Flashlight Tint/Color Reference

[McGizmo]

Electron Guru,
If I am following this correctly which is no certainty, I would assume that if you took some shots with your camera (lens covered with the ExpoDisc) of an incandescent source and bracket the shots with manual settings of CCT, as you approached the actual CCT of the light source, the recorded image would approach "white" ? Yes? Or does the image approach the spectrum of color one sees at that color temperature?

Does the manual adjustment of the camera attempt to filter out the tint and give you a white appearing white surface when you have set the color temp correctly for your light source or does the camera attempt to give you an image of a white surface tinted to match the spectrum of a black body, at that kelvin level?

 

[ElectronGuru]

I would assume that if you took some shots with your camera (lens covered with the ExpoDisc) of an incandescent source and bracket the shots with manual settings of CCT, as you approached the actual CCT of the light source, the recorded image would approach "white" ? Yes?

If we were talking about absolutes from the flashlight, yes it would be white. But in the context of computer displays, there is an extra complication. In nature, when more light is added, it just gets brighter. In a computer, white is already max, so the only way to show bright (accurately) is for everything else to get darker. So adding color reduces the brightness of a given sample. To be on the same K scale as the colors I'm showing then, white has to be darkened to gray.

Here's an example Mag85 (incan) with a measured color temp of 3500K:

The left half is what the camera sees, with 21 different K settings, starting with 2500K and ending with 4500K (zero plus or minus 10 whole steps). Down the center is a pure gray reference. Side note: you may notice that at 3500, the actual color is still a tiny bit blue. But 3600K is just a tiny bit red. Even 3550K looks a bit off, the true color the light is putting out is between the two values, but K (expressed in 00's) is not precise enough, so we accept it.

The right half is what happens when the same values are adjusted so that 3500K is near white. But as you can see, as the brightness reaches the computers ability to display pure white, the colors are almost pure white as well (as we approach maximum brightness, colors get washed out and look rather :sick2.

So to answer your question, as the CCT approaches the value of the light being measured, the image appears gray.

Or does the image approach the spectrum of color one sees at that color temperature?

The spectrum is there, but as the goal of this approach is to show the tint, the tint is all you see. With a camera, the only way to show the spectrum of color contained in a light source is with a prism or by shining the light source on something colorful and showing which colors render/reflect.

Does the manual adjustment of the camera attempt to filter out the tint and give you a white appearing white surface when you have set the color temp correctly for your light source or does the camera attempt to give you an image of a white surface tinted to match the spectrum of a black body, at that kelvin level?

From the point of view of the camera sensor, they are the same thing. Cameras recording light are much simpler than what you have to deal with, coordinating the output of a given LED with the perception of the human eye. Since all the camera needs to do is calibrate to achieve white/gray, a black body radiating at 6000K blue is just as easy to coordinate with as an LED radiating at 6000K green. Configuring the camera to match 6000K will neutralize both tints with equal ease.

Part of the confusion here is that I have introduced white balance disks at the same time as a new technique that mis-uses them. Normally, they are used to erase tint. Figure out what the tint is and calibrate to it, the camera's K moves about to adapt. Here, I have frozen the K setting of the camera, so the tint is what your actually viewing.

Think of an ExpoDisc like a grey card, except instead of measuring the reflection of light from a given surface, you are measuring light on what would be, its way toward the grey card. As with a grey card, the goal with using a Disc when taking a photo of something is to match the CCT of the camera to the CCT of the ambient light, so the final photograph appears as though the ambient light was white.

Ironically, most of this came about because of photographic improvements. Back in the dark ages, we used film. Film was calibrated to sunlight and we used it outside all over the place in natural light. It wasn't sensitive enough to take indoor shots without a flash, so indoor shots were calibrated to a single known source, the flash. And even when color was messed up, someone else was doing the developing and often correcting the color for us. Then along comes digital. Suddenly we are doing everything ourselves, but more importantly, we now have the ability to take indoor shots with ambient light. And as it turns out, the CCT of indoor lighting is all over the place, creating this problem. But I can't stand flash photography, so to me, this is a good problem.

BTW, here's a good primer on photographic white balancing:
http://www.ronbigelow.com/articles/white/white_balance.htm

 

[McGizmo]

So to answer your question, as the CCT approaches the value of the light being measured, the image appears gray.

Thanks that is what I figured. Now if there is any residual tint then I would assume this is due to the light source being viewed having its x,y coordinates off of the Planck curve.

I believe I am following to a reasonable extent what it is you are doing here but I admit I am not clear as to what is being accomplished but no matter.

With the integrating sphere and spectrometer I have, I can reasonably integrate the light source (as you are doing with your ExpoDisc) and further I can get a CCT for the source as calculated by the equipment. I could then set my camera at the appropriate kelvin number and photograph the light source through a ExpoDisc and capture an off white (or off gray) image. This would indicate a tint variation or deviation beyond a correction for color temperature, I guess. Now I understand that this exercise would have some significance in regards to the particular sample LED I was working with but as to how well this LED represents other LED's of the same bin or resident in fellow flashlights is still debateable.

