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

Spirit DataCIne

On 29 April Paul Grace submitted some questions to the group about the
Spirit DataCine, I would like to address some of those questions.

Paul asked the following questions :
>I understand the data output will be one HIPPI channel so the maximum
speed will be less than 10 fps HIPPI = 100mB/s ?

The data output from the Spirit is limited by the available external
interfaces in the computer industry. We are not developing a proprietary
data output interface for the Spirit, but prefer to adhere to industry
standards. Film data transfer rates are determined by the film format, for
example academy 35mm film is good for 10 frames per second, but
Cinemascope which requires more than 1800 lines per frame will
transfer at approximately 7 frames per second ( assuming 100mB/s
sustained HIPPI transfer speed )

> if multiple HIPPI cards are used how many and what type of data
recorders will be used and at what cost ?

It was decided to implement single channel HIPPI in the machine because
this was the easiest and most economical solution for the marketplace
today. All HIPPI hardware exists today and can be purchased for almost
all high end computer equipment, there also exists the possibility of using
serial optical HIPPI hardware which will become available in a matter of
weeks making this interface method more practical from a physical layer.
The data output is an option on the machine and the and the whole
platform is built around a modular concept. When the computer industry,
(and the whole infrastructure) jumps the next step into higher transfer
rates and new interfaces, we can deliver a new faster interface for our
machine ( e.g. HIPPI 6400 , fiber channel ect ) But not before a standard

> What is the sustained data rate from the Telecine, HIPPI cards use
cache ram to burst data out, this will limit the sustained data rate.

Our HIPPI card is our own Philips design, developed for the DataCine
application, it is not an #off the shelf# component. We use the OTS HIPPI
chips, but the rest of it is a custom design. HIPPI can handle 100
megabytes / s but some form of protocol is required which takes up
bandwidth becoming overhead. Our design sends raw film data and
does not use the IPI3 protocol. Therefore we can transfer the maximum
theoretically possible data rate using HIPPI`s framing protocol ( raw data
) the net transfer rate should be close to 100mB/s. We do not use cache
on our HIPPI output card, we have an expandable internal full resolution
frame buffer ( up to 34 frames )  to continuously #feed# the data output.
There should be no limitation on data transfer rates. The DataCine itself is
able to produce in excess of 320 megabytes / s, you would need four
parallel HIPPI channels to transfer this data - with existing computer
technology this is not economically feasible.

> Stabilizing images in a workstation is slow and defeats the object of a
fast scan. If a 10 second effect shot is transferred via HIPPI at say 6 fps
it will be boring and extremely expensive to fix these shots in inferno,
negating the need for higher speed scanning ?

Stability and the ability to transfer images via the data output and via any
of the video outputs from the machine in a stable manor is high on our
priority list. We have very good basic stability on the machine which is as
good as Jump Free, with our near term plan a Steadygate solution similar
to that shown on the Quadra at 3 to 4 frames per second, and finally our
long term goal is to complete a electronic real time stability system.

> Cost, why is the cost so high ? we need realistically proceed tools ?

Yes it is true this is an expensive machine, but so is the technology, and
processes derived to bring this product to market. This machine has
been in continuos development for many many years now ( as the FLH
1000 ) and we accept this may not the platform for everyone in the film
transfer business - it is a very different approach to conventional
Telecine technology and methods which, hopefully, will improve the
quality, throughput, reliability and consistency of film transfers. Things do
cost money to design and manufacture, and this is a relatively small
worldwide market. While it is our goal to achieve the best possible price
for technology in the marketplace we are driven by cost factors. The
machine does have the ability to apply itself into multiple applications,
525/625, HDTV, EDTV, and film scanning applications so the potential is
there for multiple revenue streams.

>What choice of color correctors are available ?

Basically any of the currently available high end correctors available
today, DaVinci, Pogle and the Copernicus systems are fully compatible
and will control our system.

> In data mode if the hi def monitor is being used to grade on , what
contingency has been made for color space issues to be addressed and
contrast range handling to be addressed ?

The Spirit DataCine is a film scanner concept, not a conventional Telecine
approach, this means that in terms of color space it does not limit or #cut
off# any of the film color space, neither does it compress dynamic range.
The optical filtering for the CCDs ( i.e. the colorimetry ) match as closely
as possible the colorimetry from say a Genesis scanner from Kodak (
remember Kodak build the imaging system for the Spirit ) There is no
reason why we cannot catch the full film color gamut of a scanned
image. By selecting different TV standards as an output, this signal will
match the color gamut of the selected TV standard e.g. CCIR 601, HDTV
1125 ect. We also have the ability to switch of CRT gamma ( or tube
gamma ) from the selected TV standard, or data output file if desired.  

The issues of film color space, tv color space, and even film recorder
color performance are important issues with no simple solution - the
same color space concerns are present in scanner / printer technology
in use today with #look up table# and color space re mapping solutions.
This is a viable but slow process using software again. We are
investigating system solutions that will hopefully address all these issues
in keeping with the high speed technology we have developed in this

Steve Russell
Marketing Manager - Film Imaging Products
Philips BTS