Shutter speed advice needed.

I need to shoot some stunts and want to do them in regular time, but show them speeded up. I'm not thrilled with results I get from just speeding up the shot in FCP, so I'd like to try adjusting the shutter speed of my Sony VX2100. The default speed is 60. I can set it anywhere from 4 to 10,000. The main action is a guy being thrown through the air due to being hit by a truck. I'll probably be moving the camera, not the actor. I'm working in FCP 7, Mac Pro 1.1, Snow Leopard 10.6.4. Any advice greatly appreciated.

I think you should go at 1/30 for 200% speed ramp.



www.strypesinpost.com

The shutter speed setting on the Sony VX2100 is really just a sampling setting; it has no impact on the speed of the frame capture. As this camcorder doesn't have an adjustable frame rate, is there anything I can do to get a better capture for speeding up in FCP? Thanks.

Rent a proper camera for the day?

I don't mean to be rude, but if you can't shoot off-speed, then you cannot shoot off-speed. It's really as simple as that.

Jeff Harrell Wrote:
-------------------------------------------------------
> Rent a proper camera for the day?
>
> I don't mean to be rude, but if you can't shoot
> off-speed, then you cannot shoot off-speed. It's
> really as simple as that.


Blunt isn't necessarily rude, but I think you missed the thrust of my query. What I'd like to know is whether there is anything I can do to get a capture that is optimal for speeding up in FCP. For example, if I was shooting with film and wanted a very slow motion shot, I'd pump all sort of extra light into the shot. Perhaps there's a similar method for adjusting the adjustable settings on my VX2100 in order to end up with a better looking shot when speeded up 200% or more. If so, I'd like to know. If not, so be it.

You could try adding motion blur. I like the blur from the foundry's furnace core package. The speed ramp tools are pretty good too.



www.strypesinpost.com

Yeah, I see what you mean, and I apologize for the bluntness. I also apologize in advance for what's going to sound like pedantry, but this is a public forum, so I feel obliged to go into a little background.

What you're talking about is shooting off-speed. You run the camera at a higher frame rate, then play the footage back at 24 as normal, and the result is slow motion.

When you shoot off-speed, you keep the shutter angle in the camera locked to 180°. The shutter angle is a way of talking about shutter speed as a fraction of frame rate. The historical reason is that film-camera shutters are metal leaves that could be adjusted to cover a fraction of a circle. The shutter spun in sync with the camera's motor drive, so having the shutter set to cover half of the circle ? 180°, in other words ? would let light into the camera for half the time that the film was in the gate. At the standard frame rate of 24 fps, your shutter would be open for 1/48th of a second. But if you run off-speed, your shutter remains open for half of the time the film's in the gate: 1/200th at 100 fps, 1/160th at 80 fps, whatever. That's why we talk in terms of shutter angles, because it disconnects the relative exposure time from the frame rate.

When you shoot off-speed, what you want is to get a motion quality that matches what you'd get if you'd been filming at normal speed. Imagine your action was literally happening at half normal speed, and you just filmed it as usual. Your frame rate would be 24, and your shutter angle would be 180°, so your exposure time would be 1/48th. Since action does not happen at half normal speed, you double the frame rate but keep the shutter angle constant. So you record at 48, with a resulting exposure time of 1/96th, but when you play back it looks like you were shooting at 24 with an exposure of 1/48th.

In a perfect world ? and this is what I meant by suggesting you rent a proper camera for a day to shoot your off-speed stuff ? you'd do exactly that, only with a digital sensor instead of film and a rotating shutter. Shoot at whatever frame rate works, keep the exposure locked to one-over-the-frame-rate-times-two, then when you play back, everything is normal.

The reason you need to pour more light onto the set when shooting at high speed is because you're exposing each frame for less time. If the amount of light coming into the lens is constant, each time you halve the exposure time, you're cutting your exposure by one stop. You could open up your aperture by one stop ? go from f/2.8 to f/2, for instance ? but that would change your depth of field, and just as you keep your shutter angle constant to preserve the motion quality, you keep the aperture constant to preserve the depth of field, so your off-speed shots look like all the rest of your shots. So the only solution is to double the amount of light on your subject.

So these are the properties of a shot:

• motion quality (i.e., how the motion of the shot feels, mostly determined by motion blur)
• overall exposure
• depth of field

And these are the things we can directly change:

• frame rate
• shutter angle
• lens aperture
• lighting

You generally want the things in the first list to remain constant. (Well, you don't so much want depth of field to be constant across an entire show, but you generally want a particular depth of field for a particular shot.) But the things in the second list are related to the things in the first list in a fairly complex way; changing one of the things in the second list will change at least one of the things in the first list. In order to keep the things in the first list constant (or at least controlled) you must change at least two of the things in the second list.

