Off Color HMI

[David Regan]

Just got some footage back shot on 16mm. I am concerned about the differing color temperature in a couple of my shots. All our light sources were daylight balanced, a 575 and 1200 HMI, and daylight Kinos. I was shooting Tungsten film with 85 filter. However in a couple shots, and most noticably this one attached below, the light from the 575 is really really off, much bluer than the other lights. Notice in the first shot, the door in FG, is the normal color temperature, it may look slightly magenta, the compression is off, and looks different on my different monitors. But if you look at the back wall, in the middle of frame it is much bluer, which is where my 575HMI is hitting. The door is lit by a Kino and spill from the 1200 through the window.

 

 

In the second still, notice the very left portion of frame, that sliver of side wall visible, is again much bluer and is being hit with the 575. The back wall, lit with the 1200 is normally colored.

 

 

Any ideas as to why this is occuring? My thought is either the bulb is old or screwed up, I don't know much about HMI bulbs if ct can vary that vastly. Or I can't remember absolutely on this, but if I only had the UV lens in, and no wide/medium/spot etc...lens in front of the light, would that alther the color? I can't remember for sure, but I think we may have had it out to get extra punch out of the light, or perhaps it was neglected to get put in. My reason for wondering this is there are other scenes lit with the 575 that look perfectly fine. Its quite annoying that this happened and I hope to figure out why to avoid it happening again.

 

Thanks for the help.

[Mitch Gross]

I'm always finding variations in HMIs and NEVER get absolutely pure color out of kinos. First move for me is to fire up the daylight units and let them run for a few minutes, then pull out my color temp. meter to see how off they are. It's usually a bit of minus green or a bit of CTO or straw. Stregnths are rarely more than 1/8th, but that can be enough to cause the issue you're seeing. My guess is that your Kino is a bit magenta and your 575 was a bit cyan.

[Alessandro Machi]

If you're running off of the same generator (assuming you were) is it possible that the generator amperage output dropped after you hooked up all the lights and that affected each light differently? Lower amperage could also affect warm up time which also affects color temperature.

 

I noticed on a video shot of mine recently that as I was zooming out it looked the image went bluer at one point in the zoom and then normalized again. It might be useful to test two lenses, one a zoom lens versus a fixed lens in the same setting and see if they both reproduce the colors similarly or not at the same lens settings.

 

Ultimately, a color meter makes a lot of sense.

 

I recall once having a mini hmi that stayed purple for quite a while. The strange part is I still use that bulb and it doesn't seem to go purple anymore. Which makes me think it could have been voltage or amperage related as now I try not to go below 13 volts outputs since my four pin cable length will knock it down a couple more tenths of a volt.

 

The door definitely has that "purple" look that either older HMI bulbs get or one that is not quite getting enough voltage but enough to light up, it also does look like it was just turned on.

[David Regan]

Yeah I think seeing this is now convincing me more than ever its time to shell out the money for a color meter <_<

 

Lamps running off color makes sense to me, the only reason I wonder is that I have other scenes lit with the exact same lights and there is no problem. Its only here and there that I see this issue, which makes me think something that can change with the light, the only thing in my head being the lenses.

 

We actually weren't running off a generator, since it was such a small package, it was all just off different circuits in the apartment we were in.

 

Thats a good call about lenses, I never thought of that, accept again I have times with and without the problem on the same lens, was usually on a wider prime in some of the tight spaces.

[Walter Graff]

Unlike the stableness of tungsten lights, multiple HMI fixtures do not always give off identical color temperatures simply because they are all HMIs. Mostly it depends on the age of the lamp. It has little to do with generators or anything else as long as your voltage is constant or in range of the transformer, you will get light. It's mostly about the lamp. There is a window where HMIs give off that light blue 5600k we so associate with HMIs. And before and after that stable period, they tend to go other ways, especially visually purple like you show and especially purple with 575w fixtures as lamps age. Very simply anyone using HMIs who wants exacting color representation needs a color meter and correction gels on hand to correct for lamp wear/variable outputs of smaller vs larger size fixtures, etc. You'll want to make sure to burn fixtures for at least ten minutes before making any color correction decisions. And a range of minus/plus greens, along with light grades of CTB, and even very light grades of CTO should be on hand to make the corrections. WYSIWYG with HMIs so while you may not have noticed the differences while shooting, they were there. The color shifting is pretty evident. Because you say you have less experience with HMIs you probably didn't know to really look. But in the end, after it is all set up simply step back and look at the quality and color of lights from all fixtures. The "purpleness' of HMIs can be seen well compared to a fixture giving off a good 'blue'. But while it bothers you, as I look at your photos I don't think it is bad at all as it is more in the 'shadow' areas of light and easily corrected in transfer or later in post. Frankly, you seem to have done well with your lighting motivations as I see it. Nice job!

