Filter
Top Products
Barco - RHDM-2301 - User manual 157
Video signal processing
But a gamma of 2.0 sounds very strange in the broadcast world! All these years, video looked good
on CRTs, where people are used to values of gamma of 2.2, 2.5, or even 2.8, but never 2.0. Thus
this mean that we have been looking at non-linear end-to-end transfer?
Indeed, according to CRT physics, the cathode ray tube transfer characteristic has a gamma of 2.5.
An early EBU specification actually uses a number of 2.8. However, a later in-depth characterization
of the gamma of a CRT (Alan Roberts: Methods of measuring and calculating display transfer
characteristics (gamma); BBC RD 1991/6) yields values closer to 2.2-2.3.
So yes, in fact the scene-to-screen transfer has not been linear from the start. With an effective
encoding gamma of 0.52, and an effective display gamma of 2.2-2.5, the system transfer is
somewhere in between 1.1-1.3. A recent EBU recommendation (EBU.Tech.3320) actually specifies
the value of 2.35 for new reference (grade 1) monitors.
Why does this non-linear scene-to-screen transfer give such good results?
The scene and the display have very different viewing conditions. Video monitors are typically
calibrated at a luminance of about 100 cd/m
2
, and viewed in dim environments. A scene can have a
luminance which is 100-1000 times larger (the sun!) than the luminance which can be produced by
the display. These different conditions give rise to perceptual differences: a video shown on a
display with a linear end-to-end transfer function will look very pale and unsaturated. Therefore a
“transfer gamma”
(end-to-end) with an exponent higher than 1 is desired (around 1.2). This
situation is referred to as “rendering intent”: the video is encoded in such a way (exponent 0.45,
effective gamma 0.52), that when a CRT decodes it (with a gamma of 2.2-2.5), the end result has
an end-to-end transfer gamma of around 1.2.