RedOne, Generators and Flicker

Matthew Parnell · June 1, 2008

I am currently 3rd electric on a feature film shooting in Australia using the RED ONE camera.

The production is shooting 24fps at 173 degree shutter.

Power in Australia is typically 240v, 50hz.

About three days into the shoot, both our DP and the productions DI Colourist(who as supervising the first few days rushes) started to notice a minor flickering being produced whilst shooting. The flickering was barely visible outside of the DI Suite and only apparent on set on the waveform monitor. A new camera body was brought in, but it has the same problems.

The flickering was not apparent when using flicker free devices such as F.F. HMIs and kinoflos, but when using smaller tungsten units (5k and under).

In the process of testing we moved from our 40kva generator(which has done countless films, tv commercials and music videos on both Film and HD) to house power, which immediately solved the problem.

We immediately tested our generator and all seemed fine on the surface. The generator was running at 240v and exactly 50hz. We proceeded to swap over all our mains, distro and the generator outlets. The problem still existed.

We then started using another generator, and the problem still existed, but slightly less apparent than previously.

The original generator was then fully workshop tested. All that the generator techs could find is a variance of +/- 2 volts and replaced the AVR unit to minimize this(which it did down to a .5-1v variance, which is typical in most generators). We tested again, and again the problem still existed.

In a number of phonecalls in recent days the Gaffer has discovered another production also in Australia that is having similar problems when using generators and the Red One. To the point of using only 5k plus tungsten units because of flicker and is not using HMIs at night, because of a ?weird pulsing? that is produced.

To me it sounds like the rules for shooting super high speed stuff.

We are getting another generator on Monday, and I will let you know the results.

What I?m wondering is if anyone else has been having these problems when running off generator power?

Is this a problem specific to running the RED ONE with generators, with 50hz generators or is there some simple solution to this problem?

Any ideas or input would be greatly appreciated.

Cheers.

Will Earl · June 1, 2008

The production is shooting 24fps at 173 degree shutter.

173 degrees? Did you mean 172.8?

Matthew Parnell · June 1, 2008

Yes 172.8 degrees.

dan brockett · June 1, 2008

Yes 172.8 degrees.

Sounds as if it "could" be a possible rolling shutter artifact? I have not shot with the RED but I am doing a lot of shooting lately with the Sony EX-1. The rolling shutter artifacts have shown up in all sorts of weird places and I do know that the RED shares the CMOS/Rolling shutter issues with the EX-1. Seems weird though. Is it possible to do some testing with different shutter angles?

Dan

Szymon Bielesz · June 5, 2008

Definitely sounds like rolling shutter artifact.

Give these two a read when you get a chance:

http://dvxuser.com/jason/CMOS-CCD/

http://www.dvxuser.com/V6/showthread.php?t...rolling+shutter

John Sprung · June 5, 2008

All that the generator techs could find is a variance of +/- 2 volts and replaced the AVR unit to minimize this(which it did down to a .5-1v variance, which is typical in most generators). We tested again, and again the problem still existed.

2 volts is more than enough to give you trouble. If the frequency of the variance is much slower than the mains frequency, say like 2 Hz or so, you can see that in person with your eyes. A just noticeable difference for human vision is roughly a change of 1%. Because of the way a change in voltage shifts the blackbody curve horizontally, the change in the visible output of an incandescent near 3200 Kelvin is more than the input voltage change. If it shows up on camera, well, that's just what was in front of the camera in the real world. The way to see that kind of variation accurately is to look at the generator output on an oscilloscope, and slow down the sweep so you can see the envelope, not just the sine wave. Did the techs do that? If they just went by an analog meter or something, the true variance might have been even more.

The thermal mass of an incandescent filament will filter out a lot of the mains frequency, but at the kind of lower frequencies you can get from loose regulation, that's not enough.

The rolling shutter could be contributing to the problem. Film camera shutters are also sort of rolling, but the differences are that they roll faster, taking maybe 45 to 60 degrees to pass over the whole frame, and the shutter edge is soft because it's between the lens and the focal plane. CMOS rolling shutter edges take the full 360 degrees to cover the frame, and they're pixel-boundary sharp.

-- J.S.

John Sprung · June 5, 2008

Yes 172.8 degrees.

That's the right number:

(172.8/360) x (1/24) = 1/50

-- J.S.

Phil Rhodes · June 6, 2008

Course, if you can get it to flicker at the right (rather high) frequency, you can get bands of light and dark down the frame, or have it brighter at the top than the bottom...

