[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

TWiGi explained from Dave Corbitt



the following was written by Dave Corbitt and sent to me for
approval-- I believe the information is relevant to many ongoing
discussions so passes the 'not marketing' test to be on this
mailinglist.

--Rob

------

There have been a number of statements or questions about TWiGi and I would
like to relate my own understanding of how it works.  This message was
reviewed and approved by Stuart at Innovation TK so I'm not giving away any
trade secrets.

We have 2 TWiGi systems installed and running here at Manhattan Transfer on
our upgraded URSA 4x4's (looks like the old URSA but has a heart of GOLD) and
the results so far are very impressive.

TWiGi is a substantial improvement in performance for any URSA, be it old 422,
422 updated to 444 "Gold Equivalent", or the factory built GOLD with it's
newer sheet metal and color scheme.  TWiGi is not just a burn corrector for
URSA, it has an entire new set of A to D cards with true 12 bit flash
conversion and oversampled 2x normal clock frequency.  Not only that but there
are 2 A-D chips per channel sampling on alternate 1/2 clock cycles out of
phase with respect to each other so each new URSA pixel is an average of 4
true TWiGi samples.  That alone cuts down on A-D errors and noise enormously. 


The linear signal out of the cell box is slightly gamma corrected before the
12 bit A-D to scale the bits better at the higher density end of the scale and
then rescaled back to linear after the A-D to satisfy the URSA color channel
parameters.  The math to do this is easy and creates new values for video that
are rounded off to the full bit range of the URSA channel.  This is in
comparison to the Cintel dual flash with 8 Big bits then the in between little
bits A-D conversion.  My own measurements of this technique shows it can't
create more than 10 useful bits sampling depth.  The rest is all noise or
error.  Once this video is gamma corrected the sampling errors become some of
the noise we have all come to expect from the various versions of URSA.

TWiGi also has rewired the cell box and replaced the head amps.  This trick
gets rid of noise pick up from the switching supplies and noise generated by
less than ideally designed ladder networks and head amps.  Afterglow
Correction is done very close to the cell box in the same box that does the
non-linear function preceding A-D.

The burn correction sensors are another well thought out design.  The sensor
assembly has Red, Green, and Blue filtered sensors to let the burn sensing
spectrum be divided to correspond to the the spectral distributions sensed by
the cell box.  That way a different signal is generated to correct the burns
for R, G, and B.  This allows a true reading of what is really going on
relative to each primary color.  CRT grain and burns can be and are different
in different parts of the spectrum.  The TWiGi sensor is the first burn
corrector (to my knowledge) to actually try to do something about that.  The
sensors are fully solid state and arranged in a circle around the face of the
CRT.  There are at least 4 sensors per channel.  

The burn signals are Afterglow corrected and digitized and then digitally
multiplied by the video after A-D to remove all sorts of artifacts of the CRT
from the video (CRT grain, burns, level differences between Run and Still). 
The burn correction works so well that the operator can change beam current
between 300 and 150 and see no difference in color balance or RGB levels.  All
he/she will see is a slight worsening of the signal to noise at the lower
setting.

I am wondering what Dave Walker's Accuglow for URSA will incorporate.  Does it
have RGB sensors for burn?  Will it have new A-D's to get rid of all those
errors so common to URSAs?  If it does not then I don't see how it can hold a
candle to what Stuart has done with the TWiGi.  If it does then there may be
some real competition in this niche.

Dave Corbitt