Re: Why all the fanfare for XPG or XML
In God we trust, all others bring data that supports me. I'd ignored the tone, but let's now review why sunlight is good light:
1. Bright. (It reaches aaaaaaaall the way across a field or valley)
2. Very bright. (Reasonably high lux is the primary factor for being able to accomplish most visual tasks)
3. Pretty good runtime.
4. Variable CCT, tint, and pretty good CRI. (Although, annoyingly, these change throughout the day)
5. Bright.
"What is the best light?"
A good light is one that works well for you for a task. Light is often modeled with these three or four parameters, which do NOT capture all aspects of color perception and task-appropriateness: Output, CCT, CRI, and tint.
Output: Artificial light output levels for all interior design are sorted by task, from 'relaxed' to 'demanding.'
http://home.wlv.ac.uk/~in6840/Lightinglevels.htm has some examples and discussion. For most tasks requiring speed or precision, more light is usually better than less.
CCT is task-dependent as well. For example, cooking works best for me under medium to high CCT light with good-enough CRI. I find that fluorescents over 85 CRI are adequate to good, and LED from 85+ is quite good. Super high CRI LEDs make a little bit of difference, but I find far more benefit from getting the right CCT and tint for a task, than trying to get a single perfect LED… Especially since LEDs of all types I know of change their CCT throughout their current ratings. At very low current the overloved Nichia 219 looks to me like an array of blue-tinted white LEDs, for instance. Crees may shift color by 800-1000K throughout their low-to-high temperature and current regime. I like medium-to-high CCT for cooking, but other tasks may be best served by lower CCT. Among these, preparing for sleep to allow correct melatonin cycles is one.
CRI is overplayed as well. It's sort of an outdated metric, and is obsoleted by the Kruichoff curve. CRI idealizes the black-body radiator as the uber-color-perception device. However, color perception requires photons of a matching wavelength to allow perception of a color. Some people exaggerate the effects of spectral spikes in modern light sources, forgetting the results of these spikes are manageable. Black-body radiators follow a bell curve spectral power distribution so that the ones that allow you to tell indigo from black have unusual tints. There are much better metrics for reporting color perception, but it's a list of ratings, not a single number. Single numbers sell, arrays of numbers get ignored. As in this discussion
Tint is quite noticeable. CCT is a major player in tint, but there are others. Cree's tint bins are fairly wide, but better than LEDs were ten years ago. That's no excuse for the current situation, but there it is. They'll respond to the market, or not. Most LEDs shift CCT and therefore tint through their normal operation range, as in the allegedly "mint green" Cree XP-Gs that came out last year (That were usually more like a drop of mint ice cream in snow at reasonable drive currents). There are varyingly-appropriate tints for different tasks, but it's difficult to separate tint from CCT. So aside from CCT effects, tint in a good light source will not be overwhelming (Few greens, purples, and so on are desired).
When I choose a light for a task, my first consideration is output: Lumens and beam pattern. A given task may call for half a lumen, floody or throwy. Or it may require one hundred, or thousands. The first consideration is, "Are there enough photons?" All else follows. I choose an appropriate CCT/Tint. Most of my lights are neutral-white because I find it to be a good all-around compromise. You may find something different to be your favorite. And then CRI comes in later, because most modern lights are adequate for color perception outside color-critical tasks.
I don't find digital photography to require especially special light sources. I find that light characteristics are far less important than matched lights. Burning kerosene will appear orange next to burning steel. Many LEDs will appear blue next to fire, and cold next to sunlight. So in photos I don't care so much what light I use, as that they all match. And as long as they do match, some CCT will have a natural appearance. In extreme cases (HID arc light, 8000K used for portrait) I have played with saturation to bring back missing red photons. But with or without that editing are striking in different ways.
So: End the tint snob religion. Find what you like, tell people about it, and don't knock their choice. I'm glad you've found what works for you (If you have), and if not? Ask!
