Tungsten Flicker

[Adrian T.]

I wonder if this will be a problem with Dedo lights? They are DC powered.

DC is never a problem. In fact, I've shot several times with my Dedos with no flicker at all.

 

[David Mullen ASC]

Christopher, you're living in a 50 Hz country, so did you shoot with a 1/50th shutter? 24 fps at 180 degrees is not technically a "safe" speed for 50 Hz lights. See:

http://www.panavision.com.au/Information/Flickerfree50Hz.htm

You should either be shooting at 24 fps with a 172.8 degree shutter or 25 fps with a 180 degree shutter to get 1/50th.

Unless you are saying that you did that and still got flickering from the streetlamps.

 

[James T Mather]

He's right -

Probably low pressure sodium or some kind of discharge lamp (if it's yellow I'd guess its the former, cooler I'd guess the latter)

 

[Dj Joofa]

One more comment I may add:

The flicker power attenuates as the exposure time increases. Hence, a longer exposure will always help even if you are not operating "flicker free".

The worst cases are:

(2n + 1) / (4 * line freq).

Therefore, for 60 Hz countries, the worst cases lie at exposure times of:

1/240, 3/240, 5/240, ...,

and for 50 Hz. countries:

1/200, 3/200, 5/200, ...,

As mentioned above the flicker power attenuates for larger exposure time. Hence, 5/240 is better than 1/240 (5/200 better than 1/200) and so on.

 

[Phil Bates]

When we rented the Phantom, we were *guaranteed* by the lighting rental house no flicker with HMIs up to 10K fps. That proved to be wrong as we had flicker even at 1000 fps. We went to tungsten 10Ks and 5Ks and had no problems. I was told by our DP that you can get "arc wander" with HMIs at that speed.

Phil
www.artbeats.com

 

[ChristopherKenworthy]

Christopher, you're living in a 50 Hz country, so did you shoot with a 1/50th shutter? 24 fps at 180 degrees is not technically a "safe" speed for 50 Hz lights. See:

http://www.panavision.com.au/Information/Flickerfree50Hz.htm

You should either be shooting at 24 fps with a 172.8 degree shutter or 25 fps with a 180 degree shutter to get 1/50th.

Unless you are saying that you did that and still got flickering from the streetlamps.

Thanks David. That's probably the answer. I wanted to use 1/50th all the way through, so set it at that and forgot about the implications. I'll experiment again. Cheers.

 

[Joe Carney]

DC is never a problem. In fact, I've shot several times with my Dedos with no flicker at all.

Thanks Big L. I really like the look of tungsten and would hate to think I couldn't use it with Red. (Get the blue channel fixed please...).

 

[Dj Joofa]

See:

http://www.panavision.com.au/Information/Flickerfree50Hz.htm

David, thanks for the link. However, I think that chart is derived for exposure with a film type shutter. For a rolling shutter digital camera, such as Red One, the situation is more complicated, as each row is not integrated at the same time. Therefore, I do not think all exposure times will work for all those frames rates mentioned as flicker free.

Somebody also provided the URL for 60 HZ:

http://www.panavision.com.au/Information/Flickerfree60Hz.htm

We shall ignore the effect of any blanking (say vertical blanking) for this illustration. As I mentioned before, for a rolling shutter sensor, the "flicker free" exposure time is:

n / (2 * line freq).

And, for the worst case flicker the formula is:

(2n + 1) / ( 4 * line freq),

so lets say we pick two exposure times for the 24 fps. Putting n = 1 in the both equations, and assuming US based 60 Hz line freq, we get:

1/120 seconds exposure time for flicker free = 72 degrees shutter at 24 fps, and,

3/240 seconds exposure time for worst flicker = 108 degrees shutter at 24 fps.

According to the chart for the 60 HZ link any shutter angle is fine for 24 fps, and hence, 72 and 108 degrees will have no flicker. That may be true for a film camera.

But, for a rolling shutter digital camera 1/120 seconds exposure time at 24 fps will be "flicker free" and 3/240 seconds exposure time at 24 fps will be bad flicker.

 

[Mitch Deoudes]

As I mentioned before, for a rolling shutter sensor, the "flicker free" exposure time is:

n / (2 * line freq).

And, for the worst case flicker the formula is:

(2n + 1) / ( 4 * line freq)

Could you explain the math behind these formulas? My inner nerd needs the dirt.

(I'm noting that the second formula gives 1/48th as a "worst case" for 60Hz, as well.)

 

[Dj Joofa]

Could you explain the math behind these formulas? My inner nerd needs the dirt.

(I'm noting that the second formula gives 1/48th as a "worst case" for 60Hz, as well.)

Assuming the sinusoidal input of the form C * sin (2*pi*f*t), where C is a constant, f is the line frequency, and t is the time, the light intensity seen by the rolling shutter sensor can be found by integrating the power intensity w.r.t time, and you will end up with a complicated formula with a number of terms, including the sinc (x) = sin x / x, function.

The zeros of sinc function are at integral multiples of pi, i.e., at n / (2 * f), and the max. values are at (2n + 1) / (4 * f).

 

[mezmo]

We've had a few questions lately from customers seeing flicker with tungsten sources especially at higher frame rates where you can't sync the shutter to 50/60Hz.