For example, if you change the frame rate while keeping shutter angle, aperture and lighting constant, you will change the overall exposure. Specifically, you'll make everything darker, 'cause you're capturing less light in each frame. If you compensate for this by changing the lens aperture, you can bring your exposure back into line, but at the cost of changing your depth of field. If you compensate by changing the shutter angle, you can also bring the exposure back (within limits), but at the cost of changing the motion quality. In order to change the frame rate while keeping all those things in the first list constant, you must change the lighting.

On the other hand, if you keep the frame rate constant ? as your camera apparently requires you to ? then you can't change any of the other things without affecting those things in the first list. If you change the shutter angle, you'll affect motion quality and exposure. If you change the aperture to compensate, you'll change motion quality and depth of field. If you change the lighting instead, you'll keep the same depth of field, but you'll still change the motion quality.

None of the changes you can make while keeping frame rate constant can help you get closer to an off-speed shot. If I were you, and had the limitations that camera appears to have, I'd let go of the idea of getting a "normal looking" slow-motion shot, and instead explore my options. For example, what happens if you shoot with a wide-open shutter ? a 360° shutter angle, or 1/24th of a second at 24 fps, which is impossible on a film camera but can be done on a video camera with an electronic shutter ? and then step-print your shot at 2:1? In other words, slow your shot down by half without frame-blending it, which has the net effect of showing each frame twice in succession, just like what you'd get if you step-printed film on an optical printer. That'd be a drastically different look from traditional off-speed slow-motion, but you never know, it might be a look you like.

Shoot some tests. Go nuts. Do crazy things. Shoot 1/24th at 24 (compensating by halving the amount of light in your shot, or using an ND filter) and then slow the clip down in Final Cut without frame blending to 50%. See what you think. If you hate it, no big loss, try something else.

Work within the limits of what your camera can do, tweaking those variables you can control (the second list) and seeing what effect it has on the properties of your shot (the first list). You might get inspired by what you find.

he wants to SPEED IT UP, basically shoot LESS FRAMES.

if the motion happened twice as fast, you'd get double the blur,
so if your default shutter speed is 60, try 30 as strypes suggested,
THEN speed it up to 200% in FCP.

shooting with shutter set to 30 may give you a "progressive" style image, one where each field is the same,
so it may not match your other footage.


nick

And if you are unable to change the shutter speed as you mentioned, you can try adding motion blur to the image. If you want to go at 200%, you drop frames. Or you can shoot at 15fps.



www.strypesinpost.com

also,
speeding up a moving-camera shot will amplify any wobbles.
and make them feel "un-real",
so stabilisation can play a part, too.

You might add Twixtor to your toolbox, it's a very good tool for slowing down or speeding up footage.

If you're working 'interlaced' there is a nice trick.
Create a copy of your clip and stack above the original with 50% transparency
Apply the de-interlace filter to both the original and the copy but with different field settings.
Now set the speed to 200 percent for both.
This will get you a pretty nice 'fake' motion blur.

Andreas

Some workflow tools for FCP [www.spherico.com]
TitleExchange -- juggle titles within FCS, FCPX and many other apps.
[www.spherico.com]

There are two separate problems here: irregularity of timeflow; insufficient motion blur.

Timeflow. The Sony VX2100 normally shoots interlaced video. When FCP makes 2x fast motion it probably does not discard every other field but every other frame. This leaves a sequence of fields representing:
time = 0/60 sec
time = 1/60 sec
time = 4/60 sec
time = 5/60 sec
time = 8/60 sec
time = 9/60 sec
etc.
The 0, 1, 4, 5, 8, 9 ... time series makes an unpleasant irregular timeflow.

To do better requires more sophisticated software that discards every other field's time from the original. It can't just discard every other field, because then half the lines are always missing, so it must do some deinterlacing. The result is a pleasantly regular 0, 2, 4, 6, 8, ... time series.

FCP likes to pretend there's no interlacing. FCP can't even display separate fields.

