Softclip Conversion

Appendix C provides the equations used to perform Cineon Softclip conversions in the Bit Reduce and Bit Expand nodes (as well as in the DPX/Fido tab sections of the Disk Input and Disk Output nodes).

Note:
The Cineon Softclip conversion algorithm works with 10-bit log data, even if the input to the Bit Reduce or Bit Expand node is linear. However, you do not need to convert to 10-bit log before assigning values to the conversion parameters.
If the input image data is linear, you should enter corresponding linear values in the Softclip conversion parameters. Chalice will convert these values internally to 10-bit log according the specifications of the "Advanced - Cineon Settings" parameter tab.

Bit Reduce Node

The following equation shows the precise conversion process for values above Reference Black:

A breakpoint is computed for Softclip:

Breakpoint = RefWhite - Softclip.

If an input value is below this breakpoint, the output will not be influenced by the Softclip setting. These values will be calculated as follows:

OUT = 10 ^ ((IN - RefWhite) * 0.002/Film Gamma) ^ (Display Gamma / 1.7) * (Gain - Offset)
Gain = (2 ^ N - 1) / (1 - 10 ^ ((Reference Black - RefWhite) * 0.002 / 0.6) ^ (Display Gamma / 1.7)
Offset = Gain - (2 ^ N - 1)

If an input value is above the breakpoint, the output must be adjusted according to the parameters specified in Softclip:

OUT = (IN - Breakpoint) ^ (Softclip / 100) * Kneegain + Kneeoffset
Kneeoffset = 10 ^ ((Breakoint - RefWhite) * 0.002 / Film Gamma ) ^ (Display Gamma / 1.7)) * Gain - offset
Kneegain = (2^N - Kneeoffset) / ((5 * Softclip) ^ (Softclip / 100))

Variables:

OUT = Converted output value.
IN = Raw Cineon Input value.
RefWhite: The 10-bit log value of the 16-bit linear value entered in the Reference White parameter. This value is calculated according to the settings of the "Advanced - Cineon Settings" parameters. If defaults are used, 4095 in Reference White will be converted to 685. The latter value will define RefWhite in the equation.
Film Gamma: Read directly from the Film Gamma parameter.
Display Gamma: Read directly from the Display Gamma parameter.
N = The Format type of the converted image. For example, a conversion to 8-bit linear would result in a value of 2^8 - 1 = 255.

Bit Expand Node

The conversion assigns output values by computing the inverse log of the input Cineon data, then scaling and adjusting according to the reference white, reference black, film gamma, display gamma, and output data format, as the following equation shows:

OUT = RefWhite + log10[((IN + Offset) / Gain) ^ (1.7 / Display Gamma)] / (0.002 / Film Gamma)
Gain = (2 ^ N - 1) / (1 - 10 ^ ((Reference Black - RefWhite) * 0.002 / 0.6) ^ (Display Gamma / 1.7)
Offset = Gain - (2 ^ N - 1)

Variables:

OUT = Converted output value.
IN = Raw Cineon Input value.
RefWhite: The 10-bit log value of the 16-bit linear value entered in the Reference White parameter. This value is calculated according to the settings of the "Advanced - Cineon Settings" parameters. If defaults are used, 4095 in Reference White will be converted to 685. The latter value will be used in the equation.
Film Gamma: Read directly from the Film Gamma parameter.
Display Gamma: Read directly from the Display Gamma parameter.
N = The Format type of the converted image. For example, a conversion to 8-bit linear would result in a value of 2^8 - 1 = 255.




Table of Contents | Index



Copyright 1996-1999 by Silicon Grail Corporation