Tungsten sources can flicker because they aren't truly continuous sources.
Most tungsten sources don't appear to flicker because their decay from on to off is slow enough that it's not visible. With some sources (usually regular incandescent) the filament will decay faster and cause flicker (smaller filament and faster cooling).

HMI's with electronic ballasts, continuous (not pulsed) LEDs sources, or DC powered tungsten are the best ways to make sure you don't get flicker.

This side benefit is that HMI and LED sources balanced to daylight will provide a much nicer blue channel whereas with tungsten, you're under exposing blue significanly.

Deanan

Hi Deanan,
What do I do if I'm shooting in an airport with several thousand
magnetic ballast warm white delux (3200K) tubes banging away
up in the ceiling on a ten hour night shoot?
Can't turn them off cause they are all in shot, can't change them to daylight
for budget/political reasons. Any Ideas?
Mezmo

 

[Ariana]

But, for a rolling shutter digital camera 1/120 seconds exposure time at 24 fps will be "flicker free" and 3/240 seconds exposure time at 24 fps will be bad flicker.

How would you calculate the same rates for film with a shutter that is wiping across the frame from left to right or top to bottom?

 

[Deanan]

Hi Deanan,
What do I do if I'm shooting in an airport with several thousand
magnetic ballast warm white delux (3200K) tubes banging away
up in the ceiling on a ten hour night shoot?
Can't turn them off cause they are all in shot, can't change them to daylight
for budget/political reasons. Any Ideas?
Mezmo

The initial note was referring specifically to higher frame rates, smaller shutter angles, and some tungsten lights whose filaments cool fast enough to cause some flicker. Shooting at 1/60th or 144degrees (or 72deg) would work fine. We're by no means saying that shooting tungsten is not possible, only that some light can cause flicker even though most don't. Likewise, shooting tungsten color temp lights is not bad, just that daylight is more optimal if you intend to white balance back to daylight.

 

[Dj Joofa]

How would you calculate the same rates for film with a shutter that is wiping across the frame from left to right or top to bottom?

Firstly, flicker can manifest itself as inter-frame (between frames) and intra-frame (within the same frame). It appears the film world is concerned more about inter-frame flicker. If you notice the flicker free frame rates for all shutter angles for 60/50 Hz, at the following Panavision links:

http://www.panavision.com.au/Information/Flickerfree60Hz.htm

http://www.panavision.com.au/Information/Flickerfree50Hz.htm

You will see that they may be calculated by taking the formula I gave as "flicker free" exposure time (n / (2 * f)) for a rolling shutter digital sensor as the frame time for a film camera. Hence, if you take exposure time as frame time and invert the formula you get frame rate as:

flicker free frame rate (for the film world) = (2 * f) / n.

For 60 Hz countries, you would get the following flicker free frame rates:
120/1 = 120,
120/2 = 60,
120/3 = 40,
120/4 = 30,
120/5 = 24,
120/6 = 20,
120/7 = 17.143, ...

And, for 50 Hz countries you will get:
100/1 = 100,
100/2 = 50,
100/3 = 33.33,
100/4 = 25,
100/5 = 20,
100/6 =16.67, ...

I.e, the exact same rate as available on the links on the Panavision links provided above.

Now coming back to intra-frame flicker (within frame) effect for the film camera. I don't see anything regarding that on the Panavision links above.

For a rolling shutter digital sensor, I am copying an image below that I posted on a different thread, which illustrates the effect of within frame "banding effect" produced by flicker if the exposure time is not calculated by the formula (n / (2 * f)):

The above image results from mismatched timing of the exposure time of a rolling shutter sensor with the line frequency.

As regards your question of if the film shutter is moving vertically downwards, then I shall expect to see the behavior as shown in the image above. And, if the shutter is moving horizontal I shall expect a behavior that we will get by rotating the above image by 90 degrees, i.e., instead of vertical banding you see horizontal banding.

 

[RobRoySyd]

We did some tests last night, under street light, and strobed like crazy, even at 24fps. We're shooting several street scenes here in a couple of weeks, so we'll need to black out all the street lights and light from scratch. (I always planned to light from scratch, but didn't realise we'd have to black out like this.) No idea what kind of light-source is in an Australian street lamp, but this is probably going to be an ongoing issue for us.

They'll be iron ballasted mercury or sodium vapor lamps most likely. They flicker like crazy at 100Hz. Iron ballasted fluros aren't quite as bad as the phosphor has some afterglow so it doesn't stop completely emitting light when the current crosses zero.

 

[Dj Joofa]

They'll be iron ballasted mercury or sodium vapor lamps most likely. They flicker like crazy at 100Hz. Iron ballasted fluros aren't quite as bad as the phosphor has some afterglow so it doesn't stop completely emitting light when the current crosses zero.

Has anybody done a CCD-based camera vs. CMOS-based camera at the same set location for a comparison?

 

[mezmo]

The initial note was referring specifically to higher frame rates, smaller shutter angles, and some tungsten lights whose filaments cool fast enough to cause some flicker. Shooting at 1/60th or 144degrees (or 72deg) would work fine. We're by no means saying that shooting tungsten is not possible, only that some light can cause flicker even though most don't. Likewise, shooting tungsten color temp lights is not bad, just that daylight is more optimal if you intend to white balance back to daylight.

Thanks for that.......Mezmo