Choosing a new desk lamp for digital processing.

[Alan Klein]

Sure, why not? The values of the bulbs don't necessarily match what comes out, at least the one's I've measured. The higher CCT that one might suspect are preferable due to the value themselves don't necessarily produce the most pleasing color rendition, the CCT 3500K bulbs look great. I find the CCT 4700K bulbs can appear 'too cool' depending on the task.

The Kelvin of the lamp is going to effect the light balance and colors depending on which lamp you use.  6500K is natural, 5000K is natural white, 3500K is warm.   Shouldn't you pick a lamp and its Kelvin rating based on the type of light you intend to show the prints under? 

http://www.planetbulb.com/pages/What-is-Kelvin-Color-Temperature%3F.html

Handy chart of Kelvin for artificial lights and outdoor sunlight at differing times.  http://www.wpsphoto.org/KelvinTemperatureChart.pdf

[digitaldog]

The Kelvin of the lamp is going to effect the light balance and colors depending on which lamp you use.  6500K is natural, 5000K is natural white, 3500K is warm.   Shouldn't you pick a lamp and its Kelvin rating based on the type of light you intend to show the prints under? 

I don't know where you got that idea about 'natural' anything and further it's important to keep in mind that any CCT Kelvin value is a large range of possible colors!
http://www.ppmag.com/reviews/200512_rodneycm.pdf

Showing your prints could occur under any number of different possible illuminants that are not anywhere near the illuminate of your viewing booth sitting next to the display. Do you think one has to (one can) match exactly the illuminant for the print and whatever is backlighting the display? Unlikely. And yet we can produce a close visual match between the two with some trial and error in how we calibrate the display white point.

[WayneLarmon]

Found a place online selling the Soraa 00807 indicating it should ship by July 31, 2015...

http://www.elightbulbs.com/catalog_product.cfm?source=AmazonCSE&prod=SC00807&pageurl=http%3A%2F%2Fwww.amazon.com%2Fgp%2Foffer-listing%2FB00P97UKCM%2Fref%3Dolp_tab_new%3Fie%3DUTF8%26condition%3Dnew&sku=SC00807

Thanks.  Ordered.

I've been studying some color science textbooks and it looks to me that there are lies, damned lies and color rendering indices.  The CRI metric is dicey at best.  (Page 216 of Wyszecki and Stiles's Color Science, 2nd edition shows the spectral graphs of two extremely spikey and bogus looking illuminants that are CRI 100.)   And CRI is all there is (officially).   

"In the eye of the beholder" is the best we can do.  I want to try those Soraa bulbs and see if they work any better for copy stand work than my "high CRI" CFLs.

Wayne

[digitaldog]

The CRI metric is dicey at best.

It's kind of a joke. It was primarily designed by Fluorescent light manufacturers to make their products look much better to unsuspecting consumers than they really are. There are BCRA tiles used to compare under a reference light source but only eight. That's too small a set of tiles. That make it easy to create a spectrum that will render the 8-14 tiles and doesn't tell us that the light source is full spectrum. It doesn't tell us how the other colors will render. My understanding is there are two reference sources; Tungsten for warm bulbs and D50 for cool ones: Above and below 4000K. That means that a normal tungsten bulb and perfect daylight both have a CRI of 100! As such, a high CRI is a decent gauge of how well a light will preform in your home but not such a great indicator of how well it will work for photography and proofing. Both a Solux bulb and a "full spectrum" tube from home depot may have a CRI of 97. I can assure you the Home Depot bulb has a giant mercury spike and some spectral dead spots.

[Alan Klein]

Andrew:  the reason I asked the question about Kelvin, was because I was thinking about my 25 year-old lightbox that I believe was set at 5000K so that slides would display at the closest to what is actually in the film.  Now is realize we're talking now about prints and reflected light, but wouldn't the light's Kelvin effect the colors you see in the print. 

In other words, if you are comparing a print with the monitor's colors and white balance, wouldn't these be different depending on which Kelvin you selected?  Just like 5000K was the usual standard I believe for lightboxes, is there a recommended Kelvin for a viewing booth light?

[WayneLarmon]

As such, a high CRI is a decent gauge of how well a light will preform in your home but not such a great indicator of how well it will work for photography and proofing.

Yes.  CRI (such as it is) is based on the eye's tristimulus response.   The tristimulus response of cameras is only an approximation of the human eye's tristimulus.

Both a Solux bulb and a "full spectrum" tube from home depot may have a CRI of 97. I can assure you the Home Depot bulb has a giant mercury spike and some spectral dead spots.

