It has to be more difficult than this...
November 25, 2007 8:17 PM   Subscribe

How do the editors/producers of music videos achieve the "slow motion" effect as seen here in James Blunt's video?

Is it as simple as I think it is (and this person, too)? Just record video it at a slower speed and match the audio? And really, what about Coldplay's video for The Scientist? Did he really lip sync it backwards and it slow motion? There's no fancy-schmancy CGI stuff to make all of this happen?
posted by cdmwebs to Technology (14 answers total) 2 users marked this as a favorite
 

Hmm... I'm no expert, but in theory since they know how fast they're over-cranking the camera, they know what percentage it's being slowed down. The on set audio playback could be slowed down the same percentage and the musician could mimic the words as slow as the recording. The takes with the best fit could be matched up in post.
posted by sharkfu at 8:33 PM on November 25, 2007


The same question is asked here, but doesn't seem conclusive.

sharkfu: wouldn't he need to sing faster if the video is being slowed down, or am I missing something?
posted by null terminated at 8:36 PM on November 25, 2007


I have no actual knowledge of this area, but here is my guess: I think that both of the videos you link to take advantage of our tendency to "reverse lip-read." That is, given an audio track and a person (or something else) with moving lips, we tend to read them as more synchronized than they actually are. Think about cartoons like the Simpsons and live-action video with 'talking animals' like Mr. Ed. or Lancelot Link.

You'll notice that, in the James Blunt video, they never show him singing more than a few words at a time. My guess is that they found short phrases that didn't match up exactly, but were close enough that they seem synchronized in real time.

In the Coldplay video, in a lot of the shots if you look very closely the lip movements and audio track don't actually sync up particularly well at all. Also, there seems to be some very creative editing that allows them to switch back and forth between shots that are 'forwards' (and well-synchronized) and 'backwards' (not so well synchronized).

I'd be interested in what a really good lipreader would think of this.
posted by googly at 8:36 PM on November 25, 2007


Best answer: I just looked at the Coldplay video again, and I'm even more convinced that it is just good editing between the following kinds of shots:

(1) Regular real-time shots that have no external clues to the time-sequence (example: the initial close-up of him singing on the bed).

(2) Shots that appear "backwards" because of external clues that could be easily manipulated through careful planning (example: shots of the singer walking down the street while everyone is walking 'backwards').

(3) Shots that are clearly filmed in reverse, but generally are long shots that don't allow you to closely observe the singer's lips and/or don't actually contain singing at all (example: jumping backwards over the wall).

Seamless editing encourages us to assume the simplest explanation: that the singer is moving through the world 'backwards' in all the shots; rather than the more complicated but accurate one: that these are a bunch of different shots on different timelines woven together.
posted by googly at 8:49 PM on November 25, 2007


Well back in the analogue days when this famous surf movie was made, they used a triple print technique, - get three prints of the film and splice the three copies of each frame together to get the film slowed down by three times. I guess that similar techniques are used in digital production.
posted by singingfish at 8:50 PM on November 25, 2007


Well, I'm a TV engineer, not an editor, but it kinda looks to me that they are intentionally dropping frames to make it look slow when it's not.

That's just a guess.
posted by Argyle at 9:01 PM on November 25, 2007


Before the Scientist, a band called Electracy used a similar effect for a video to their song Morning Afterglow. Not sure if it is available on the web, but the CD single had a great "Making Of" video that showed them playing the song backwards during filming so that the singer could lip sync to it. Then they did weird stuff like rigging a Polaroid camera to suck a picture in, so that when they reversed it it would look normal, and you had a mix of reversed and normal things going on at the same time.

Phew - so in essence, yes, in that one he just lip-synced to the song in reverse!
posted by csg77 at 9:02 PM on November 25, 2007


If you watch a Making-Of with the Weird Al "White and Nerdy", there are a few sections where you can see (real-time) Al lip-syncing to a much sloooowwweeerr vvvveeerrrsssiiiooonnn oooooffff ttthhhhheeee sooonnngggg. This 'results' in a video where everything seems speedier, and where the lip-syncing matches.

