One long continuous shot, sped-up until it becomes "live"

Question

I have an idea for a video project (that I will need to commission a studio to produce for me), but at this early stage I don't even know whether my idea is feasible.

The idea involves turning (hours of) continuous/uninterrupted footage from a single camera into a short (circa half-hour) film by varying the speed of playback so as to speed over "boring" sections whilst still drawing attention to key moments of interest: that is, some parts might be sped up by as much as 64x, whilst other parts could be slowed right down (even to a frozen frame). Is this sort of sped-up/slowed-down effect from a single continuous shot known by some particular name?

Where this idea gets more complicated is that playback is to begin whilst the footage is still being shot, becoming real-time/live once it catches up to the present. Is this sort of transition from recorded footage to live broadcast known by some particular name?

Practicalities

I envisage the recorded footage being transferred from camera to editor in "chunks" (e.g. initially of a half-hour each, reducing in duration toward the end until the feed is live). Each such chunk could then be worked on to produce a (shorter) "chunk" of final film that is immediately forwarded to the presentation display and enqueued for playback.

When it is time for playback to begin, the enqueued chunks would start playing in turn whilst further chunks continue to be added to the queue. Once the footage is live, these chunks would amount to little more than a typical video stream (so some small buffer will be required to cope with network and processing delays—but the smaller the delay, the better).

I appreciate that (in order for the different sections of film to playback smoothly and comfortably, without any judder/strobing/etc, yet have sufficient definition to be interesting), various camera settings such as shutter speed will need to be varied according to how that particular section of footage is to be played-back.

I also understand that some computationally expensive interpolation may be required to smooth out the transitions, and this will add delay to the editing process...

Questions

My principal question is whether this idea is feasible, or completely unrealistic? Perhaps the time that would be required to edit the footage would alone make such a project impossible?

If it is feasible, then I'd like to understand the key risks to the project from the videographer's perspective (aside from network/comms failure between camera/editor/presentation). What other challenges do you foresee, and how might one tackle them?

I've tried contacting a few professional studios about this idea, but so far haven't had much response. I've probably approached the wrong sorts of places, who perhaps don't have the resources that would be necessary to turn such a crazy idea into reality. What sorts of studios should I be looking for, and what capabilities should I ensure they have?

Answer 1

The answer above answers half of your question, and yes, it is trivial for anybody who has used a non-linear editor (NLE) such as Premiere Pro, Final Cut, Resolve, etc., to do speed ramp effects. The film Koyaanisqatsi features many visual effects that include speeding up and slowing down time, which may help you better imagine the possibilities of the medium. But NLEs are really only for off-line production.

To answer your second question about doing this while actually shooting, the short answer is no, nobody has built an off-the-shelf system for doing this. The concept of "instant replay" (encompassed in a system such as the NewTek 3Play 425) illustrates how live feeds and recently (immediately) recorded clips can be used together while slowing down the playback rate. And while it's theoretically possible to speed up playback rate (subject to available content in some buffer), it's not anything anybody has tried to do. Instead, people edit offline and present a finished product after editing.

Answer 2

"The idea involves turning (hours of) continuous/uninterrupted footage from a single camera into a short (circa half-hour) film by varying the speed of playback so as to speed over "boring" sections whilst still drawing attention to key moments of interest: that is, some parts might be sped up by as much as 64x, whilst other parts could be slowed right down (even to a frozen frame). Is this sort of sped-up/slowed-down effect from a single continuous shot known by some particular name?".

It's called "Speed Ramp", assuming it's a smooth transition and not a 'jump cut'. See "ARRI Ramp Tutorial - David E. Elkins" (.PDF) or ARRIFlex D-21, but it only goes from 1 to 60 FPS.

In summary: Continuous long shot video, with duplicate or low activity, frames deleted; alternatively, manually control the frame rate with smooth ramping.

• Use AVISynth on a Laptop with a Camera tethered.

• Use a "Duplicate Frame Detector" or Fizick's "GetDupe" and "MvTools V2" to detect motion more precisely and in specific portions of the frame.

The output can be a smooth video of single frame rate from variable frame rate input.

Technical Note: It goes without saying: If you speed up 64x and there's one boring hour then either it will show in less than a minute or take 64 hours to create one boring hour - whichever way, AVISynth can automate the process. If you want a half hour output and there's nothing interesting, so it's sped up 64x, then from the time you start shooting until the time you can go live will be (30*64)-30 minutes ; a 31.5 hour delay, then a half hour video can be streamed live. Of course that's a worst case scenario.

Create one continuous video or chunks that you can join. Whatever mashing you want can be done in AVISynth either directly or with one (or several) of the 1000's of Plugins; if by chance there isn't precisely what you want then the source code is available and you can create a new effect.

There are 10,000's of experts but it's easy enough to do yourself; since you're willing to pay finding someone is all that much easier. There are dozens of Forums where people will help for free.

For a 'hardware only solution' consider a 'motion detecting camera', that will shoot a series of stills at a regular rate while there's motion and slow to a preset value when there's no activity (depending upon the camera and the settings). That will give you 'jump cut video' (no smooth ramping) that you'll need to edit before you can go 'almost live'.