As I posted in another thread I got a ColorMunki Display + ArgyllPro ColorMeter to measure SPDs.   The first thing I measured was my Home Depot CRI 93 CFLs.  Yep, they are spikey.   (All the 5500K CFLs I measure had a CRI > 90, no matter whether they are marketed as being "full spectrum" or not. The garden variety 5500K CFLs from Home Depot (etc.) all have approx. the same spectrums and measurements of CRI > 90 as the Home Depot "Full Spectrum" CFLs do.)

However, I also measured the Solux 3500K 120V screw in lamp I have and it's spectrum was a bit gnarly.  It has a lumpy droop in the middle of it.  (Attached--click to see the graph.) This was why I raised my eyebrows earlier in the thread when you recommended them.  It is also CRI 86.

Wayne

[digitaldog]

Andrew:  the reason I asked the question about Kelvin, was because I was thinking about my 25 year-old lightbox that I believe was set at 5000K so that slides would display at the closest to what is actually in the film. 

I'm not sure about this idea that because the Fluorescent lights in your lightbox is 'rated' as CCT 5000K, your slides display what's actually 'in the film'.
I also had such light boxes and tried to use the same make and model as my E6 lab so we were on visual parity. That's useful!

Now is realize we're talking now about prints and reflected light, but wouldn't the light's Kelvin effect the colors you see in the print. 

AFAIK, print, display or slides on a box, it's all the same issue.

In other words, if you are comparing a print with the monitor's colors and white balance, wouldn't these be different depending on which Kelvin you selected? 

The numbers are mostly meaningless because the don't tell us the exact color (unlike say D65 or any other standard illuminant), the numbers don't tell us anything about the spectrum, the numbers don't guarantee a visual match. It's only one part of the descriptor.

Analogy: I can tell you I've got a pixel who's value is R45/G8/B120. Does that describe the color? No. It's only a partial ingredient, we need the associated color space. Saying we have a CCT (that's important) 5000K anything is about as ambiguous as saying "I've got a pixel who's color is R45/G8/B120."

[digitaldog]

However, I also measured the Solux 3500K 120V screw in lamp I have and it's spectrum was a bit gnarly.  It has a lumpy droop in the middle of it.

Using an EyeOne Pro and various packages, I can replicate that.

[WayneLarmon]

However, I also measured the Solux 3500K 120V screw in lamp I have and it's spectrum was a bit gnarly.  It has a lumpy droop in the middle of it.

Using an EyeOne Pro and various packages, I can replicate that.

Your plot looks more like an older (CRI 66) 5000K LED than the 120V 3500K screw in Solux bulb.  Are you sure that you posted the correct screen shot?

Wayne

[digitaldog]

Your plot looks more like an older (CRI 66) 5000K LED than the 120V 3500K screw in Solux bulb.  Are you sure that you posted the correct screen shot?

It's the now discontinued desk lamp from Solux (the black one).
The CRI numbers are not IMHO pertinent for reasons I've outlined already.

[Tim Lookingbill]

Yes.  CRI (such as it is) is based on the eye's tristimulus response.   The tristimulus response of cameras is only an approximation of the human eye's tristimulus.
 

As I posted in another thread I got a ColorMunki Display + ArgyllPro ColorMeter to measure SPDs.   The first thing I measured was my Home Depot CRI 93 CFLs.  Yep, they are spikey.   (All the 5500K CFLs I measure had a CRI > 90, no matter whether they are marketed as being "full spectrum" or not. The garden variety 5500K CFLs from Home Depot (etc.) all have approx. the same spectrums and measurements of CRI > 90 as the Home Depot "Full Spectrum" CFLs do.)

However, I also measured the Solux 3500K 120V screw in lamp I have and it's spectrum was a bit gnarly.  It has a lumpy droop in the middle of it.  (Attached--click to see the graph.) This was why I raised my eyebrows earlier in the thread when you recommended them.  It is also CRI 86.

Wayne

Did you or have you made a direct connection to how the spiky spectra of these artificial lights claiming high CRI affect perception of colors on a print viewed under these lights in preventing a match to a display that uses its own approximation visual system to counter the spiky LED/fluorescent backlight?

How can anyone claim accuracy according to a full spectrum standard D50/D65 lighting device by analyzing its spectra in order to get a match on another device that uses spiky or completely different lighting hardware?

[WayneLarmon]

Did you or have you made a direct connection to how the spiky spectra of these artificial lights claiming high CRI affect perception of colors on a print viewed under these lights in preventing a match to a display that uses its own approximation visual system to counter the spiky LED/fluorescent backlight?

Not really.   I haven't been doing Color Management as it is classically defined (matching a monitor to prints.)  Since I got my ColorMunki I've been dabbling with Color Management techniques a bit, but I don't print for real at home.  So I haven't tried to match prints with any real vigor.

How can anyone claim accuracy according to a full spectrum standard D50/D65 lighting device by analyzing its spectra in order to get a match on another device that uses spiky or completely different lighting hardware?

