Home of Professional Cinematography since 1996

Color Phasing – Fluoro Light & F900

Published : 20th November 2003


I grabbed a quick shot last week with someone under an 8' cool white fluorescent fixture with an F900 at 23.98fps, 1/48 sec shutter.

The editor was pulling selects from a Beta SP down-conversion today and called to say that the skin color of the guy cycles from normal to bluish white and back again over a period of about 5 seconds. Since it was a quickie pick-up shot I didn't bother to use a monitor so this was a complete surprise to me.

I didn't see any flicker or exposure shifting in the viewfinder at the time.

Has anyone seen this problem before? I'm assuming that the 900 at 23.98fps with the shutter at 1/48 was out of sync with the fluoro which must have been drifting over time but can't guess why the fixture/tube was exhibiting color phasing.

Any thoughts from anyone would be appreciated.

Thanks,
Randy Miller, DP in LA



Yeap : I have seen this before but only on video equipment.

I think what caused the problem - but I haven't confirmed - was the ac was holding the camera in pause mode and was right up against a huge HMI, the lens was looking straight into the HMI - and to be technical I think either the brightness or heat mashed the ccd's causing an apparent white balance shift which seems to take about 3-5 secs to cycle through.

Now even outdoors with no additional lights the colour balance seems to cycle. Has anyone else seen this problem?

Thanks

®
Rolfe Klement
www.creativesunshine.com



>. . . I'm assuming that the 900 at 23.98fps with the shutter at 1/48 was out >of sync with the fluoro which must have been drifting over time

23.98fps is not an HMI-Safe (...or 'flo safe) frame rate. The effect you describe can be quite subtle so a lot of people seem to overlook it. This is not a problem with a light fixture.

There is probably some archived discussion on this.

David Perrault, csc



David Perrault wrote :

> 23.98fps is not an HMI-Safe (...or 'flo safe) frame rate.

Odd, I was trying to get the answer to this very question for an HD shoot w/ HMI's a week ago. No one seemed to know, even called Plus 8 in NYC and the tech there *thought* it was fine (safe). I consulted a few other gaffers who said they never saw a problem and w/o a definitive answer by shoot day I went ahead w/ it and did not notice a problem. However we did not have a very good monitor. No bad reports from editor, but I'm wondering if it shows only under *certain* conditions, what they may be I don't know.

If any one has links or additional info to read up on this I'd appreciate it.

John Roche, gaffer
NYC



Were you shooting with any electronic shutter dialled in?

Art Adams, DP
Mountain View, California - "Silicon Valley"
http://www.artadams.net/



>23.98fps is not an HMI-Safe (...or 'flo safe) frame rate…w/o a definitive >answer by shoot day I went ahead w/ it and did not notice a problem.

It's a pretty easy thing to test. Just put up a vector scope and you can see the phase changing back and forth.

David Perrault, csc



John Roche wrote :

>Odd, I was trying to get the answer to this very question for an HD shoot >w/ HMI's a week ago.

Usually the phasing effect is so slow and relatively small because the frequencies are so close together that the "beat" is very slow that makes it difficult to pick out on a monitor visually. look for something in a mid tone to mid level highlight on a wave form breathing 3-5 units or so. If you are really attentive you can sometimes see it in a VF with zebras.

In any case you have to put your eyes on the monitor for a while to see it happen.

Mark Smith
Oh Seven Films



One would think the error would be miniscule :

120/24=5
120/23.98=5.004

None of the electronic shutter speeds seem designed to put one right in the middle of an HMI safe window.

Art Adams, DP
Mountain View, California - "Silicon Valley"



>120/24=5
>120/23.98=5.004
>None of the electronic shutter speeds seem designed to put one right in >the middle of an HMI safe window.


I believe the difference is that at 24fps you do *not* get a beat frequency going. You may end up with a *very* slight density variation from take to take at 24fps BUT this would not vary within a take.

At 23.98fps you have a slight fluctuation *within* the take.

What I don't understand is why there is a phase variation, as seen on a vector scope, which indicates a colour shift back and forth at 23.98fps. I would have thought this issue would manifest itself as a density change - video level change.

David Perrault, csc



>At 23.98fps you have a slight fluctuation *within* the take.

So what's the solution? Shoot at 1/48th shutter?

Art Adams, DP
Mountain View, California - "Silicon Valley"



>At 23.98fps you have a slight fluctuation *within* the take. So what's the >solution? Shoot at 1/48th shutter?

I'd assume that 1/60th would be more ideal for 60 hz lighting.

