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

Re: Telecine

Hi Larry,

<< I want to do a story for our Feb '98 issue on film-to-data transfer as
opposed to film-to-tape transfers. >>

Here are some comparisons and contrasts between these two techniques:


Mostly done in telecine suites to produce standard video formats from film,
usually running in real time.  Typically used for TV programming or for post
production.  Color correction,  image enhancement, and generation of edit data
usually part of the process.  The majority of telecines now in operation
(Cintel Mk. III C and Turbo, URSA and variants, FDL-90, Quadra, etc.) scan at
approximately 1k by 1k, which is interpolated to 525/60 or 625/50 and then
processed and recorded directly or encoded.

Generally speaking, these machines were built for SDTV operation and do not
have the optics, scanning, servos, or signal processing necessary to provide
satisfactory HDTV operation.  The extent of the modifications and new
components added to Mk. III C telecines to turn them into "Hi-Res 1440"
machines bears this out.

The "next generation" of telecines (Philips Spirit, Cintel C-Reality--when it
becomes available) scan at approximately 2k by 2k, making them suitable for
HDTV and certain classes of film scanning.  Of course, the output can always
be converted down to 525/60, 625/50, or whatever, and the results are usually
quite good, since the interpolation starts out with more information than was
present in the older machines.

At present, the chroma channels of the video from most telecine suites are
either sampled at half the luminance rate (e.g. 4:2:2), or they are eventually
interpolated to that level, since there are no standard RGB videotape
recorders or transmission systems.  It appears that this will continue to be
the practice in HDTV, since the digital TV schemes currently being promoted
all seem to be based on Y/Cr/Cb or variants thereof.


Usually done on scanners (or telecines that can act as scanners), not in real
time.  The slow speeds of scanning are due to the cost and difficulty of
implementing data interfaces, the cost of storage media, and mechanical
limitations of obtaining perfectly steady images.

Scanning is typically done when it is desired to perform computerized effects
on images.  These include--but by no means are limited to--the addition of
titles, fixing image defects (such as wire removal), or creating special
effects.  As a general rule, film-to-data transfers are done on specific
frames, or blocks of frames, not entire shows.  Output may be to video, but
more likely back onto film, or to print.  

The resolution of film scanning is often adjustable, and may be different
depending on what the customer needs.  2k by 2k scanning is the norm for 16mm
and some classes of 35mm work; however, 4k by 3k can be used for 35mm where
the output will be back to 35mm or larger film.  There are about a dozen image
file formats in use, and many programs either use or do conversions between
more than one.  Film scanning is almost always done with equal numbers of
samples for red, green, and blue.

Best regards,
Christopher Bacon

No advertising on the TIG without permission [contact rob at alegria.com]
TIG subscriber count is 908 on Mon Jan  5 09:43:10 PST 1998
complete information on the TIG website http://www.alegria.com/tig3/