Exposing Provia under 5000k LEDs

[NortheastPhotographic]

Can anyone help me with this, as it's not listed in the Provia data sheet. I'll be doing a project in a woodshop that is lit with 5000k LEDs. (5000K for Kelvin). Do LEDs in and of themselves require exposure modification?

Re color, I think a 5000K base temp on Provia may require an 81a, if anyone has anything to add there I would be grateful.

I'll be shooting 4x5 and using a strobe for some portraits so I'll also need to gel the strobe if I'm going to filter, but I'd like it to be balanced as much as reasonably possible with the space...

Thanks!

 

[AgX]

There are two things do be aware of:

-) the difference between the colour sensitization for 5600K of the film and the light temperature of the ambient light of 5000K. It depends on the very job whether to filter for neutrality or not. Basically it would be an easy task with filters.

Likely more important is to

-) be aware that colour temperature strictly speaking only applies on radiation caused by temperature. And only in such cases different colour temperatures can be compared and filtered up to each other.
Mathematically a blue and a red light can give 5600K mixed together, though omitting part of the spectrum.
LEDs emit by luminescence and such radiation is more or less non-continuous. Think of common TL tubes which light seems neutral visually but typically leave a green cast on film, due to some peaks.

LEDs got better by now. One hint at usefullness may be a colour rendering index of above 90. Current Phillips household LEDs got one above 80, what was only second class in times of TL tubes.
That index of course is hard to get to know once the lamps are installed...

There is some test on light quality, but likely you have to do a test on your film.
But as said it depends on the job. Yours it not a reproduction job.

 

[NortheastPhotographic]

Yeah the lights are installed and I can't get make or model from the folks running the place, I tried. Not sure how to approach this one...

 

[Mr Bill]

Yeah the lights are installed and I can't get make or model from the folks running the place, I tried.

If they'll let you unscrew a bulb you can probably get the data off base. One question: if you don't know the specific lamps in use, how do you know they're 5000 K?

There is some test on light quality, but likely you have to do a test on your film.

I definitely agree.

The typical "white LEDs" nowadays are a mix of an actual bluish LED combined with some fluorescent brighteners to fill in the non-blue colors. They give a fairly full spectrum except for a big spectral "sag" between the two components, just off the bluish color. Because of this sag, I wouldn't trust them if you needed really good color reproduction. The range of "color temp" available seems to be based on the balance between the two. An actual film test is probably your best bet, in my opinion.

If it was me, I'd probably do some preliminary tests with a digital camera, then once I had a starting point do a test shot on the film. (With the digital you can photograph a white card, then use the color histogram to see how closely the color spikes line up.) But I guess it really depends on how big the job is and how critical the color.

Ps, there is also some variety in the "color temp" of electronic flash; the best source of info is probably from the flash manufacturer, who often has both "UV-coated" and "non-coated" flash tubes. (I think anyone using non-coated tubes is just asking for headaches, at least for people shots.)

 

[NortheastPhotographic]

The problem is, I'm not going to be able to do any tests. This project is taking place at a state prison, and I only have the time that I'm there. Also getting them to just unscrew some bulbs that are very high off the floor is not really an option.

 

[AgX]

If they'll let you unscrew a bulb you can probably get the data off base. One question: if you don't know the specific lamps in use, how do you know they're 5000 K?

The LED-substitute bulbs I got at home have no colour rendering index on them. One would have to look them up. I bought them this year, they already vanished from the Philips catalog. But all I looked up got an CRI of 80.

 

[Mainecoonmaniac]

LEDs can vary from manufacturer to manufacturer. From my experience, daylight LEDs are pretty amazing. Where I work, we have a few daylight balanced LED light panels and the color rendition is amazing. They can be dimmed without shifting the color temperature. I have under the counter LED lights and color rendition is very good. If you scan the transparencies, you can correct them in Photoshop.

 

[Mr Bill]

The problem is, I'm not going to be able to do any tests. This project is taking place at a state prison, and I only have the time that I'm there. Also getting them to just unscrew some bulbs that are very high off the floor is not really an option.