Oh, rolling shutters are great.

P

Matthew Parnell · June 6, 2008

We have now tested on two seperate generators. Both were synced to 50hz both using the meters on the genies and a flicker meter on the lamps and also by winding up and down the frequency of the generator while watching the waveform of the monitor.

Unfortunately both generators controller units would not change the frequency down to 48hz or 60hz so we could not do a 180degree test in the hope of achieving a wider window of exposure. That is one test for another time i think.

Our theory is, that while house power is essentially as close to perfect as possible, by the sheer nature of a generator it is constantly 'hunting' for sync. Normally, the tolerances on the wander, which are rather small are fine for standard film work(these generators have been used on stuff up to about 250fps according to the memories of the owners, without a single issue). Unfortunately because of the sensitivity of the cameras rolling shutter, in combination of the shorter window of 1/50th of a second(because we have to run at 50hz) these minimal fluctuations are outside of the tolerance of the camera.

We obviously have a lot of testing and work to go to try and find a fix to this issue, because the red camera is already quite widespread, and in a production environment that runs of generators, obviously this is a major problem.

Thanks so much to everybody. If you still have any ideas it would be great.

Cheers,

Michael Collier · June 6, 2008

Do you suspect this problem exists only in areas that are on the 50hz standard? I understand you couldn't test a 60hz with 1/48th exposure, but from what you've seen is it likely that RED users in America would see the same problem?

Matthew Parnell · June 7, 2008

From everything i have read this issue is specific to 50hz countries, and the problem only exists when running of a generator. There are other productions shooting in Australia with the same issues.

Essentially, unless you have a magnetic ballast discharge lamp running off the generator, the flicker is barely noticeable, and for TVCs or even TV production you could get away with it, just. As for a projects with an eye to going to a filmout, its quite worrying.

In terms of this issue occuring in the states, I believe that the only issue is that you cant run magnetic ballasts or electronic ballasts in silent mode. I think the problem is less likely in the states because your power allows for an ideal shooting window at 180 degrees at 24fps.

At the end of the day all my research points to this being a fundamental problem with Rolling Shutter CMOS chips and the simple nature of a generator. There are a number of Global Shutter chips in existance and in development, and hopefully RED looks at going in that direction in the future.

June 7, 2008

Hi mathew,

Have you spoken to RED about this?

timHealy · June 7, 2008

In terms of this issue occuring in the states, I believe that the only issue is that you cant run magnetic ballasts or electronic ballasts in silent mode.

Hey Matt,

I think you may have mistyped, but there are no silent mode settings on magnetic ballasts.

Best

Tim

Matthew Parnell · June 9, 2008

Tim, 'magnetic ballasts' or 'electronic ballasts in silent mode' is what i meant. Sorry if i didnt make that clear. My bad.

Stephen, Im in two minds as to approach red about this issue or not. Really, now that we have made it clear that it is not a problem being caused by our department we have handed the ball back to the camera department to sort out with Red and the camera house. However, considering the fact that quite a lot of money was spent getting full generator diagnotics done, new AVRs fitted, and second generators being hired, and the fact that we're going to have to face up to these problems no doubt again shortly, I would really like to see what the true resolution to this problem is, so i might write a couple emails in the next couple days once I have finished post work on the truck (we wrapped shooting on friday).

Cheers,

Matt.

June 9, 2008

I would really like to see what the true resolution to this problem is, so i might write a couple emails in the next couple days once I have finished post work on the truck (we wrapped shooting on friday).

Cheers,

Matt.

Matt,

Please keep us posted if you do.

S

Brian Drysdale · June 9, 2008

Tim, 'magnetic ballasts' or 'electronic ballasts in silent mode' is what i meant. Sorry if i didnt make that clear. My bad.

Stephen, Im in two minds as to approach red about this issue or not. Really, now that we have made it clear that it is not a problem being caused by our department we have handed the ball back to the camera department to sort out with Red and the camera house. However, considering the fact that quite a lot of money was spent getting full generator diagnotics done, new AVRs fitted, and second generators being hired, and the fact that we're going to have to face up to these problems no doubt again shortly, I would really like to see what the true resolution to this problem is, so i might write a couple emails in the next couple days once I have finished post work on the truck (we wrapped shooting on friday).

Cheers,

Matt.

People from RED do monitor this forum. Although, I expect RED ONE camera people on REDUser would be interested in this issue. The discussion there tends to be more practical these days.