If I had to make the James Blunt effect on a cheap budget, I'd do the opposite - lip sync the song at double time and then slow the video by 50% (yadda yadda math), overlay the normal song, and voila!
posted by unixrat at 9:18 PM on November 25, 2007


Argyle's got it. It's at normal speed, but they are dropping frames so it looks to us like it's in slo-mo.
posted by miles1972 at 10:06 PM on November 25, 2007


I don't know if they did it for James Blunt's video, but I remember watching a behind the scenes feature on *cough* Celine Dion's "My Heart Will Go On" video; they sped up the audio double-time, recorded her at regular speed lip-syncing to the chipmunked version, then cut the video speed in half for playback, bringing the audio to normal speed. As long as she's able to keep up with the super-fast version, her lips will match in the final video.
posted by sarahsynonymous at 10:59 PM on November 25, 2007


Best answer: The Blunt video isn't dropping frames. "Dropping" frames would not result in a slomo effect, and in fact the film would look faster if that were the case. The only other practical methods would be to slow down (in post) film which was shot at normal speed, which would either get you a "strobey" look (where frames are repeated in order to get a slowed down look, which is certainly not the case in this video) or shoot it to begin with overcranked, which will give you discrete, non-repeated, "slo mo" frames upon playback at 24fps or 29.97fps. If you step through the video frame by frame, you will see that there are no repeated frames. This is a clear indication of an overcranked shot.

This means that Mr. Blunt did indeed have to lipsync his song at chipmunk speed in order to get proper sync at normal playback speed.

googly has the rest of the equation correct. It's all editing and pre-planning. One of the secrets of VFX is that we do a LOT more non-CGI sleight of hand than the end viewer might expect. If you sprinkle sporadic "hero" shots (i.e. shots where you're supposed to be paying attention to the effect and/or talent) within the video, you can get away with "faking it" in the connecting shots. Because the wider shots are meant to be taken in by the viewer as a whole, Mr. Blunt didnt have to be quite so exacting in his chipmunk lipsyncing. This relieves the (ahem) "talent" from getting it perfect for every single shot, and makes our lives--as VFX post dorks--easier, as we dont have to spend as much time finessing those shots.
posted by melorama at 11:39 PM on November 25, 2007


Best answer: I edit music videos. I've done slow and reverse motion.

It is as simple as you think. Before we shoot we make a CD of the music that is sped down by a nice divisible amount (normally doubled) or reversed. The artist then learns to lip-sync the lyrics faster, or reverse.

Working in PAL 25fps, we then shoot with a camera such as the Panasonic Varicam, or on 35mm film at 50fps. When we process that as 25fps footage we get a 50% slow-down and when the artist's lipsync at doubletime is slowed down it fits in with the original track.

For reverse, it's the same thing - but we play the footage in reverse.

The other great example of this sort of trickery is the Nickelback video (can't remember the title, and don't want to wade through YouTube) where the camera moves through the crowd with the band on stage, but the crowd is moving in reverse. This was shot with a motion control system. A clean plate shot of the band perfoming, then the motion control reversed and the crowd do their thing in the foreground with no band on the stage. The audience footage is reversed, the movement matches the performance exactly, and the band are 'comped' back into the shot.

The biggest challenge of these types of effects is that we (as humans) have a really strong sense of 'lip sync' and we can see when something's not right. That's hard enough in a normal music video, but even harder when the artists have to perform off tempo, or in reverse. It's difficult to accurately reproduce the right movements to get it to look convincing. That is why these types of videos are often largely in quite wide shots, where we are not going to see the lips so clearly. To get around this, as editors we rely on cutting to shots where the lip sync is better, or no lips are visible, or something else. Which makes long (or single) take videos with these techniques really difficult.
posted by sycophant at 12:57 AM on November 26, 2007 [1 favorite]


Response by poster: I've always wondered about this and now, thanks to Ask MeFi, I know. Great answers everyone!
posted by cdmwebs at 6:50 AM on November 26, 2007


I've wondered about this, too... I'm in the process of shooting and editing my own very homemade video and have been practicing lipsyncing at double speed to be able to create my own slo-mo effect. I was feeling kind of silly and amateurish doing it, but now... well, I still feel silly and amateurish, but at least I feel more confident that I'll get the results I'm after.
posted by myrrh at 5:45 PM on November 27, 2007


« Older Help me find a simple speaker solution!   |   No woman is a good woman? Newer »
This thread is closed to new comments.