Problem is there's too much light at once and the eyes have to adapt to it immediately.
This isn't a problem with light level changes on the order of 5x..10x, only far more significant changes
The LED spectrum is definitely less desirable than natural daylight. Incandescent light is probably also better (if you can tolerate the subdued blues). The rest is a question. Some people complain about fluorescent lighting in offices, some other would probably complain about LEDs. Low CRI HID lighting has indeed lots of critics...
Incandescent is, in fact, an example for your next question :
Sorry that's difficult to imagine. Sounds like an oxymoron. That's like if colours didn't respond to the light spectrum.
The CRI of Incandescent is 100, but it is very lacking in the Blue
In contrast, the CRI of SON is around 20..25 and the light is Yellow color, yet it represents Blue even better sometimes. The way i think this happens is a correction process happening in our mind. This correction responds to the preceived CRI, not to spectrum :
- Under Incandescent light, the CRI is high (different colors in the surroundings are visible) so we believe that what we see represents what there is. When a Blue object is under the light it will appear dull, but we believe that it indeed is dull
- Under SON light, the CRI is low (everything appears in shades of the SON light) so we pick the subtle still visible differences, and extrapolate from them to what the color would have been. Since this happens in the subconscious, it just results in us seeing colors well under SON (well above what we would expect thinking of it as Yellow light and not knowing about the correction process)
Generally, there is a mismatch between CRI, ability to see colors properly (with the correction in action) and actual light quality :
CRI is an artificial unit, made by calculations based on few values in the D-chart. It's been designed to be representative of a would-be empirical test : A test sheet with 100 color samples is presented under the light. The number of different shades that can be identified is the CRI (same shade in different brightness still counts as same shade). So for example, if CRI would go between 0 and 4, the test sheet contain 4 color samples, and under the tested lamp 2 look the same shade and another 2 look the same shade, then the CRI would be 2. I see 2 places where an error is intoduced into the method :
The color samples are printed with pigments, which are strong colors and have narrow reflection spectrum, but most objects around us have wide reflection spectrum where only the overall balance shifts more or less towards some color. For example, Blue sky is nowhere near strong electrifying Blue, it is just Blue-r than neutral
Lamps with continuous emission spectrum (Incandescent, LED), even if it is not uniform from 400 to 850nm, would always cover the reflection spectrum of the test sample, allowing it to show its color (maybe darker than it really is, if it is off the peak emission of the light source, but still the correct shade)
Lamps with discontinuous emission spectrum (Most discharges) would miss any test sample which reflection spectrum falls inbetween the peaks of the light source reflection spectrum
However, for evaluating the overall balance of an object that reflects to some extent the entire visible light range, a light source which covers uniformly as much of it as possible (even if discontinuously, i.e. an array of sample points spread uniformly through the visible light range) would be preferrable over a light source with continuous but extremely non uniform spectrum
2. Since the 100 samples test is only a theoretical test and not performed in real life, its not clear how well the calculated values would even reflect its results, and how consistent would the error be (under different light sources which have different oddities, the test might have error sometimes in one direction and sometimes in the other)
Light quality is a qualitative measure (there is no scientific way to evaluate it), and depends not only on the light, the objects under the light, but also on its application. What's more, it is affected by the choices of lighting outside of the area where we are measuring the light quality :
If a shop and its parking lot are illuminated by the same White light, they look the same and both the good and bad aspects of the ligth look the same
If a shop is lit with MH and its parking lot with HPS, then when walking from the parking into the shop, crossing the door into the shop changes the light and it appears as higher quality light with lots of color information, so good for the application in the shop. When walking from the shop back to the parking lot, the light changes from White which shows many details in everything, to Yellow which lowers the level of distraction from the surroundings, lowering the quantity of information which is irrelevant to finding the car and driving it
If the shop would be lit by HPS inside (as well as the parking lot), it would be something like : When walking into the shop, we would very actively be using the correction process to choose our stuff in the shop under the Yellow light. The light is inadequate for the application, but lets say we succeeded in doing our business in the shop anyway. When walking back from the shop to the parking, there is no change in the ligthing, so the same looking into every detail and highly active correction process would possibly distract from finding and driving the car instead of helping it