[Mike Williamson]

HMI's rarely seem to match each other. If you're trying to balance them to each other or daylight, they're going to be off in basically two different directions: 1) orange/blue and 2) green/magenta. I try to think of it like this, as it becomes simpler than trying to remember what's going to fix "purple" or "yellow".

 

In the example you posted, it looks like the HMI is going towards green (as Mitch points out) and the Kino is going towards magenta. It's possible that they are also off on the orange/blue axis (aka color temperature), but if I were trying to correct these on set, I'd try to match them on the green/magenta axis first and see how that looks.

 

If you're trying to match to sunlight, HMI's are generally too cool and need to be warmed up a little. For skylight, they're often too warm and need to be cooled down some. So with natural daylight, I'd probably make those tweaks before trying to chase down the green/magenta issues.

Edited January 13, 2008 by Mike Williamson

[David Regan]

That all makes sense and thanks for the thorough answers, I'm very inclined to think the answer lies in just bulb age/variety. My only concern is that if it was a bulb I would expect it to be in all my scenes. But if you look at this still, there is certainly no deep blue from the HMI like in the earlier still, and this is using the exact same HMI and Kinos. I'd like to add I'm not sure what compression might be doing to the color. The magenta that seems to be coming from Kino-lit areas is not at all visible in the transfer footage I have. When I compress it and upload it I do see it much more heavily however, and it depends on the display I see it on. But Kinos regardless, if that 575 is really heavily blue, why not so here? Is it possible the color could change if somehow a lens was neglected to be put in in those off color shots? Or perhaps the lab was able to time out the color in this still but not the others?

 

Thanks again for all your help

 

 

[robert duke]

a SECONDARY ANSWER is that the color response is more from the wall paint/reflected light. Painted walls can look one way to the human eye but reflect light in a totally different way to film. It is just a theory. The deep blue in the stills seems to only be on the door and door frame, a different paint than the wall in the background. Granted I am only looking at stills on a laptop, not streamed video or projected video. It would account for the appearance and disappearance of the color shift.

[David Regan]

I just talked to the guy where we checked out the HMI, and he told me he had just changed the bulb in that particular HMI, and it was running at about 9000K :o So you guys were right it was the bulb. I'm still a little confused as to why it wasn't always present, my best guess, is in the stairway scene, which was just the 575, it was easy to time out, whereas when it was running alongside the normally operating 1200, it was harder to fix.

 

Annoying, but on the other hand I'm now convinced on acquiring a color meter.

[Mike Williamson]

Yes David, one of the main issue is the age of the globe, HMI globes get cooler as they age, unlike tungsten globes which get warmer. The other problem is that they're less stable than tungsten globes, and they can change color temperature throughout the course of the day.

 

Another possibility is that the colorist was able to balance some of the shots where you were using only one of the two light sources (just the Kino or just the HMI), as you mentioned. It's extremely difficult to time out color temperature differences within the same frame, you'd have to start messing with secondary color controls to go after it which is not something a colorist is going to do in an unsupervised dailies session.

 

A color meter is useful but expensive, so pick one up if you can afford it. I think it's still a smart idea to try and train your eye to see the color differences, it's a skill you need when doing color correction as well.

[Matthew Parnell]

HMI's produce the least stable colour at the begining and end of the bubbles life. I always check hired lamps in the workshop using a CT Meter, and label the lamps with tape either on the back of the head or the header cable with the required gel to correct it so all my lamps are the same temperature.

 

The kino flos are also another place to look for trouble. Kino tubes tend to become unstable at the end of their lifespan, and also for the first couple hours of use i find their CT to be a bit off. When dimmed the kinos change CT, also the CT can change with various factors to do with the balast. Again, check them with a CT meter and gel to compensate if need be.

 

If you are using Kino Flo Divalites take care to ensure the lamp is rotated the correct way around. If in vertical mode the ceramics must be facing upwards if in horozontal mode, angle the light 10 degress so the ceramic is again higher. This stops the mercury heading towards the cathode, which causes CT instability, and also causes the balast to heat up.

