Easel and Palette Lights

(Peter Höhsl) #21

I did some extensive research on light sources last year to find an an optimal solution for my situation and thought I’d share my findings on the subject. (Sorry, it got a bit long. I’m a bit of geek when it comes to these things. :blush:).

  1. There is no cheap solution to good quality artificial lighting.

  2. Incandescent light sources provide the best light quality with a totally even distribution across the entire spectrum ie. a CRI of 100. They’re even better than natural daylight because they’re not influenced by the atmosphere and the environment, and don’t change during the course of the day. This applies to everything from light bulbs to halogen lights and even candles, the latter just having too low of a colour temperature (ca. 1800K) to be practical. Halogen sources usually have a colour temperature of around 2800K with those that we use in the film industry operating at 3200K. The latter, however, have their lifespan reduced to about 180 hours, run extremely hot, and have a tendency to explode when not handled correctly. The above mentioned Solux lamps are normal halogen lamps that get pushed to 3000K and then filtered to provide light at higher colour temperatures - there is no other way to achieve this since tungsten melts at about 3700K. This makes them probably the least efficient light source on the market because to modify a 3000K lamp to emit light at 5000K you loose at least 2 stops of light output. A loss of 2 stops in photographic terms means that the lamp then only has 25% of its original brightness. The rest is converted into heat. So, the best light source, but inefficient and costly with a short lifespan.

  3. CFLs are dead. The spectrum of even the best ones doesn’t hold up against average LEDs. They usually have a massive spike in the green part of the spectrum that even manufacturers like the ridiculously expensive Kino Flo lights can’t suppress. And it is noticeable. Manufacturers can hide this spike by placing it in between colour patches that are measured for CRI so it won’t show up.

  4. LEDs are probably the best choice - but which one? I found that when a manufacturer indicates at least CRI 90 with R9 of 80+ you’re quite safe. It is very difficult for us to differ between CRI90 and CRI100, provided that measured fields such as specifically R9 are high as well. If a manufacturer doesn’t disclose specs on their product don’t touch it. I also found that lamps that are “specifically made for artists” aren’t any different to any other product. Some, as I recall reading in independent tests, are actually quite mediocre. So, don’t fall for marketing lingo. Also LEDs with exceptionally high lumen/watt output have a lesser quality. These lamps usually push the green part of the spectrum to achieve brightness at the cost of quality. So, a lower lumen output is an additional indicator of a possibly good quality lamp. Consequently good quality LEDs have a tendency to produce more heat than ones of lesser quality.

  5. Colour temperature is a matter of personal preference. For colour matching, however, the most neutral is around 4000K - 4500K. Good quality light at this temperature really has a “white” appearance. It’s also very much in between two possible main viewing scenarios in which your paintings may end up in, ie. cool daylight from a window or warm artificial lighting in your average home. I eg found that when painting with 6000 - 6500K lamps my yellows would leap of the canvas when viewed in incandescent lighting.

  6. When working with prints as reference, colour temperature is very important and should be around 4500K - 5000K, due to an effect called metameric failure. At 5000K this effect is least noticeable. In an average photographic print colours are made up of cmyk inks, which means that most colours don’t really exist in the print such as an olive green or skin colour eg. If you now match a colour with different pigments it might look correct under one light but completely different under another. One might imagine that the colour shift affects both the mixed paint and the print, which is true, but in a different way. Hence some colours will no longer match at all under different lighting unless matched under ideal lighting conditions.

  7. Mixing different light sources is generally not a good solution because the chance of improving the situation over just the single worst one in the mix is less than 1%, unless you have proper measuring equipment to assess and find exactly those lamps that complement each other in the spectrum positively. Mixing anything other to an incandescent light source will worsen the result. Also filtering light sources with a CRI of less than 95 should be avoided because the resulting CRI can be highly unpredictable.

