Frame Rate Slow mo/Speed up

[Chamber005]

So Good Will Hunting and Braveheart were not shot on video, but they both explored something I'm interested in (and neither one really succeeded in doing, IMO), which is to take an action scene (or, fight scene) and use the process of slowing down and speeding up at specific moments.

I originally was looking into this because of rolling shutter, examining that if frame rate was sped up in post if it would elminate artifacts, etc.

So the question is, how would one go about shooting a scene (which could have ADR issued in post, as there would be no dialouge) where characters are hitting one another in 24fps action. E.g. I'm wondering how I would film something where the actors are actually making brutal physical contact with one another while moving at a slower pace in real time, and then speeding that up to appear as if it's happening in traditional "fight" speed?

And to further the dilemma, assuming that it is the frame rate that needs to be altered (i.e. 8 frames to 25fps or whatever), once that piece of the video was then transferred to 35mm and projected, would everything remain in synch?

Am I over thinking this?

Let me know what you guys think!

THANKS!!!

 

[J. Eric Camp]

The biggest problem with that is that gravity won't slow down. Lets say you are filming at 12fps. Twice normal. So everyone moves at half speed to appear at normal speed when played at 24fps. The physical forces of the world won't change speed. Sweat won't get knocked off of someone's head let alone at the "right speed." Blood won't spout at the "right speed." It could make things interesting. As far as this being used to deal with "rolling shutter artifacts" I don't see how you expect it to accomplish that.

If you want to re-time peoples actions shoot the whole thing in slow-mo and retime in post.

 

[Jeff Kilgroe]

The ramping effects your describing are typically accomplished in post by shooting each shot or scene at high frame rates and then altering playback speed at the appropriate times for the right effect. ...Like J. Camp said above -- re-time in post. Ramping can be triggered in camera, but to actually time that with a human event (fight coreography) can be challenging to keep timing consistent or even somewhat accurate.

It's not uncommon for fight scene footage to be shot a bit undercranked for a small speed up in post. Extreme effects like in "300" are a combination of over/under-cranking and lots and lots of digital enhancements. But even those shots that get a speed up are often shot overcranked (as for doing slow-mo) and then re-timed to look undercranked. With good visual interpolation algorithms, there's a lot more information there to create a more realistic illusion of fast movement, as opposed to shooting at 12fps and playing back at 24fps.

You will want to experiment or test with various frame rates and shutter speeds in regards to the lighting you wish to use to see what yields the results you are after. Also consider what post tools are available to you and your budget and how that may effect your shooting choices.

 

[Chamber005]

The biggest problem with that is that gravity won't slow down. Lets say you are filming at 12fps. Twice normal. So everyone moves at half speed to appear at normal speed when played at 24fps. The physical forces of the world won't change speed. Sweat won't get knocked off of someone's head let alone at the "right speed." Blood won't spout at the "right speed." It could make things interesting. As far as this being used to deal with "rolling shutter artifacts" I don't see how you expect it to accomplish that.

If you want to re-time peoples actions shoot the whole thing in slow-mo and retime in post.

So I'm not sure if the videographer's original intent was to deal with rolling shutter (this is actually in reference to the EX1, which I'm going to be using assuming the Scarlet does not become available soon), but something is definately happening during the process here --

http://www.vimeo.com/1025962?pg=embed&sec=1025962

Because aren't some rolling shutter artifacts occurring primarily due to the frame rate? I understand CMOS reads from top to bottom so that it creates specific kinds of artifacts, but doesn't it all still come down to frame rates?

Again, I might be confused on this matter.

 

[Chamber005]

Could someone jump in on this thread?

I guess I'm trying to discern why certain artifacts exist in certain fps, while they don't exist in others. And if you used the fps wherein the artifacts do not exist, couldn't you then just speed up the fps in post and create fast moving pictures without artifacts?

 

[Jeff Kilgroe]

Considering how vimeo butchers the video, it's hard to tell just what specific artifacts or anomalies come from where.

I'm unclear as to why the video was shot just the way it was... I didn't see any motion or retiming effects to actually make it appear that fast. Other than the motion graphics work with the green aura and then the flames, it still looked like the bike never went above 40mph and that's with the claimed 6X speed-up.

I somehow doubt this was done over concerns with rolling shutter. I see lots of stuff shot on the EX1 with much higher motion and the rolling shutter or skew effect is not a serious concern. If you want to know why it was shot overcranked only to be re-timed to 6X faster than real-time in post, why not ask the guy who shot it? Or I'm guessing you have, and received no reply...

I think the whole rolling shutter concern often talked about on this site and on many others, is simply blown out of proportion. People are more often focused on issues like that rather than proper lighting or other factors that will dictate how your work really looks. It seems people often forget that the only motion picture systems out there that do not exhibit a rolling shutter effect are [most] analog tube and CCD-based video camera. CMOS sensors skew, film skews. You want to see bad rolling shutter or skew? Look at the HV20/HV30, Canon 5Dm2, Nikon D90. Good rolling shutter/skew? Sony EX series, and the RED ain't bad at all considering the size and resolution of the sensor. Skew with film is a bit different as it's caused by the motion of the mechanical shutter and potentially motion of film through the gate, depending on frame rate, shutter angle, which camera, etc..

The rolling shutter effect can be effected by exposure time. Faster exposure times will lessen the amount of skew, to a certain degree, but can also have the effect of making skew more noticeable due to having less motion blur. I'm not intimately familiar with shooting on the EX1, so can't really comment on all its shooting options. But will say that you need to pick a frame rate and shutter speed (exposure time) that is appropriate for what you are trying to accomplish. You may need to conduct some tests. Or give us a better idea of what you want to do rather than having us make wild-ass guesses about some clip on vimeo without much info.

What is it that you like, or don't like, about this motorcycle video. Or specifically, what about the final result are you trying to analyze?

 

[J. Eric Camp]

It it my understanding that the rolling shutter in the R1 is independent of the frame rate. It is always going as fast as it can.

So you could shoot it at a faster frame rate or higher shutter to lessen the amount of exposure time (motion blur) per frame, but when the signal is sent to turn off or on the photosites it always moves at the same speed.

I don't know about the ex cameras. Maybe their shutter activation rate is tied to their frame rate.