Don't give up on fluorescent lights quite yet, although giving up on the cheap import stuff could be a wise move. Lately I've been experimenting with several different 4 foot linear bulbs available through various sources, and one seems to be a potential candidate.
For reference, here is a photo of a Macbeth color checker made under a 50 watt, 12 volt MR16 halogen track light. The color rendering index (CRI) of this bulb approaches 100.
Here is the same scene, illuminated by a Philips T32F8/TL930 fluorescent bulb. This bulb has a rated CRI of 95 and a color temperature of 3000K. By the way, the camera white balance in Adobe Raw was set to precisely the same settings as in the previous photo. The exposure time was different.
When comparing these two, except for the shadows they should appear fairly similar. The halogen produces a realistic sunlight shadow but the fluorescent does not (which could cause some modelers, in a futile attempt to better simulate reality, to dry-brush fake looking highlights on models.)
Now for an acid test. I digitally cut out small squares from the halogen photo and overlaid them on corresponding squares of the fluorescent photo. Got that? The big picture was shot under the test fluorescent lamp, and the tiny cut from the halogen illuminated photo just above.
Now for the comparison. How obvious are the cutout pieces?
From my experience of doing industrial color for over 15 years, this is an excellent visual appearance match under affordable lights. The light from the halogen flood lamp was not as even as that from the long fluorescent lamp. Some, but not all of the patches that do not match in brightness can be attributed to that. Also note that the halogen appears slightly greener; hopefully field testing on a layout will not show an objectionable difference.
I have not yet photographed a mix of this fluorescent lamp with ordinary tungsten track light fixtures. Since ordinary tungsten filament bulbs
(i.e. "real light bulbs") are yellower than a halogen, we would anticipate somewhat-directional tungsten light to be yellower than fluorescent fill light, just as the sun is yellower than sky fill light. If so, that could be very good, if the visual color shift was moderate. The shift between cool white fluorescent (4000K) and tungsten (about 2800K) is too extreme to be visually believable.
Also note that the normal fluorescent lamps available from the "home store" have a CRI of 74 to 85. They produce a color appearance far from the more natural look of colors under CRI 95 or greater. Color under a CRI 74-79 lamp appears to my eye quite "dead" and dull (and I seem to become a sort of "zombie undead" under all cheap but common fluorescent lights!)
I have yet to measure any spiral compact fluorescent (CFL) lamp that was anything but extremely poor junk. That is not to say that a good CFL does not yet exist, but it does not (perhaps yet) exist in my personal experience. Surprisingly, the brand name from major brands to cheap in-house brand, did not matter. For the ones I bought and measured, they were all the same poor color but sold at different price points. When I measured their emitted lights spectrum (SPD in color lingo), all bulbs I tried at a given color temperature might as well have come off the same low-cost production line.
What's the next step? Field testing. If the layout owner approves, we move forward. Otherwise, I have a lifetime supply of bulbs for my home metal shop!
