[BigMackCam]

Staff note: This post may contain affiliate links, which means Pentax Forums may earn a small commission if a visitor clicks through and makes a purchase. If you would like to support the forum directly, you may also make a donation here.

---

So...

I'm wanting to create my own dual illuminant camera profiles for various cameras that I own, and software that I use, via a couple of different software approaches. I've done this before, but always using whatever estimated natural lighting conditions I've had available to me at the time. This worked OK, however... here in the UK, especially at this time of the year, it's very difficult to find natural lighting at 6500 and 2850k.

For some time, I've been thinking that an artifical light source in a darkened room would provide better conditions for the colour chart reference shots I need in creating these profiles.

I've come across a low cost product that I might be interested in, here:

Aputure AL-MX 2800-6500K CRI95+ LED Video Light Lamp for DSLR Camera Camcorder 635946656549 | eBay

This is an LED panel lighting source that allegedly offers stepped adjustment of temperature between 2800 and 6500K.

Given the rather low price, I'm assuming the colour temperatures specified aren't especially accurate, and yet I wonder if they might be good enought to provide illumination that's approximate enough for my intended application?

Alternatively, I'd be interested to know of other options. I could afford to spend more, but would rather not spend a lot more unnecessarily. Then again, I'd rather spend nothing at all if the budget approach is going to be useless...

Thanks in advance for any views, opinions and ideas

[BrianR]

I was looking at that light the other day, it looks like a beast. I've no idea how accurate it's colour temps are, but I'll add in another option - lighting gels. Rosco has dirt cheap swatchbooks *and all the gels have fancy little graphs with details on how they transmit light. If you start with a known light source, it might be more accurate way of getting a range. It would be limited to the steps of the gels of course.

*I point this out knowing that shipping to England may make a single cheap item not cheap. I tacked a set of these gelpacks onto a larger order I was already making from BandH years ago otherwise it wouldn't be worth it for this alone to Canada either

[sqrrl]

Some LED's, especially cheap ones have distinct spikes in colour spectra when viewed via a photo spectrometer - they blend several narrow bands of colour to balance at the number they quote.

For a cheap solution I'd be inclined to run with off camera flash (specifically speedlights) and get a couple of books of lighting filter samples to gel them to the temperature you want. Strobes tend to be broad spectrum like daylight, sample filter books are a good size for speedlights, and don't tend to be very expensive for the 100 odd filters you get.

[swanlefitte]

I looked into flash color accuracy a while back. It's imprecise. Even measuring it is. I had found a group doing this. The brand specs were off, light measurements varied by tool. Even what is defined as daylight is pegged to noon on the equator on a specific day. (Strobist talked about that)

So what is close enough? I suggest shooting outside at different intervals. Then see what daylight shifts become too much. That should get you a ballpark guess as to how accurate you will need for it to be worth it with little time and no expense.

[swanlefitte]

Found this here I thought 5600k was daylight color temperature?!!



"Daylight in motion picture film world is 5600/5500 degrees kelvin. We (film industry peoples) made it the standard because some guy in some point in history determined that is what the average daylight color temperature is... although that number is not accurate most of the time, the standardization is important as it allows all daylight lighting in film/tv to be close to 5600 so our lights match other lights and daylight balanced film stock... and viceversa.

But in the real world, the light outside can vary tremendously based on weather, time of day, geological location, etc. That is why whomever came up with the daylight standardization may not match what actual daylight is in other parts of the world or in different weather.

Daylight can vary by thousands of degrees kelvin. Thus, lighting manufacturers who manufacture or incorporate non-tv/film bulbs into their products which are not specifically made for tv/film can pick whichever color temperature in the appropriate daylight spectrum they determine as daylight... in this case the bulb manufacturer thought 5000k would be a better fit in homes and etc.


...the light(s) you have linked to above have the lighting fixture/head and accessories designed for film/tv use (although not professional) but they utilize standard non-tv/film manufactured compact florescents bulbs, like the ones you buy at home depot and replace over the kitchen sink. I'd go a step further and say that those bulbs will also not be completely treated to have a balanced green/magenta hue, thus you will most likely have an off spike of color in that realm as well.

So that is why. They made a cheap light housing that uses cheap household bulbs, which are made by manufacturers who deemed daylight as something just a bit off what standardized motion picture lights and cameras deem as daylight."