After all, at 24 fps, a 144 degree shutter is supposed to be optimum for 60 hz, not 180 degrees. What I don't understand is if 180 degrees at 24 fps is within a safe window, why not a constant 23.976 fps?

David Mullen
Cinematographer / L.A.



>I'd assume that 1/60th would be more ideal for 60 hz lighting.


Too bad 1/30 isn't an option. I can't see why it wouldn't be... but it ain't.

Art Adams, DP
Mountain View, California - "Silicon Valley"



>At 23.98fps you have a slight fluctuation *within* the take. So what's the >solution? Shoot at 1/48th shutter?

No. The fluctuation is still there at 1/48th. You can try 1/60th but that looks a little different. Or... Shoot 24fps and transfer (conform?) at 23.98fps. This keeps post (audio) happy.

This whole issue is all sound related : if you can shoot double system then why not just shoot 24fps? Like film...

The seductive savings of having sound recorded on the HD video cassette doesn't just bring a cable headache!

I keep thinking someone is going to come up with a simple, elegant, time-code/sound solution for this issue. But since no one has...

Maybe there isn't a good solution.

David Perrault, csc



>What I don't understand is why there is a phase variation, as seen on a >vector scope, which indicates a colour shift back and forth at 23.98fps.

I don't have any direct experience with an F900 at 23.98, but I do have a couple of decades of experience with other video cameras and fluorescent lights...

I think people are discussing 2 different issues here...the "HMI Safe" question, which would as David says produce a density change...and the "weird fluorescent phase cycling" that Randy originally asked about.

On regular 60i cameras, as soon as you put a shutter in at any speed, you risk running into this problem. I don't know what causes it, but it varies considerably depending on the location - sometimes you have no problem at all, other times it's very noticeable. My personal belief is that it has something to do with the power at the site and/or the condition of the ballasts for the fluoro's and/or the fluoro tubes themselves...or some combination of the above.

Now, here's the thing...a regular video camera shooting 60i is really running at 59.94, which is off .06 from the AC...so you'd think that if "HMI Safe Speeds" applied to this fluoro problem, 23.98 would be close enough to 24 that it shouldn't be an issue.

So I wonder if 60 Hz is the ONLY safe speed to avoid this color phasing - that is, rather than being close to a traditional "Safe Speed," you actually have to be close to 60? I've shot 30p on my IMX camera, which would really be 29.97, and haven't run into this problem. So maybe "fluoro phase safe speeds" must be full multiples of 60, or 30 (exactly half), while the other HMI safe speeds would be "exposure safe," but not necessarily "phase safe."

It would be really interesting to return to that location and shoot tests at 23.98, 24, 29.97, 30, and 59.94i (For the purists, I'm rounding to 2 decimal places here, I know it should really be 29.976 etc....) to see which speeds, if any, are safe.

Even more interesting to shoot tests on film at the same shutter speeds and compare.

George Hupka
Director/DP
Downstream Pictures
Saskatoon, Canada



George is right. There are two different issues being discussed here. From what I gather from several previous posts here by others, 23.98 has led to an exposure change during a shot. If only by a few IRE.

The question I originally asked about was a color shift only during a shot. I haven't had a chance to look at the video that demonstrated the problem so I can't say whether the color shift only happened in the fixture that was illuminating the one human in the shot or if all (maybe 10) fixtures lighting the whole room are fluctuating.

I can say that on Thursday I spent almost the entire day shooting under fluorescents in several different buildings and used a monitor every time to make sure I didn't get fooled again. Not a single case of phase shifting was found. Same F900 at 23.98 with 1/48th shutter.

Since we didn't use a scope I can't speak to exposure changes. I also agree with George that the problem I had can probably be traced to a bad ballast or some other problem with that single fixture that I had the bad fortune to place a subject under.

Randy Miller, DP in LA



>…Not a single case of phase shifting was found. Same F900 at 23.98 >with 1/48th shutter. Since we didn't use a scope I can't speak to >exposure changes

If you set up an F900 in front of a face lit with HMI light and shoot at 23.98 (1/48th) you will see a (cyclical) phase shift in the flesh tones. This is easy to see on a vector scope. The same effect is in evidence with fluorescent sources.

This is *not* a source related issue.

What is more difficult to specify is the degree to which this bothers people.

>…So maybe "fluoro phase safe speeds" must be full multiples of 60, or >30 (exactly half), while the other HMI safe speeds would be "exposure >safe," but not necessarily "phase safe."

Maybe. The main question is, "Is there something changing within the take?"

At 23.98 (1/48th) there is. At 24.00 (1/48th) there is not.

David Perrault