Motion blur. 2x fast motion requires 2x motion blur. It was suggested to employ 1/30 sec shutter speed in order to double the motion blur. I'd guess the 1/30 sec shutter setting in this camera makes each field from the sum of two exposures: one during that field's 1/60 sec and one during the previous field's 1/60 sec. Since the CCD readout time (for a row) is tiny, the two motion blurs adjoin to make one motion blur, like a 1/30 second exposure should. So far so good, but the interlacing can spoil it. When an object really moves 2x as fast, or equivalently when the camera really runs 2x as fast, there is no overlap in the motion blurs from field to field. However if FCP gives the time series 0, 1, 4, 5, 8, 9 ... there will be overlap between 0 and 1, none between 1 and 4, overlap between 4 and 5, none between 5 and 8, etc. That's funny looking. The more sophisticated software gives the time series 0, 2, 4, 6, 8, etc. so there is no overlapping. However more sophisticated software doesn't require shooting with the 1/30 sec shutter speed at all, since the software can double the motion blur (using the information from the discarded field times) as well as the camera can. If Andreas Kiel's suggestion does this, great!

Dennis Couzin
Berlin, Germany

To Andreas: Twixtor sounds nice.
This is software that reconstructs the reality from the original snapshots (frames), and then takes new snapshots at new times and with new shutter times of the constructed reality. My first instinct, always, is to defeat the godlike software: to create some video on which it makes hilarious artifacts.

Dennis Couzin
Berlin, Germany

Its pretty easy to make the software creating hilarious artifacts if you know how it works.

I (sometimes) do use Twixtor with AE since several years to create extreme slow motion - so just the other way round of what needs to done here. I really know what will happen at certain frames when I see the source video. But then I make just a list of the 'offending' frames, let the plugin do it's job, proof the things I expected and start to do some roto paint and roto spline for these frames to send those bunch of layers to the plugin again. The result can be exceptional good if you know what you do or know what needs to be done.
In the moment I got a project which needs a 90 minute minute clip to be rendered from 25 fps up to 500 fps at maximum and Twixtor does a perfect job on this if you do feet it with well prepared footage and parameters *. And this plugin does really show where big Intel machines with a lot of memory do shine.

In this current case the plugin 'only' has to deconstruct 2 interlaced frame into 4 full frames, create motion vectors for each pixel to construct a new frame by taking those 4 frames into account. This won't create any serious artifacts as long the scene is somehow well lit and contrast is not washed out seriously, but even then it will fall back into a kind of blending those frames and the result is still very good in most cases. So Rick should give it a try.

Andreas

*) In the current case it's a car ride. The installed on board computer samples current speed of the car every 1/100 of a second which is used to create the keyframes for the slow down parameter of Twixtor to get a constant speed for all the 90 minutes - speed is from 0 to 130 km/h. There is other stuff sampled like GPS data with a 1 cm accuracy (they say) and front tire angle which are used to pan around in the slow motion clip. All the other captured data are just matched to each frame somehow like EXIF data for a still picture.
It's kinda fun job which wouldn't have been possible some years ago since storage had been way more expensive -- we need around 24 TB which includes space for backup.

Some workflow tools for FCP [www.spherico.com]
TitleExchange -- juggle titles within FCS, FCPX and many other apps.
[www.spherico.com]

0 to 130 in 90 minutes?! No one will buy that car, Andreas - FAR too slow I like the ones that go from 0 to 300mph in under six seconds.

Quote
Jude
0 to 130 in 90 minutes?! No one will buy that car, Andreas - FAR too slow I like the ones that go from 0 to 300mph in under six seconds.

Jude, not 0 to 130 in 90 minutes . The car is able to go from 0 to 130 in 8 seconds and it does. It's about to have a 90 minute movie where the car speed is anything between 0 and 130 km/h. The challenge is to simulate different speeds with just one movie without getting a ticket for driving too slow or too fast when shooting this movie. The rendered movie will have a constant speed of 0-10 km/h and playback speed will be controlled by a kind of 'USB gas pedal'

Just an idea about the car I had two years ago to do something similar, thee new one is not that stylish and fast but has a lot more electronics installed.
pic1
pic2
This car was fast with a 500 kg carbon body and above 200 PS - though it didn't reach your preferred 300 mph, you definitively would love it

Andreas

OK you got me. That's a really cool looking beasty.

A note on the post I made above about creating a 'virtual motion blur'.

I just found a movie where I used this method to create a motion blur for a 100 to 100 speed. Was an art project. Below a frame of it. The merge time must have been 10 which meens you loose 10 frames at the beginning -- 1 frame per layer. Result had been a real nice motion blur for the running movie.
pic
Aliasing occurs because of the source codec (MPEG2)

Andreas

Some workflow tools for FCP [www.spherico.com]
TitleExchange -- juggle titles within FCS, FCPX and many other apps.
[www.spherico.com]