I don't think that a true D50 or D65 light source exists.  And real daylight is spiky if you look at it closely enough. Page ten of Wyszecki and Stiles's Color Science has a plot of daylight measured at .25 nm that is very ragged.

You know that LED monitors are spiky.  I'm attaching spectral plots of monitor white, red, green and blue.  This is of a Samsung 2333HD  monitor.  The spikes in my NEC PA241W look very different.

I think that the consensus that I read in my color science textbooks is that measurements, no matter how precise, are no substitute for matching by eye.  The human tristimulus mechanism isn't understood well enough yet to be replicated with instruments.

As Andrew has noted earlier in this thread, the CRI spec that is based on measuring eight (or fourteen) colors leaves a lot to be desired.

Wikipedia:

Ohno (2006) and others have criticized CRI for not always correlating well with subjective color rendering quality in practice, particularly for light sources with spiky emission spectra such as fluorescent lamps or white LEDs
https://en.wikipedia.org/wiki/Color_rendering_index#Criticism_and_resolution

With this said, it would be helpful to have better color rendering metrics.  They could at least try; instead of leaving us hanging with the existing CRI.

Wayne

[GWGill]

The CRI metric is dicey at best.  (Page 216 of Wyszecki and Stiles's Color Science, 2nd edition shows the spectral graphs of two extremely spikey and bogus looking illuminants that are CRI 100.)   And CRI is all there is (officially).   

The current "fix" to the limitations of CRI is to make sure that the R9 value is greater than zero.

[ R9 is the CRI saturated red patch, which is not part of the average CRI Ra value. ]

The lighting industry appears to have successfully sabotaged several attempts to update CRI, although TLCI seems to have snuck around this by being a "Television Standard".

[GWGill]

How can anyone claim accuracy according to a full spectrum standard D50/D65 lighting device by analyzing its spectra in order to get a match on another device that uses spiky or completely different lighting hardware?

Standards like CRI attempt to do that in a practical manner by evaluating the delta E's that will result when viewing colors under each spectrum.

Where the current CRI falls down is in the limited range of colors it uses (R1-R8 are all pastels), and the way it averages the individual patch delta E's together, allowing a few good patches to swamp a really bad patch. Something like TLCI attempts to address these problems, as do the other proposed replacements for CRI.

[Tim Lookingbill]

Seems to me the people we should be asking are those that have to match paint swatches to existing and/or older paint they sell to customers. There's nothing worse looking in interior design than to have subtle or pronounced paint mismatches on walls and a lot of that is compounded by mixed lighting in homes. Interior paint sellers must have to coddle a lot of picky customers and come up with excuses why their wall paint isn't matching.

That industry must have a better matching system than what's available to photographers or provided by existing color science matching and measuring methods.

[pluton]

I offer this, as a relatively inexpensive 'daylight' solution:
http://www.bhphotovideo.com/c/product/728827-REG/Kino_Flo_488_K55_S_6P_4_Kino_800ma_KF55.html

 They work in any  standard 4 foot flo fixture("shop lights" that you get a Home Despot, etc).
Yes, they are fluorescent, which means they have a discontinuous spectrum.
On the other hand, all LEDs have discontinuous spectrums, and tungsten-halogens with imperfect filters in front of them probably have fairly  discontinuous spectrums as well.
Just a suggestion...
Oh, and KinoFlo is a well-established brand (for over 25 years) in the motion picture and televsion industry.
Ask any cinematographer if they 'make good stuff'.

[Tim Lookingbill]

How is a $130 fluorescent tube considered relatively inexpensive?

[Ernst Dinkla]

Typical Olino.org lamp test report that gives more than the CRI value and goes into the details of CRI values on another page.

http://www.olino.org/us/articles/2013/12/18/installerdirect-com-ge-led6-5d-gu10-840-fl-bx-dimmable

Lots of similar lamp test reports there. Dutch site with pages in English.

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots

[howardm]

How is a $130 fluorescent tube considered relatively inexpensive?

Well, actually, it's a 6pack of tubes. 

You can also purchase individual replacement tubes from Just-Normlicht or GTI for approx $15-20 each

[GWGill]

Seems to me the people we should be asking are those that have to match paint swatches to existing and/or older paint they sell to customers. There's nothing worse looking in interior design than to have subtle or pronounced paint mismatches on walls and a lot of that is compounded by mixed lighting in homes. Interior paint sellers must have to coddle a lot of picky customers and come up with excuses why their wall paint isn't matching.

They don't have these problems because they do spectral matching. The match will be good under almost any lighting.

(And note that CRI is not about matching - it is about the appearance of the colors compared to what you would see if you were using a broad spectrum illuminant).