I recently got a new HMI light (a LupoLux Daylight 1200) and noticed the quality of skin colour is not as good as tungsten.

I did a quick test comparing tungsten to HMI to daylight and put the results on my blog: Flash vs Tungsten vs HMI vs Daylight | blog.8Networks.net (http://blog.8networks.net/2012/04/19/flash-vs-tungsten-vs-hmi-vs-daylight/).

Does anybody have any tips on getting better skin colours using HMI lighting?

I have found that HMI are a little blue
You might want to try some cosmetic gels.
peace.

Some random thoughts in no particular order:

Tungsten is *not* a full range source. It lacks, some say severely lacks, blue. But... it can create some gorgeous skin tones. Because skin tones are based on red (blood) which tungsten handles well. In particular, if you like that glow that golden hour often gives skin tones, tungsten lighting is hard to beat.

Unfortunately, the lack of blue in tungsten lighting can lead to noise in video camera sensors (the blue channel can be light starved and thus noisy). This is part of the reason (along with efficiency and heat generation) that people are moving away from tungsten and toward 5600K lighting. Not sayin' it's good or bad, just that it is.

Cloudy overcast sunlight is extremely blue -- 9000K and higher depending on the time of year, longitude, and altitude. If you want something closer to 5000K, you'll have to forgo the clouds. A clear blue sky actually results in *less* blue in the sunlight. Where I am, I get about 5000K around noon on the spring and autumn equinoxes on a cloudless day.

I don't have a lot of experience with HMI (can't afford it), but what I've read says that HMI bulbs have to break in for a while -- 100 hours is what I keep reading. And they have to warm up and stabilize (10 minutes or so) every time you turn them on. That said, HMIs, like fluorescents, are discharge lamps. This basically requires spikes in the spectrum, and most lights of this type (fluorescent, HMI, LED, etc.) have a fundamental green spike which is more or less controlled depending on the lamp, ballast, etc. And yes, the ballast does have an effect on the color and quality of the light, so your bulb may not be a good match to your ballast. But I suspect that as your blub ages in it will "calm down" some and provide a better quality of light. That said, most 5600K HMIs are actually bluer than that (6000K or so), so you might find some CTO gel will help you get the skin tones you desire. You might also want to try some minus green, maybe a 1/4 gel? IDK.

It's not "fair" to compare different units unless they are corrected to the same color temperature via gel and metered with a color temp meter. Skintones will always look better in a comparative test with a warmer rendition.

Beyond this,iIt's tough to do a true apples to apples comparison unless the size, intensity and color temperature of the sources are matched. The differences between the daylit scene in your test and the HMI, for instance, go beyond the sources themselves. It wouldn't be the easiest thing to set up but I could see placing the subject into a tent with a given size "window" of diffusion cut into it (say, 2x2) that would start with ambient daylight, then a secondary tent placed adjacent to it and the light source duplicated by tungsten, HMI, LED etc. that was corrected in intensity and color temperature to match each other as closely as possible.

In my experience, it's indeed possible to get decent skin tones with HMI.

My experience with LED light is that weird sickly skintones are normal. But HMI is used very widely in Hollywood with good results. It's certainly possible to get it right, albeit probably a bit more difficult having to keep an eye on the color temp of aging HMI bulbs and perhaps apply color correction.

This doesn't knock the original poster at all; HMI is not going to give you great color "out of the box".