid lines is desaturated. The lines and numbers appear black, but actually are a dark red on the test chart.
This example is of a grid of black lines and numbers on a more saturated green background of the target. Here the algorithm correctly interpreted the color between the lines, but the green is over saturated.
The achievable resolution is dependent on the contrast.
|Moiré Artifacts in the Image|
With very fine lines a Moiré pattern can appear. This is caused by the Bayer pattern grid interfering with the structure of the image. There are methods to suppress the moiré but they lead to a reduction of sharpness and resolution.
|Over-Saturation and Over-Sharpening|
In photography, bold colors are desirable for most applications. Camera algorithms are often biased to produce highly saturated color. In this color chart test the red channel was oversaturated and made it more difficult to achieve accurate color reproduction and maintain subtle color variations.
|Loss of Fine Detail with Low Contrast|
|The texture of the paper is not evident in the plain white area even with sharpening. Some structure will be seen in the color areas depending on how it was printed.|
B. Scanner with Trilinear Sensor
Principals of Operation of a Scanning Camera
With a scanning back, the image is created one pixel line at a time as a highly-optimized trilinear sensor is physically moved in a smooth, continuous motion across the entire imaging area of the camera. Within the image sensor, three rows of light-sensitive photodiodes are each covered by a red, green, or blue color filter, making the entire pixel row sensitive to only one primary color.
The three sets of pure color data are combined to create the final color digital image. No other internal processing, algorithms, or interpolation is necessary. This “raw scan” is often saved for archival purposes or additional processing steps can be done such as applying a color profile, adding sharpening, or other modifications to optimize the image for printing or digital access.
|Enlarged view of a small section of a painting.||Raw image from a Trilinear sensor. No additional processing is necessary.|
- Achievable Image Quality of a Trilinear Sensor
|The Scanning Back results are independent of color, contrast and structure of the image content|
|This image example shows the correct reproduction of the color and saturation of the red lines and number and the green background. The texture from the ink jet printing can be seen in the green field. To properly print this image, sharpening can be applied and apply a proper printer profile for color management.|
|The achievable resolution is independent from the contrast of the image.|
|Clear resolution of the finest line details without moiré.|
|Fine Details will be Perfectly Reproduced|
|The texture of the paper is visible even in the plain white area. In the printed green area the paper texture is even more visible.|
- Use of the Systems
The table below compares some of the technical and operational differences of the Bayer sensor camera systems and the Trilinear sensor of the scanning backs. Fixed and enclosed scanning systems, such as flatbed scanners or book scanners also will use trilinear sensors, however these are used for specialized applications.
|Bayer Sensor: DSLR and Medium Format Cameras||Trilinear Sensor: Large Format Scanning Camera|
|Type of Lighting||Flash or Continuous Light||Continuous Light|
|Camera Shutter||Mechanical or Electronic Shutter||No Shutter Needed|
|Exposure||The Whole Picture
(Shutter Speed: 1/1000s to several seconds, up to 60 minutes)
(Shutter Speed: 1/500s to 1/20s)
|Resolution||max. 100 MPixel (Photokina 2016)||max. 2800 MPixel (Photokina 2016)|
|Time to Record 100 MPixel Image at Maximum Quality.||A Few Seconds||About 15 Seconds|
|Lifespan||Limited by Mechanical Shutter||Millions of Scans|
|Types of Subjects||All types of subjects:
People, Sports, Events, Still Life, Architecture, Products, and Landscapes
|Only static subjects with little or no movement: Still Life, Architecture, Products, and Landscapes|
|Types of Objects||Photographs, documents, books, sculptures, transparencies and negatives.
|Large size originals, objects with fine details, fine surface structures like metal, wood and stone, offset prints, Photographs, documents , books, sculptures, transparencies and negatives.|
|Color Management||Integrated, not specific to finished product (e.g. printed media).||Adjustable workflow for the finished product and use of image.|
|User||Photographer||Trained operator or Photographer|