MARC

The MARC project focused on improving the path from accurate imaging to accurate printing. A conventional RGB system is used in a portable camera. However, a calibration technique similar to that used in VASARI produces very accurate images in the CIE Lab color space.

The MARC system uses a scan-back camera where a stepper motor system is used to move an array CCD in the image plane of a conventional large format camera.

The MARC camera is a digital back attached to a standard 6cm by 6cm Bronica camera body. The digital back contains a masked 500 pixel by 290 pixel CCD sensor mounted on a two way motorized stage. The camera uses both micropositioning and macropositioning. In micropositioning, the motors drive the sensor to 6 by 8 positions within the pixel separation distance of 11um by 17um, repeating 4 times for red, green, blue and green lacquered sites. In macropositioning, the sensor is moved across the 6cm by 6cm focal plane of the lens to seven by nine positions.

During manufacture, two calibration steps are performed. A set of motor calibration files are found through an iterative process taking several hours with the camera mounted on an optical bench and pointed at a defocused checkerboard pattern. Measurements of apparent brightness at each of the 500 by 290 sensor sites are taken with the camera pointed at a smooth evenly illuminated white target.

Three calibration scans are made. A lens cap black is taken. This is used to determine the dark current for each sensor site. An image of a Macbeth Color Checker chart, placed in front of the center of the painting, is made. This is used to determine a color calibration matrix that will convert RGB to CIE XYZ values with minimal errors in Lab space. A scan of a piece of gray card, placed where the Macbeth chart was placed, is made. This is used to remove non-homogenates in the illumination of the area covered by the chart.

The painting is scanned with the camera that is capable of making a complete scan in about 15 minutes at maximum speed. The time taken can be reduced to about 5 minutes by skipping some positions and interpolating values. The fields taken during acquisition are compressed and saved. Little calibration is done during acquisition to keep the exposure time short.

After scanning is complete, image calibration, an automatic process taking several hours, is performed. After corrections are made, the calibrated frames are blended together to form a single corrected, high-resolution color image. The high-resolution image and raw data are stored on tape. The high-resolution image and a low-resolution image are stored to optical disk.

The Marc book is printed using an automatic gamut mapping procedure to represent colors outside of the gamut. Almost the entire gamut was printed unaltered, colors near the edge of the gamut were moved in slightly and color way outside the gamut was clipped. Paper cast was corrected for by moving the D65 neutral axis onto the paper axis. One proof was printed for one image from each press plate. During the final print run, the proof is used to keep the press within tolerance. The color match in the final book is generally very good but some differences are apparent, especially in paintings with large areas of deep brown, a very difficult color to reproduce well.

Applications of the MARC system include assembly of infrared reflectograms, non-destructive assessment of transport damage and print publishing.