|
|
Posted by Derek Gee on 10/24/91 11:46
"MovieStuff" <moviestuff@sbcglobal.net> wrote in message
news:1146281097.540069.21090@j33g2000cwa.googlegroups.com...
>> Derek Gee wrote:
>
>> I have seen problems when using NLE editors to interpolate film speed.
>> For
>> example, 16mm film shot at 16-18 fps was transferred at 24fps. Using two
>> different brands of NLE editor software to bring the rate back down close
>> to
>> original shooting speed left all kinds of weird visible artifacts.
>> Please
>> explain how your interpolation works differently from the NLE software.
>
> I think the problem you are describing starts with the film being
> transferred at the wrong speed. If, for example, you have 15fps footage
> and you transfer it, instead, at 24fps on a Rank, then it already has
> pulldown frames built in, which are difficult to remove and will
> definately produce some weirrrrrd artifacts if you try to change the
> playback speed back to 15fps after transfer using any sort of speed
> change software. But, if you ask the Rank operator to simply transfer
> at 15fps (or actually 14.97) to begin with, then the Rank will
> automatically use a different pulldown pattern that puts each film
> frame neatly on two consecutive video frames. Or, you could ask the
> Rank operator to transfer at 30fps (29.97) and end up with each film
> frame on an independent frame of video, which is really ideal because
> then you can apply ANY speed change you want without the types of
> pulldown artifacts you are describing.
I believe you are correct. The film was shot with a spring wound 16mm
camera at what should have been 16 fps, but in reality could have been
anywhere from 15-18 fps. Transferring at 24 fps with a 2:3 pulldown would
have yielded a bunch of extra frames which I think should have been dropped
completely from the file before attempting to change the speed. Maybe just
for curiosity's sake I'll take a clip from that early transfer and see if
that works.
Taking film that was shot at 18 fps and transferring at 20 fps, and later
slowing down to 18 fps in the NLE seems to work fine without generating any
weirdness.
> Likewise, with the WorkPrinter and Sniper units, each frame of film
> exists on an independent frame of video with no interlaced motion
> artifacts within each frame. In other words, even though a standard
> interlaced NTSC or PAL camera is being used to capture each frame,
> nothing moves during the first and second field scans because the frame
> is held stationary during exposure, so the computer will end up seeing
> those two fields together as a single, progressive scan frame. As such,
> you can then apply any pulldown pattern necessary to create the speed
> desired, just a Rank does.
>
> For instance, if you need 15fps, then CineCap will just double every
> video frame. If you need the classic 2-3 pulldown, then it will create
> an interpolated frame pattern, just as the Rank uses. If you need
> 18fps, then it will create yet a different pattern. But what you can
> NOT do is apply, say, a 15fps pulldown pattern to footage that has
> already been altered to 24fps with a 2-3 pulldown pattern. THAT appears
> to be the problem you are describing above, where you have footage that
> was shot at 16fps but was inadvertantly transferred at 24fps, then run
> through NLE software (which is never the greatest for speed changes,
> anyway) to try and bring back down to 16fps. You can do it but you
> would need to first use software of some kind to remove the pulldown
> frames from the 24fps footage, render that out, then use that new
> pulldown-free footage to convert to 15fps (or 16fps, if needed).
Hey, I figured that out without even reading this paragraph yet... The last
time I used software to remove the pulldown though, it INCREASED the size of
the output file. Any idea why it would do that?
> So the problem you are describing isn't a problem with the CineCap
> software, because the CineCap software is depending on you having total
> frame discretion, where the initial captured file has one film frame
> per video frame. Even though NLE systems typically do a poor job of
> telecine speed changes compared to CineCap, any NLE system will do a
> much better job if you remove any pre-existing pulldown frames before
> applying a new speed change pattern. Likewise, if you try to use
> CineCap software on film that already has a pulldown pattern in it,
> then it is going to look very, very funky. :)
>
>> > More specifically, regarding the qualitative difference
>> > between our units and a Rank, please see the comparison done in PC
>> > magazine where footage was sent to us and to a lab with a Rank Turbo:
>> >
>> > http://www.moviestuff.tv/whats_new.html#Sniper%20vs%20Rank%20in%20PC%20Magazine
>>
>> A favorable comparison, but in the still photo above, the Rank transfer
>> has
>> better shadow detail and more pleasing flesh tones.
>
> Certainly not $249,000 worth of shadow detail. And if you look at the
> last photo, the Sniper still is clearly better, with the Rank transfer
> washed out with lousy color. But in truth, the results are so close
> that operator input probably accounts for more of a difference than the
> technology used.
Agreed - the skill of the operator is paramount here. If you get a cheap
rate from some post-houses as a test/stand-by/take pity on me rate, you may
get an inexperienced operator still being trained.
>> The big problem with the Moviestuff products is the inability to transfer
>> sound films.
>
> Agreed. We have not concentrated on sound units because the demand has
> mainly been for silent film footage. However, we have been pestered so
> much by archive houses that we are actually about to release some new
> sound units that will allow frame accurate transfers with perfect,
> frame accurate lip-synch. We're pretty excited about it. They should be
> very popular.
That's great news! I just might be buying from you if you can come up with
equipment to do that!
Derek
Navigation:
[Reply to this message]
|