SDI Technologies iH30 Especificações descarregar pdf (Página 61)

Advanced Topics

Quadro SDI Output PG-03776-001_v06 | 56

The Quadro SDI processes the input values at 12-bit precision. In the case of 8-bit input

data, the data is shifted up by 4 bits and the top 4 MSB bits are copied to the least 4 LSB

bits. For 16-bit per-component input data, only the upper 12 bits are processed. The RGB

data values come from the GPU output. The coefficient, offset and scale values are

determined as described in the following sections.

10.3.1 Coefficients

The coefficients as well as the offsets in the formulas above are signed 16-bit integers

which represent the values from -1.999 to +1.999. Each coefficient effectively has 14 bits

of precision which should be adequate to provide high precision. All multiplications and

additions are performed without truncation. Truncation only occurs at the final stage at

which time 10-bit or 12-bit output values are extracted.

Table 10-1. SD ITU 601 Coefficients

R G B

0.2989 0.5865 0.1150

-0.1684 -0.3310 0.5000

0.5000 -0.4181 -0.08095

Table 10-2. HD ITU 709 Coefficients

R G B

0.2130 0.7156 0.0723

0.5000 -0.4535 -0.0455

-0.1145 -0.38450 0.5000

Custom coefficients may also be programmed via the API as outlined in Code Listing 43

and Code Listing 41. For applications that wish to scale the results within a particular

range, the scale values must be considered in the computation of the coefficient values.

10.3.2 Scale

By default, the Quadro SDI performs color space conversion into what is commonly

referred to as Video Range. In the 10-bit case, Y ranges [64 – 940] and CrCb [64 – 960].

The default scale values are then computed as follows.

scale

y = (940 – 64) / 1024 = 0.85547

scale

cr = (960 – 64) / 1024 = 0.875

scale

cb = (960 – 64) / 1024 = 0.875

1 2 ... 56 57 58 59 60 61 62 63 64 65 66 ... 146 147

Sem comentários

SDI Technologies iH30 Especificações Página 61