Adding Plus Green to LED Panels

D

Digitally Pleasing

Active member
Hi All,

I've got an Ikan IFB576 Bi-Color LED kit and I notice a magenta spike at 5600k. It supposedly has a CRI of 87 on the daylight bulbs and slightly lower on the tungsten end. Will adding a plus green gel help me get closer to a pure white daylight? If so, how much? An 1/8th, 1/2... just guessing?

Not sure if this is a dumb question or not; but does adding plusgreen actually change the net CRI? Or is it still 87 with or with out the gel?

These lights are no longer returnable. I could trade them in, but I'd be taking a pretty big $$$ hit. Would love to find a solution other than fixing in post. Thoughts?
 
Last edited:
Your instincts are correct to add some magenta...as to how much...it varies with different LED systems. If I were you, I would start with 1/8....or likely 1/4 minus green and go from there. Do everything you can to not have to "fix it in post." Probably the single most dangerous phrase on set.
 
Thanks for the reply. So I'm going the other way around; the lights are too magenta so I'm thinking of adding PLUS green. Same thought process though, right?
 
You could also go to a local rental house and they'll probably just give you a few cuts off of used rolls for free.
Those swatch books are also free if you get them from somewhere besides B&H (like a rental house).
 
I wouldn't worry about changing the CRI - it really doesn't mean much anyway. What matters is how the light looks. However don't make your decisions about gelling with a color meter or by eye. See how it looks on your camera with a good monitor. Every camera seems to respond a little different to FLOs and LEDs
 
Stop guessing and find the right values.

Shoot a raw still using a DSLR of a known-accurate white & gray card, and open in Photoshop or a raw editor. Shoot so the white chip isn't blown out but right on the edge. In the Raw converter, make sure the color temp slider is daylight, and tint is 0. Check all controls to make sure nothing is affecting the color balance.

(I guess you could use a frame grab from your video camera if you don't have a DSLR - but many video cameras jack with the color matrix, etc).

Open the file and eyedropper the white and the gray with the "info" palette open. Look for even RGB values (like gray being R150, b150, G150). Check the whites, too.

If you have a red spike - say that middle gray is R175, B 150, G149... add a new layer, and make a square of the closest EVEN gray you can - in the red-spike case, do it 150-150-150. Overlay it on the photographed gray layer for a visual reference, keep it on its own layer. Having the filter layers BELOW the photshop-generated "perfect" gray really helps me see which way things are going.

Now add a "photo filter" adjustment layer - here's where it gets a little guess-ey, as some of the filter choices are real photo filter (80A, etc) and some are colors, none are gel colors.

But the "magenta" photo filter is a good minus-green. At 12% it does a passable 1/8 minus green. So compare the photo filter swatches to what you know of CTB, CTO, etc. Keep in mind most gels are 1/8, 1/4, 1/2, 3/4... so you have to add filters in realistic increments.

The filter combos you use in photoshop will give you a good road map for gels - BUT THEY ARE IMPERFECT in how they relate to grays vs. whites vs. blacks... if it seems like 1/8 green and 1/4 CTB are what's working, re-shoot the test with those gels on the fixture, and see how close your grays are now. Your whites and grays should be more balanced to each other with gels vs. Photoshop layers, too.

I have some 6000k DIY HIDs that can get dead-accurate with 1/2 CTO, for example. And most of my asian knockoff kinos need magenta for their yellow cast, like minus green with some CTB - this is how I've dialed in all my "daylight" stuff.

Someone really needs to make a photoshop filter or app that takes an image of a gray card and calculates commercial gels needed to correct - color meters give you over-the-lens filters, but for this you need gels.

One of these days I'll at least come up with a set of photoshop filter colors that approximate common gels, that shouldn't be really difficult with a good film scanner.
 
Michael Carter said:
Someone really needs to make a photoshop filter or app that takes an image of a gray card and calculates commercial gels needed to correct - color meters give you over-the-lens filters, but for this you need gels.
Click to expand...
Most gaffers I've worked with (if not all of them) own a color meter, he's certainly not asking the 1st AC to drop in a magenta filter because the meter is telling him that the lights are too green. You just have to interpret the readings they give you differently to get the right gels. I think they usually carry a little chart in the case for converting the values. But then again, the preferred method these days is to just look at the monitor and see if it looks right or wrong...
 
Color meters are notoriously inaccurate for judging green & magenta in LED's and Flo's and every camera seems to respond differently. A good monitor is your best friend here.
 
I thought the issue was discontinuous light spectrum- do modern colour meters actually produce a spectrum? The older ones I've seen just have a black body temperature with a plus/minus green/magenta tint adjustment. That's just not the appropriate way to measure something with a spiky spectrum or big spectral gaps (like most cheap LED's).

I gave up on filtering LEDs because I found the results were unpredictable. I could tune to a reasonable balance by eye, or on the monitor. But the spikes in the light spectrum can interact in very subtle ways with the dyes in the R/G/B Bayer filter on the camera, resulting in (say) magenta shifts in the blacks and green shifts in the highlights. If you have mixed colour temperatures (different manufacturers' lights, mixing with ambient daylight etc.) or multiple cameras can becomes unmanageable.

My best recommendation is to shoot not just a mid-grey card but a full colour chart. At least that way you have a constant reference to match to in the grade. I found that balancing white, black and mid-grey separately gave quite a good result, but sometimes at the expense of zombie skin tones (which is really due to the lack of reds and oranges in the spectrum emitted by cheap "daylight" LEDs).

My final solution was Gekko Kelvin tiles, which have six LED colours per LED, all controlled independently with a DMX toolbox. Having many more spikes in the spectrum - including important missing ones like amber and red- lets you do a much better job of colour matching. Unfortunately these aren't being manufactured any more, and some of mine have developed faults (when they get hot, the top third of the panel goes dead. Happened to several different panels so obviously a design flaw).

This isn't unique to LED: I had the same problem with fluorescents back in the day.

There's a fabulous reference here:
http://www.screenlightandgrip.com/html/emailnewsletter_generators.html#anchorHigh Output AC LEDs


Cheers, Hywel Phillips
 
You must log in or register to reply here.
Back
Top