Canon EOS C300 Mark II PL 手册
4.0 Formulation of 2K or HD RGB 444 Video Components
The image sensor parallel readout architecture breaks down the Bayer sampled 4K / UHD data
into four parallel 2K / HD components as shown in Figure 2.
into four parallel 2K / HD components as shown in Figure 2.
Figure 2
Showing the unique component video signal processing in the C300 Mark II that
formulates the 2K / HD RGB 4:4:4 video component set that is sent to the XF-AVC codec for recording
A low noise analog column amplifier implements the ISO gain control when the scene
Illuminance is very low. This is followed by the A/D conversion process and digital gain controls
that implement the camera White Balance adjustments at a high bit depth. At this juncture the
video processing radically departs from that of the earlier C300 camera. First, a new
processing feature is added – lateral chromatic aberration correction (of any EF lens that might
be employed) – and this is then followed by Canon’s unique Over Sampling HD Processing.
Illuminance is very low. This is followed by the A/D conversion process and digital gain controls
that implement the camera White Balance adjustments at a high bit depth. At this juncture the
video processing radically departs from that of the earlier C300 camera. First, a new
processing feature is added – lateral chromatic aberration correction (of any EF lens that might
be employed) – and this is then followed by Canon’s unique Over Sampling HD Processing.
5.0
Oversampling HD Processing
This innovative video processing begins with the upsampling of the video components that had
been directly readout from the 4K image sensor – to form three 8 Megapixel RGB frames as
outlined in Figure 2. That upsampling moves the first order sideband (from original image
sensor sampling) to a higher frequency which in turn opens spectral space to implement pre-
filtering prior to a subsequent downsampling back to 2K / HD RGB444 or YUV422 frames. The
net result of this sample rate conversion process – termed “Oversampling HD Processing” by
Canon – is to produce three 2K / HD RGB components that have minimum aliasing and a more
benign appearance to noise (in high ISO settings). At this stage these essentially linear light
RAW representations of the image sensor outputs have a linear matrix applied to them to define
the specific color gamut selected for the production (there is a choice of four such gamuts as will
be discussed later). This is followed by application of the camera optoelectronic transfer
function (OETF) which digitally maps the 14-bit depth linear components to a nonlinear set at
either a 12 or a 10-bit coding. The OETF can be selected to be the new Canon Log 2 –
optimized for the 15-stop dynamic range of the C300 Mark II – or the former Canon Log.
been directly readout from the 4K image sensor – to form three 8 Megapixel RGB frames as
outlined in Figure 2. That upsampling moves the first order sideband (from original image
sensor sampling) to a higher frequency which in turn opens spectral space to implement pre-
filtering prior to a subsequent downsampling back to 2K / HD RGB444 or YUV422 frames. The
net result of this sample rate conversion process – termed “Oversampling HD Processing” by
Canon – is to produce three 2K / HD RGB components that have minimum aliasing and a more
benign appearance to noise (in high ISO settings). At this stage these essentially linear light
RAW representations of the image sensor outputs have a linear matrix applied to them to define
the specific color gamut selected for the production (there is a choice of four such gamuts as will
be discussed later). This is followed by application of the camera optoelectronic transfer
function (OETF) which digitally maps the 14-bit depth linear components to a nonlinear set at
either a 12 or a 10-bit coding. The OETF can be selected to be the new Canon Log 2 –
optimized for the 15-stop dynamic range of the C300 Mark II – or the former Canon Log.
3