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

Re Ursa vs Spirit / Ursa NT

With reference to;
>>I was quite surprised
>>to see, in the transfer done on Tape House's Ursa Gold, unbalanced
>>whites and blacks, and some sort of contours problem or coring issue
>>that caused significant extra aliasing and line-pairing artifacts.  We
>>don't have problems of this magnitude on our Cintel machines.

I do not wish to join in the mud slinging with John, Tape House and Rank,
but I thought you might be interested in our most recent tests done just 
before IBC.

As part of our acceptance we visited Germany with various negative rolls and
recorded our efforts onto D1. As the facilities there are not set up to the sort
of level one would look for at home, and anyway we were more interested in
stability than colorimetry, the actual grading is not  "Client Perfect" but the 
signal recorded is quite startlingly clear, crisp and noise free.

We then had the notion to do a split screen test and remember this was 
before John appeared at IBC. I should also say that we have tweaked our
Golds a lot and, with the Hush Kits fitted, we think that their pictures are 

As the Spirit transfer had no noise reducer or post Tk processor in the chain 
we had to also remove these from the Rank path and grade the negative 
using only the PEC's and colour channel. This is one reason the Rank 
looks so bad in the split test.  It is impossible to match the transfer curve
of the machines, especially the gamma characteristic, and to try and do so
forces the Gold into a state where the system noise is totally unacceptable.

In normal neg transfer mode we stretch the Rank gamma to the max and 
force the PEC's as  "low", ie neg bright, as possible and use the far greater 
range of the DCP gamma to re compress the contrast whilst forcing gain
through the image. The actual Rank output is thus a very light, low contrast
image but one that is essentially noise free.

In a split test the contrast has to be supplied by a combination of PEC and 
colour channel balance and this really shows the old problems of system
noise being clamped at black, ( whites on neg ), and the inadequacy of the
PEC's once their output is compromised by grading requirements.

The noise produced in this way is compounded by the enhancement of the
A to D errors which we have minimised with the re-ranging of the Hush Kit.
On a standard machine, especially one that has not had the Rank op amp
mod, these errors must be very obvious and will be worsened with VAC

So if you set up the Rank in stand alone mode against the Spirit it is not 
good and the next thing we tried was to set the room output, ie DCP and
DVNR, against it.  Now, by using the Rank in its normal  "noise minimized"
mode and grading on the DCP the picture quality is back to its normal 
standard but you still cannot match the curve of the Spirit.

Doing the matching by eye we found that where the Spirit is producing a
signal that is fully unclipped at both ends and is holding detail right up to 
the roll off point at both ends, the Rank chain is invariably black crushed 
and white clipped with less detail in the mid range. The Spirit is handling 
the detail with the same toe to shoulder curve as the neg handles the 
exposure with none of the linear characteristics of the Rank. 

It is very difficult to be unemotional about such a test as we have all spent 
a long time being very critical of image quality whilst constantly seeking
improvements to the available technology so when the threshold changes
so radically it is bound to produce extreme reactions. These reactions were
similar in the Ursa / MK111 discussions and there is no doubt that a 
split screen between those two produced results that showed the improved
stability, resolution and low level colour detail of the Gold. It was on such a
that we finally decided to retire our own MK111's especially when other features
such as Scan Effects and Low Speed were taken into consideration.

There was a valid point in doing such a test then because we were looking
at like technologies where improvements bore a relationship to an existing
cost structure which,as Dick Hobbs has pointed out, is not the case with
the Spirit.  A far more valid test is between a Gold fitted with the new Twiggy
setup for instance and a pre enhanced one. Twiggy has gone far beyond
the initial A to D improvements and is now, via a diode array  "burn (tube)
corrector", removing nearly all of the CRT problems including no zoom 
through zero errors, runstill errors, or patch size change errors during
varispeed or 625 / 525 operation. I belive there is also diode stabilization
of the PEC output and an end to the well documented drift errors. 

The other Rank update of enabling the store freeze in stop mode, which
halts all PEC noise, is less impressive and seems designed to confuse
as it is not then the same as a Quadra or Spirit and certainly not when
you run. It does make the picture demonstratabily noise free though, even
if it is contrary to one of the main thrusts of Cintels marketing in that you
are grading a  " live " image.

Thus far the introduction of Spirit has been beneficial to all Ursa owners 
as the acceptance of Twiggy etc by Cintel will finally allow the machine to
function as the original marketing blurb suggested, even if there is more 
costs involved, and will let owners pass on the improvements to their clients.
Spirit customers are looking to apply a new technology to the market and to
push the traditional telecine transfer process into new areas. This adaptability
coupled with superior image processing was the Spirits main attractions for us.

Seamus O'Kane 
VTR Ltd. London.