I also read an article by the naval academy that tested daylight in Spain and the chroma of the sky. Seems there are 4 perceived astigmatic perceived levels. So if you photograph someone from 4 points even with the sun high you can get 4 variations depending if the light is more eastern or western angled up or not. Not to mention the bounce off colored objects.

[swanlefitte]

The reason I wanted defined white light or a defined derivative was to see if my nikon was cooler and my 77 did produce the truest color. I decided my purpose was a rabbit hole. Lens rentals has expensive equipment and did this for the expensive lenses.
Lens Rentals | Blog

[Dartmoor Dave]

I just use direct mid-day sunlight, then use the eyedropper to correct the white balance to the white patch on the test card and save the raw file with that WB. It's not a laboratory grade approach, but for creating dual-illuminant .dcp profiles using Adobe Profile Editor it works well enough for me. I actually base my profiles on test cards shot at noon on June 21st, because the variation in the ambient light from pure daylight in my part of the world is what I'm interested in capturing in my photos.

Creating ICC profiles is more difficult and I've never done it. Custom profiles are so crucial to my own approach to digital photography that I'd only ever use a raw converter that could handle .dcp.

[kaseki]

Swanlefitte:
The color temperatures applying to various things have origin in measurements made a long time ago and published in various optics and electro-optics publications. The sun's color temperature, determined by a best fit to a blackbody curve is shown as 5700K in the GE Radiation Calculator. The best-fit blackbody temperature for solar irradiance outside the atmosphere is 5900K (RCA Handbook). The peak luminous efficacy of the human eye viewing blackbody radiation occurs for a 6600K blackbody (RCA Handbook). Blue sky can be fitted to multi 10k blackbody curves depending on conditions. No doubt that there is some scattered/reflected illumination condition observable outdoors at some sun angle and natural environment that is best fitted by a 6500K black body curve. Establishing monitors to have a particular color temperature is only relevant to tri-stimulus eye response and I suspect to some historical decision I haven't yet seen described; camera focal planes may have different tri-stimulus wavelengths so any generalization about them and their white balance is suspect.

BigMackCam:
Thanks for that link. Here is one to a source source I found in preliminary Internet trolling for my own recent question at:

Correct illumination for use with a color target - PentaxForums.com

FilmGrade? HYBRID LED Flex Panel ? Waveform Lighting

There is enough data in its associated PDFs that one could decide if it was good enough. A back-of-the-mind calculation last night convinced me it could illuminate a x-rite Passport sufficiently to reflect 120 nt from a white square. (But full disclosure requires that I admit that it was very late at night.)

Edit: Link to Aputure products:
https://www.aputure.com/collections/led-video-lighting

Note: To reflect 120 nt (lm/m^2-sr) from a pure white surface, multiply by pi to yield the required incoming irradiance, viz., 377 lux (lm/m^2)

Looking at the Aputure product line, I think the mini20c with 720 - 4200 lx at 1m depending on beamwidth would be best for my x-rite Passport illumination interest. The barn doors could be used to keep the light off of the monitor if desired.

[PDL]

@Dartmoor Dave
Check out the ColorChecker Passport beta that is out to create ICC profiles:
X-Rite Custom Camera Profiling for Capture One Pro - PentaxForums.com
You should be able to find the appropriate settings for
ICC profile and Curve default settings in your favorite software. The ICC profiles created using this method should be usable in any image processor.

[Dartmoor Dave]

@Dartmoor Dave
Check out the ColorChecker Passport beta that is out to create ICC profiles:
X-Rite Custom Camera Profiling for Capture One Pro - PentaxForums.com
You should be able to find the appropriate settings for
ICC profile and Curve default settings in your favorite software. The ICC profiles created using this method should be usable in any image processor.

Thank you, that looks like it's definitely worth trying. The main reason I'm still using Camera Raw/Photoshop is my need to use .dcp profiles, so a user-friendly way of creating ICC profiles could finally liberate me from the clutches of Adobe. And that would get rid of the last reason why I'm still using Windows, so I could go full Linux.

[Class A]

Some LED's, especially cheap ones have distinct spikes in colour spectra when viewed via a photo spectrometer - they blend several narrow bands of colour to balance at the number they quote.

+1

Even better LED-based lighting solutions are typically inferior to incandescent lights or regular flash when it comes to spectrum smoothness.

Even relatively high CRI figures like >97 don't necessarily mean a lot as the lowest CRI bar is very low indeed, using patches with low saturation levels only.
There are indeed several CRI standards and even competing standards which are much more demanding on light sources than the low-bar version of CRI.