[J. Lamar King IMPOSTOR]

The only way to know what's going on with HMI's is to fire them up before the shoot and keep the same ballast/head combination for the whole shoot. Take a color temp reading and mark the head and ballast.

 

What you will find if you carry a color meter with you is that you will sometimes see a difference from what you read before. My experience is that they can be absolutely maddening and very inconsistent. I've found that a big reason for this is some kind of bad connection in the line. The ballast power feeder is a big culprit.

[Chris Schlaghecke]

I'm always finding variations in HMIs and NEVER get absolutely pure color out of kinos.

 

I am doing a television shoot for ESPN at the moment. All the Kinos that I am using are new, the tubes are all ostensibly from the same batch, also new. They difference in the magenta output to the naked eye is amazing!

some of them look fantasic, but the rest are just so different! I am finding a huge magenta spike with some of the tubes.

I have a very good ccu department that I am working with and so far they have been able to correct it without me using plus green, but I was very surprised!

cheers

Chris

[Jess Haas]

The color of Fluorescents will very depending on temperature. When they get hot they tend to have more of a green spike and I believe more blue. This is why they recommend not putting gels inside of the barndoors on kinos because it can cause them to get hotter which will shift the color temperature. Also, if you underdrive the tubes (ex. run 4ft bulbs in 2ft mode) they will look more magenta. I would also assume that there would be some differences depending on the temperature of the location, etc..

 

One thing that I have both been told and observed is that the human eye(atleast mine) tends to be more sensative to differences in the magenta/green level than film seems to be. I have been on shoots where there was a noticable inbalance but it looked fine on the resulting film even without color correction.

 

~Jess

[Chris Schlaghecke]

The color of Fluorescents will very depending on temperature. When they get hot they tend to have more of a green spike and I believe more blue. This is why they recommend not putting gels inside of the barndoors on kinos because it can cause them to get hotter which will shift the color temperature. Also, if you underdrive the tubes (ex. run 4ft bulbs in 2ft mode) they will look more magenta. I would also assume that there would be some differences depending on the temperature of the location, etc..

 

One thing that I have both been told and observed is that the human eye(atleast mine) tends to be more sensative to differences in the magenta/green level than film seems to be. I have been on shoots where there was a noticable inbalance but it looked fine on the resulting film even without color correction.

 

~Jess

Thanks for the tip re the 2ft 4ft mode and magenta spike, I had not realized that. I checked the ballasts when I noticed the problem and they are running in the correct mode. They seem to have stabilized somewhat in the last few days, so I probably should chalk it down to new tubes! They are running in an air conditioned studio, so temp is not an issue.

Thanks

Chris

[Brad Dickson]

To get an interesting picture of HMI and other arc sources look at their spectral distribution and then look at the sources spectral distribution you are trying to match. Obviously this will not be done on set but you can get an idea of spectral distribution differences by looking at lamp manuafacturers data or search the internet.

Judging an Arc on it's kelvin rating doesn't give you the full picture. A good article relating to Kelvin and Fluorescents is located here http://www.cameraguild.com/technology/kelvin.htm

 

As others have noted their are alot of factors that influence the output of an Arc, bulb life, burn time etc.

Put 60 moving lights in the air point them onto a stage floor and look at the differences even when they use the same type of arc bulb.

When working in video I find the vectorscope the best reference as to where the spikes and shifts are to any arc. As many have said check your lamps and don't just go by what you assume they are suppose to be because they have been given a Kelvin rating by the manufacturer.

[Jess Haas]

I just want to add that while there are many tools available for checking such things they aren't all that inexpensive so often people do not have them on set. Luckily all you really need is your eyes to do a perfectly fine job of matching lamps. It takes some practice to be able to look at a light and tell what kind of correction it needs, but it doesn't take much to know when the color doesn't look right. A color temperature meter does tend to make life easier, but it is far from required.

 

~Jess

[Brad Dickson]

I agree the first step is to be observant of lamp differences with your eyes.

If shooting video another method for comparing lamps is to bend a white card in half allow the light source you are trying to match to light one side of the card and light the other side with the other light source. Simply frame the card so the fold sits in the middle of frame and the white card fills the frame. You should see on the monitor the difference between the two whites. Simply add you correction gel to the light you are correcting until you get both sides looking white.