How do they speed up TV shows?

[lemmy999]

Many classic TV shows have been edited to allow for more commercials. But many networks also play the shows around 4-5% faster the gain the extra time instead of (or in some cases, in addition to) cutting. Does anyone know how they do this? Does it have be done at the point where they convert the film to Video (ie, during the telecining process) or do they simply drop 1 frame/ second and hope you don't notice? Or maybe every second they just take out 2-3 of the repeated fields per second (the repeated fields added during the telecining film->video process)? Thanks.

[sitlet]

they usually either take out 1 or 2 frames a second, or just speed up the program 1-2% or so. if a program was originally 30 min, and you increase the speed by 1%, then the new program would be 27 mins, and you most likely wont notice the speed difference. another way they would do it is to speed up only some scenes, say a shot of a landscape or something thats not important will be sped up

[babyboo]

...BTW...1 percent of 30 minutes is 18 seconds...not 3 minutes

[lemmy999]

But what I am asking is physically how do they do it. If it was a film projector showing 24fps, then I could understand how you just speed it up and project maybe 28fps and it would be faster. However, since NTSC is 29.97fps and many of the shows are shot in film (24fps) then 3:2 pulldown is performed to get that to 30fps (by introducing duplicate fields here there). So it is not the same as just speeding up a film projector. If you could speed up your VCR to output 32fps, the TV wouldn't be able to play it. So they must either be dropping entire frames from the 29.97fps telecined video, or dropping out some of the duplicate fields (which I think would look much better. Of course it could be done during the 3:2 pulldown.

[AngusMacGyver]

I have a similar question:
How can they convert from NTSC to PAL or vice-versa on the fly (live events)? The conversion methods I know are by accelerating or slowing down video and audio. Maybe there is an interpolate method to create or remove frames.
Does anybody know it?

[Cornucopia]

I don't know if they start with tape and run it through some buffers, or whether they use a file as a source (simpler IMO), but they just knock off 1 or more Frames per second. So 29.97 -->28.97, but running at 29.97fps.

Ex. 30 minute show
at 29.97 fps = 30 minutes. Normal speed
at 28.97 fps = 29 minutes. 3 1/2 % speedup
at 27.97 fps = 28 minutes. 7.15 % speedup
at 26.97 fps = 27 minutes. 11.1 % speedup
at 25.97 fps = 26 minutes. 15.38 % speedup
at 24.97 fps = 25 minutes. 20 % speedup

I don't think they would do more than that.

Unless they have some really sophisticated and $$$$ machinery, they would have to do it with whole frames and not single fields, even though fields would keep the motion smoother.

BTW, audio would be buffered and processed separately, using standard TimeCompression/PitchShift algorithms, and delayed to match sync with the video (probably via timecode).

(Just a guess, but a pretty educated one)

Scott

[vhelp]

I'm going w/ "take out frames/fields" because this answers the questions
on why many IVTC methods fail when source is a capture of a TV movie
or show.
.
.
You see, when you capture such a show, and KNOW for sure that it was
telecined or filmed, and then you apply your IVTC method on it. If yours
fail or produces poorer results, this is the reason. Here is another example.
Last year, I recorded to miniDV tape, Star Wars. And, after FW'ing it to my
hd, and encoding it, I could not get a "locked" in IVTC on it, if we assume
that this source if true Film (which it is) then we should get a "locked" IVTC
on it. The problem was, that every so often, a frame would hickup or studder
or whatever, throwoing off the IVTC process.

The above is a very good explaination (cause) for IVTC to fail. Mind you,
when I say "IVTC fail" I'm operating under the assumption that the source
is true film (and that you know this to be true) and as such, when you do
a clean IVTC (not those methods that mix/match fields/frames or deletes
etc etc) but those that do an exact IVTC (aka, 3:2) these should pass w/
flying colors and w/out any hitch what-so-ever. Perfect example of a TRUE
telecine/film source is DVD's.

In my opinion, this clarifys my past issues.

Thanks guys for this eye-opening topic !!
-vhelp 2396

[BJ_M]

much of what you see -- there is no pulldown as the type you expect..

its done during the film scan -- and not the expected pattern ...

if its played back off a video broadcast server or off tape -- the speed is already compressed .. this is done ussually by just running it through a avid for time compression (not by dropping frames).. often the supplier/distro would pre-supply it this way to the station ..

[Cornucopia]

Equally valid answer BJ_M (coming from a more valid source).

BTW, have you seen the new Time Compress/Expand feature in Composer Adrenaline? Night and Day compared to the old (bad) framedrop/interpolation/duplication methods!

Going in for a bit of the old Ultraviolence these days, I see...

Scott

[BJ_M]

yes -- its pretty good , still not great for 70mm film work or top notch 35 (IMO) , Phoenix Tools TimeMaster or Shake (with the right plug-ins) can do a better job -- time consuming proccess though, much faster on avid

[lemmy999]

much of what you see -- there is no pulldown as the type you expect..

its done during the film scan -- and not the expected pattern ...

if its played back off a video broadcast server or off tape -- the speed is already compressed .. this is done ussually by just running it through a avid for time compression (not by dropping frames).. often the supplier/distro would pre-supply it this way to the station ..

so are you saying that it happens during the conversion from 24fps filem to 29.97fps video? So instead of doing a standard 2:3 pulldown they are doing something else? So I guess if you captured this video, you wouldn't see the 3 progressive, 2 interlaced, 3 progressive, 2 interlaced.... pattern?

[W_Eagle]

I'm going w/ "take out frames/fields" because this answers the questions
on why many IVTC methods fail when source is a capture of a TV movie
or show.

What is IVTC and what is it used for?

[lemmy999]

Inverse Telecine.

From the Glossary section of this website:

Inverse telecine (IVTC) is when a codec takes a 29.97 frames per second interlaced NTSC video that has gone through the telecine process and reconstructs the original 24 frames per second progressive FILM video.

[lchiu7]

I have a similar question:
How can they convert from NTSC to PAL or vice-versa on the fly (live events)? The conversion methods I know are by accelerating or slowing down video and audio. Maybe there is an interpolate method to create or remove frames.
Does anybody know it?

I don't know specifically how they do it, but doing it in realtime can't be hard. I used to have a box that did NTSC to PAL conversion on the fly - feed in NTSC and out came PAL and it was in realtime. Box cost me about US$250. not much use now with VCR's and TV's having multistandard being common

[W_Eagle]

Inverse telecine (IVTC) is when a codec takes a 29.97 frames per second interlaced NTSC video that has gone through the telecine process and reconstructs the original 24 frames per second progressive FILM video.

What is the use, or benefit, of IVTC? What's wrong with watching movie in NTSC?

[lemmy999]

Video shot in 24fps is converted to 29.97fps using the telecine process. This introduces repeated fields into the video. This is done so that it will play on our 29.97fps NTSC TVs. When you capture 29.97fps video that has had the telecine process performed on it and you want to save it to DVD or SVCD, a) you would like to have progressive video on the disc so that it plays nicely on your computer, and b) there is no need wasting space with the repeated fields. IVTC will convert it back to 24fps. When you put the resulting video on a DVD, the appropriate 3:2 pulldown flags are set so that the DVD player will insert the extra fields for playback on an NTSC 29.97fps TV. And when played on your computer, you have a progressive source that doesn't have the comb effect associated with motion in an interlaced source played on a progressive computer screen.