I would not base any profiles (dual-illuminant or not) on an LED light unless you want to specifically use them with this LED light. With all other light sources (including natural daylight) you will have to expect more or less significant problems due to the bumpy if not spiky LED spectra.

[kaseki]

+1

Even better LED-based lighting solutions are typically inferior to incandescent lights or regular flash when it comes to spectrum smoothness.

Even relatively high CRI figures like >97 don't necessarily mean a lot as the lowest CRI bar is very low indeed, using patches with low saturation levels only.
There are indeed several CRI standards and even competing standards which are much more demanding on light sources than the low-bar version of CRI.

I would not base any profiles (dual-illuminant or not) on an LED light unless you want to specifically use them with this LED light. With all other light sources (including natural daylight) you will have to expect more or less significant problems due to the bumpy if not spiky LED spectra.

A few comments: Incandescent sources have smooth blackbody spectra, but cannot reach anywhere near 6500K. Flashes can manage that color temperature, and if driven hard enough can have smooth spectra (electron heating of the xenon gas can get it hot enough to overcome its intrinsic line structure), but flashes are not suitable for any eyeball evaluation of color. Other than LEDs, one needs a continuous arc source.

The Cermax line of xenon arc lamps, which have been manufactured in various forms via heritage companies since at least the late 1960s, seems to be available as used items on eBay for relatively low cost. However, these need a power supply that can generate ca. 20 kV to ignite them, with an intermediate voltage to get them into their normal operating range, and then a current-controlled ~20V source at moderately high current to keep them in their operating zone. The lamps by themselves have a negative resistance V-I characteristic at their operating point, so the power supply is more complex than those many may have experience with. The historical 1-inch lamps use 300W, but modern lamp models are available from 150W to 1000W or more.

http://www.excelitas.com/pages/product/Cermax-MX300.aspx illustrates an example. Excelitas seems to carry some of the historical line along with power supplies. Both the collimated (parabolic configuration) and the projection type (elliptical configuration) seem to still be in production. Note that even these 300+ PSI xenon fill lamps do not have perfectly smooth spectra, and as a result the CRI can fall between 95 and 99 per http://www.excelitas.com/Downloads/Cermax_Eng_Guide.pdf. I wonder just how different a color checker would look illuminated by a 95 CRI compact arc and a 95 CRI LED array.

Edit: Caution: These lamps are a significant UV hazard and need supplimental UV protection.

[Class A]

Incandescent sources have smooth blackbody spectra, but cannot reach anywhere near 6500K.

Yes, but the OP was enquiring regarding dual-illuminant profiles and incandescent sources could well play the part for the low temperature component of the dual-illuminant profile.

...but flashes are not suitable for any eyeball evaluation of color.

Interesting. In my own experience, test shots of colour checkers with flash look remarkably similar to test shots done in daylight (white balancing may be required, depending on daylight conditions).

It is very well possible that I wasn't looking hard enough. I think it is fair to say though that most commonly occurring light sources differ far more significantly from each other than flash emitted light and daylight do.

I wonder just how different a color checker would look illuminated by a 95 CRI compact arc and a 95 CRI LED array.

I'd expect differences, given that 95 CRI does not mean that much in particular regarding perceptual differences (hence the efforts around R96a, CIECAM02, etc.).

Some LED panels would come closer than others, but I understand the respective CRI numbers would not be a very reliable indicator, meaning even a high scoring CRI source is not necessarily an actual high performer.

[BigMackCam]

Thank you, everyone, for the useful responses and information. Most interesting... I hadn't understood about spikes in the colour spectra with LEDs.

It seems like I might be better served using off-camera flash and gels - and it's a cheaper solution, too, since I already have the flash units

[kaseki]

Originally posted by kaseki ...but flashes are not suitable for any eyeball evaluation of color.

Interesting. In my own experience, test shots of colour checkers with flash look remarkably similar to test shots done in daylight (white balancing may be required, depending on daylight conditions).
....

Sorry, I think I was commenting w.r.t. my thread in the digital processing sub-forum. I can't compare a physical color chart illuminated by a flash with an on-monitor rendition of a chart. Monitor renditions of flash photographed charts vs. sunlight photographed charts might well look very similar, even if there are a few xenon lines still poking up and the sun through the atmosphere is not itself a perfect representation of a black body.

[If I can get Xsane to work, I'll provide an example solar plot.]