I finally settled on 4000K PAR38 Soraa Vivid LEDs fitted to a track lighting system (https://www.soraa.com/products/lamps). Sadly Soraa no longer produce the 4000K versions so I can’t really recommend them unless you find a retailer who still has some in stock. However, if 3000K is your preferred colour temperature to work with, I can highly recommend them. They are also increasingly used in museums. Their Vivid series has a CRI of 95 with R9 of 95. What really sets them apart though is the fact that they use 3 phosphors to correct the light instead of 2, like most other manufacturers. Background info at this point is that most LEDs are initially blue light sources that require the addition of green and red phosphors to bend them into shape so that we perceive them as “white”. Soraa use a different process to manufacture LEDs which initially turn out purple and in combination with 3 phosphors provide not only a more even distribution across the spectrum, but also compensate for the uv light portion which is usually missing with other manufacturers except incandescents. This uv light portion is important to render certain pigments with fluorescent qualities correctly. A lot of fabrics that use whiteners fall into this category, as well as zinc white (not that you should use any :grimacing:).

If I had to refit my studio with new lighting today I would probably choose lamps from Scangrips Multimatch series of colour matching lights. These lights are made specifically for the car paint industry to match paints that need to look the same in all lighting conditions. Also their technical specs are beyond anything I have seen anywhere else. They measure their CRI of 95 accross all 15 patches with R9 at a staggering 99. They also have variable colour temperature, which can be useful to quickly test how artwork may appear under different lighting.

(Anthony Waichulis) #22

■■■■! Thank you for this spectacular walk-through Peter. I am sure that this will be VERY helpful to many in our community here!!! :heart:

(John McCuin) #23

Thanks for the great post and suggestions. Very helpful!

Just curious why you would say:
“Mixing different light sources is generally not a good solution because the chance of improving the situation over just the single worst one in the mix is less than 1%, unless you have proper measuring equipment to assess and find exactly those lamps that complement each other in the spectrum positively.”

Suppose we start with what you list as a terrible source, CFL, and suppose we add either an incandescent or LED. After that, from what I understand, the spectral power distribution of our two bulbs is just the summation of the two individually, so by adding the incandescent or LED, we improved upon the SPD of the CFL.

It seems the game would be more about seeing if we could find a combination of bulbs that produced the desired SPD. As you mentioned, incandescents are smooth but too high in the warm colors. LEDs give a boost in the middle of the spectrum and a boost in the blues. CFLs (or any flourescents) are full of spikes, usually in the green and orange, sometimes in the blue. Seems like a combination of LEDs and incandescents could give something closer to daylight than either could alone.

(Peter Höhsl) #24

Hi John,

Mixing light sources is an additive process. In the graph, you posted, you can see that incandescent has the most even distribution across the spectrum. If you now add the CFL you are creating a curve that still has all the spikes and bumps, only now at the angle of the incandescent light. In proportion to the total light output the spikes are less, but they are still there. If you had a light source that had spikes in places where another light has gaps then you might stand a chance of improving the situation. In reality, though, this is very unlikely to achieve and most impractical, because you really need to know the individual spectral curve of each lamp. The curves in your graph are just illustrations for the typical behaviour of the different light types. Each CFL, LED, etc has its own curve and they can differ vastly.

To change the colour balance of a tungsten light to daylight you would use a linear colour correction (or more specifically daylight conversion - also known as CTB) filter. These filters have a blue appearance and level out the curve to a more horizontal shape without changing the overall characteristic of the curve ie. the CRI. This will give you the best light source, but at the cost of losing 75% of your light output, because a filter can’t add more blue; it can only remove the yellow, orange and red chunk on the right hand side of the curve.

Be careful not to confuse the colour temperature and the CRI. You can change the CT by mixing two light sources with different CTs. This only makes sense though, if both lights are of the same quality. A lot of LED film lights work like this and allow you to dial in the CT precisely.

(John McCuin) #25

Agreed, obviously the CFL is a bad choice, but I was just pointing out that adding an incandescent to a CFL results in an improvement above the CFL alone, and thus mixing bulbs can (at least to more than 1% likelihood) result in a better SPD than the worst one.

All that being said, I agree. Avoid CFLs & Flourescents. They’re super spiky.

To my eye, it seems that a combo of incandescents/halogens/tungsten and LEDs could come closer to sunlight than these common options do alone. An important factor with the LED seems to be the size of the blue hump. Either way, best of luck painting! Thanks for your input and responses!