[tig] HDnet You gotta read this!!!
Fri Apr 2 11:02:02 BST 2004
Martin Snashall, a design engineer I work with (he goes back to the
early days of digital video -- he designed the Abekas A64 and A84)
wanted to add this to the discussion. (He's not a TIG subscriber, but
I forward him posts of interest.)
1] The reason for showing films in the cinema double or triple
due entirely to persistance of human vision. This effect is well known,
should be understood by all TIGers.
2] When displaying images on a CRT, both human persistance of vision,
CRT lag come into effect. Most people notice the relative flicker of PAL
against NTSC, but people in the PAL world adjust to the flicker, and
notice it unless PAL and NTSC monitors are placed side by side. With a
the phosphor starts to decay in luminous intensity immediately after it
charged. If the decay is too long, then you get motion blurred images.
Again, these effects are well known. Computer monitors tend to have
persistance phosphors, which means that they need to be refreshed at a
greater than 50fps.
3] LCDs are significantly different from CRTs. When a pixel is written
the state of the crystal starts to change, and may take up to 50mS
from opaque to transparent (black to white), and vice versa. However,
significantly, once the pixel has been written to, the pixel state does
change, in other words there is no decay. Lag (both on and off), but no
We have carried out various tests using a 1920 x 1200 TFT connected
custom hardware. We drove the TFT native in LVDS, and the only
was that we kept within the frequency specification of the LVDS link.
results may be surprising.
We initially drove the TFT at 60Hz frame rate (16ms per frame
using a fixed test pattern, and with around 10% vertical blanking. The
picture was solid, as would be expected. We then drove it with 120%
blanking. This meant that the screen was receiving refresh for 16mS,
was then not updating for the next 16mS. Again, the results were a
picture. Increasing the vertical blanking to 6000% (yes, read the
also produced a rock solid picture, with no flicker and no decay.
This meant that we were only updating the picture every second, taking
16mS to do
so, and then effectively not addressing the display for the next 980mS.
CRT terms, this is equivalent to displaying only one frame in every 60!.
These results lead us to the conclusion that it is not neccessary to
attempt to drive a TFT at 72fps, as there would be no visual difference
between 72fps and 48fps.
The 1920 by 1200 monitors (such as the Apple Cinema
display) can qiute happily be driven DVI at 48fps. Note however, that
one connected a CRT monitor to such a formatted signal, then you would
either get flicker, or a notice saying that the frame rate is lower than
I'll certainly forward any response on to Martin, who will be at NAB
with me showing some of the fruits of our labors.
More information about the Tig