Hi, I don't know how cooperative they'll be with you, but the maintenance department must have some spares in stock.

If you can't get any info or they won't allow time to experiment with a digital camera, I'd personally just concentrate on the lighting with my strobes and make that be right. At best the background lighting will be pretty close; at worst it'll probably have a yellowish cast, possibly pretty strong.

If I had to guess about the "color temperature" of the LEDs, I would not expect it as high as 5000 K. Traditionally, institutional use, in general (sorry, I don't have prison experience, at least not yet!) had been at something more like 4100 K fluorescents per lighting suppliers I've worked with.

If you want to try to narrow it down onsite, with a digital camera, try this: set camera white balance manually, at 5000 K, shoot a white card and look at the color histogram. If the RGB color spikes all line up, then your light is close to 5000K. If not, try another color temperature and repeat. (If you bring your rack of color filters along, maybe you only test for the color temps that you can correct, and just pick the closest one.)

 

[MattKing]

The problem is, I'm not going to be able to do any tests. This project is taking place at a state prison, and I only have the time that I'm there. Also getting them to just unscrew some bulbs that are very high off the floor is not really an option.

I'll be doing a project in a woodshop that is lit with 5000k LEDs.

Can you get someone to take a digital picture of the location and send you a raw file? You could gain some colour information from that.
I'd plan on aiming for warmth. A warming filter for flash and ambient light should give pleasing skin tones (given the location!) and may go a long way toward natural colour reproduction.

 

[DREW WILEY]

At this stage of the game, LED's can be mixed to fool the eye and simulate this or that color temperature; but because you're dealing with a discontinuous spectrum
to begin with, as well as half-baked RGB peaks well outside one or more respective bullseye, I'm rather skeptical about the possibility of precise color rendition. You can obviously test to see if anything is "good enough".

 

[TheFlyingCamera]

Plan on bringing some warming filters, and if possible, shoot Portra 400 negative film. It's the most forgiving (other than Portra 160) of mixed lighting conditions I've ever used.

 

[wiltw]

Having recently upgraded our kitchen, we put in recessed cans with LED emitters. We chose a nominal 4000K temperature unit, having CRI of 92.

Just now I shot a Macbeth Colorchecker card under our kitchen lighting (left) and in bright sunlight (right). Using Lightroom's color balance sample tool to sample the #4 gray patch in bottom row, this is the result...

The LED unit's manufacturer claims a nominal 4000K, Lightroom's sampler tool detected 3850K. Eaton Halo LED.

 

[Lachlan Young]

At this stage of the game, LED's can be mixed to fool the eye and simulate this or that color temperature; but because you're dealing with a discontinuous spectrum
to begin with, as well as half-baked RGB peaks well outside one or more respective bullseye, I'm rather skeptical about the possibility of precise color rendition. You can obviously test to see if anything is "good enough".

Good LED's are out there, but they're often really expensive - and the test that matters is not CRI but the TLCI test - where anything that scores over an 85 is rated as 'would not be noticed by a skilled colourist' - and there's an awful lot that fall well below that - including from some well known names.

 

[AgX]

The TLCI is a rather new european test. It is basically the same as for establishing the classic Colour Rendering Index, exept that it uses not 8 but 24 colour patches and their hues are differently chosen than before.
The total range too is from 0-100, but the "fine" range is wider in number.

 

[DREW WILEY]

wiltw - It is impossible to convey hue nuances precisely over the web; but even in the side-by-side chart images you posted I can detect a significant difference, yet have no way to know if that is just an artifact of the scanning or whatever, or is inherent to how the chart was lit. And ironically, it's the image on the right which looks distinctly muddier. What film were you using, and did you correctly balance the color temp outdoors by using an appropriate filter? That kind of fuss might not be necessary every day, but should be integral to any objective test. In this case, you should have balanced the outdoor shot
itself to 4000 K.

 

[philosli]

Since you plat to use a strobe, is it possible to rely your strobes as the dominant light source? So you expose to your strobe lighting, which can be several stops stronger than your ambient light. Given that you're using Provia, I guess that's already the case.

Or use black and white films, or negative films so you can correct the color shift in printing.