[Techtalk] video editing on Linux

[Maria Blackmore]

On Sun, 29 Dec 2002, Almut Behrens wrote:

> just out of plain curiosity: with "telecine", are there any specific
> measures taken against getting interference artifacts, as one would
> normally expect when resampling already sampled, discrete data? When
> not synchronized in one way or another, I'd expect to see more or
> less pronounced flickering at a frequency equal to the difference
> between the two original frequencies (film/camera shutter-freq.).

ah, interference, a beat in terms of sound/music

<snip much interesting discussion>

This does of course assume that the mark/space ratio of when the shutter
is open is as you described, which is a vaste oversimplification of what
actually happens.  Also, you should bear in mind that film runs at 24
frames per second, whereas NTSC runs at 60 frames per second, interlaced 
... plus the actual frame rate that the camera is taking pictures at does
not have to be 60 frames a second, it could actually be taking 180 and
averaging them, nor does it have to be at a 50% mark/space ratio, for
example it could be taking 1/1000th of a second exposures 60 times a
second, or it could be taking 1/70th of a second exposures 60 times a
second ..

thus, for a period of 0.25 seconds (with the assumption that film and
video are not in sync) with long exposures

film:  #########-#########-#########-#########-#########-#########
NTSC: 000-111-000-111-000-111-000-111-000-111-000-111-000-111-000-111

# indicates time when light is shining through the film
0 indicates an even frame of NTSC is being taken
1 indicates an odd frame of NTSC is being taken

or with short exposures

film: #########-#########-#########-#########-#########-#########
NTSC: --1---0---1---0---1---0---1---0---1---0---1---0---1---0---1

This is also an oversimplification, but I hope it conveys what i'm saying

You might like to notice that with the short exposures, it's much less
likely for an exposed frame to take a picture of nothing ...

it's simply time to fiddle with settings until you find something that
works with a particular cameras characteristics

nevertheless, I think that for a non-professional system, it should work
out quite well.  Professional stuff methods of dealing with this involve
showing some frames twice or (in non-NTSC locales) running 24 frame per
second film at 25 frames per second (because the sensible world uses PAL
:)

The way I did it once was to rig something up using a light gate on a
projector to get a pulse train at 24 pulses per second to form a basis for
synchronisation to the projector, then use this pulse train to drive a 48
frame per second PAL black generator.  Once you have this 48 frame per
second black signal you can use it to genlock a studio camera to the
projector using the conventional trimmers on the camera to introduce the
correct delay so you only take pictures of film.  The catch with this is
that you've now got 48 frame per second PAL, yuck, timings are all wrong,
nothing on earth will display it.  However if you happen to have a
framebuffer from an old DVE (digital video effects) unit, it will quite
happily (or at least without noticable complaint) read in this unholy mess
and store it in the frame buffer.  At which point you can genlock the
output with a nice standard 50 frames per second PAL black signal, and
you've got something you can give to the world. (for better or worse)

What a bodge :)
not all cameras will tolerate this level of abuse, most will simply refuse
to play along, some will work, some will require modification to
work.  All in all, it's not something I would recommend trying at home,
especially since the first hurdle would be finding a camera to use that
has a genlock input and susceptable, and the second would be finding a
framebuffer to abuse. Niether of these are things likely to be found lying
around.

It was fun though :)

> It's the visual system of our brain that integrates the individual
> frames into a seemingly continuous stream of light and movement.

true

> Cameras unfortunately don't possess this ability (at least not at
> present)...

don't they? I thought CCDs integrated the light falling on them over the
time which they are sampled? :)  quantisation

> (devil's advocate ;)

heh, yes.

It's been a while since I did this, I'm much more into still photography
now.

Maria