I guess another issue here for me is the fact that our eyes do make adjustments as we acclimate to a light source and I have no idea if we adjust both for color temperature as well as tint or not. My guess is that our personal adjustments can vary from individual to individual and it may not be a simple adjustment based on absolutes or the measures we make with our technical equipment (cameras, light meters, spectrometers, etc.). My gut is that we have our own means of white balance adjustment. Once we do adjust for white balance, we are still dependent on the light source and its ability to render the spectrum of colors out there in the field of view. I know that I personally can't correct to the point that I recognize a purple as a purple if the light source is too far off in its red and or blue output levels. At least I don't think I can but I may not have given myself time to acclimate to a light source and then revisit a purple object. :shrug: :thinking:

We have evolved in the presence of black body sources but who knows how critical it has been in our survival and resultant in our ability to compensate and make perceptual adjustments. We can put on a pair of tinted sun glasses and in no time no longer be aware of the tint. By the same token, we can use a light source that has an obvious initial tint and yet soon be essentially unaware of it. This is particularly true if we use the light source for viewing a range or real world objects and don't limit our focus to a white wall, for instance.

Beyond our ability to adjust and compensate, a given light source may still fall sort in its ability to allow us the perception we desire. There may be insufficient illumination in terms of overall lux or there may be insufficient light in specific spectral bands (not enough red from a LED or not enough blue from an incandescent).

When members of CPF express dissatisfaction in a particular light source and attribute it to tint or the perceived warmth or coolness (color temperature) in some cases this is based on white wall observation but often this is based on real world targets and in either case, the member has the right to be dissatisfied.

Is there some way that we can identify the nature of a light, a member has yet to see, that will aid them in determining whether the light will meet their needs and standards? This seems to be a worthy goal and I suspect part of the intent, with this thread.

I appreciate your responses to me here and I don't want to sidetrack you any further than I have already.

 

[ElectronGuru]

With the integrating sphere and spectrometer I have, I can reasonably integrate the light source (as you are doing with your ExpoDisc) and further I can get a CCT for the source as calculated by the equipment. I could then set my camera at the appropriate kelvin number and photograph the light source through a ExpoDisc and capture an off white (or off gray) image. This would indicate a tint variation or deviation beyond a correction for color temperature, I guess.

Since I included numbers to create my scale, its easy to get caught up in them, but relying on them to this degree is making things overly complicated. If the goal is to precisely calibrate a camera to the projected tint, CCT as expressed in hundreds (00's) is not accurate enough. Capturing a CCT, inputting that value into your camera, and then taking a picture based on that CCT is not as accurate as just using the camera's own fully-internal system. Here's another test: Take a sample of a Mag85 with the camera set to 4300 (top), then calibrate the camera (aka, custom white balance) to the first photograph and take a second sample (bottom).

Running down the middle of both is another reference gray stripe. Measuring the two areas with Photoshop, there is a 1% variance (toward blue). But with my naked eye, I cannot see a difference between the reference gray and the rest of the bottom half. This, compared to the test in my last post shows that using the camera's built in color system is more accurate than setting the camera to 3500K and doing the same thing.

The larger point here is that everything I am doing here is about photography, visually recording what the bulbs and emitters are doing. Trying to use this qualitative measure as a quantitative one will quickly exceed the capabilities of the camera. In both cases, any variance from true color is about the limits of the camera/settings, not a deviation of the bulb.

We have evolved in the presence of black body sources but who knows how critical it has been in our survival and resultant in our ability to compensate and make perceptual adjustments. We can put on a pair of tinted sun glasses and in no time no longer be aware of the tint. By the same token, we can use a light source that has an obvious initial tint and yet soon be essentially unaware of it. This is particularly true if we use the light source for viewing a range or real world objects and don't limit our focus to a white wall, for instance.

Anecdotally, I would attribute this to the same biological tendency we have to smells or sounds. We are relativistic beings, so adjustment to new 'normal' is necessary if we are to perceive a new 'different'. It would badly suck, for example, if when the sky got twice as bright, we could no longer see traffic signals. Being unable to perceive absolute [anything] is a small price to pay for these abilities.

When members of CPF express dissatisfaction in a particular light source and attribute it to tint or the perceived warmth or coolness (color temperature) in some cases this is based on white wall observation but often this is based on real world targets and in either case, the member has the right to be dissatisfied.

I agree. The trouble with numbers is that they are abstract. Just as we understand graphs more completely and easily than a table of numbers, having to imagine what 3300K looks like is also an obstacle to understanding. My hope is that by showing a color, alone and in context with other colors, that we can more easily imagine what it looks like and predict its performance.

Is there some way that we can identify the nature of a light, a member has yet to see, that will aid them in determining whether the light will meet their needs and standards? This seems to be a worthy goal and I suspect part of the intent, with this thread.

The problem is actually one of labor. Were it practical, you could just profile every emitter in every light before sale and every customer would know what they are getting. But this is impractical even with a small volume. In your case, if a given bin was narrow enough in color/render variation, we could show sample performance for that bin and people viewing that sample would know what to expect. If not, we would either show the variation possible in a bin (map the lottery) or create our own more accurate bins.

 

[ANDREAS FERRARI]

Re: Flashlight Color Temperature Reference

There are members who speak with authority on the lights and yet it is clear to me that they don't really understand the difference between lux and flux.

I agree-CPF has more then it's share of members who are either delusional or have a high level of self-importance.

ElectronGuru-despite the number of valid concerns about your project I have to applaud your creativity.I always appreciate it when someone comes up with a new concept instead of just doing 'more of the same'. :twothumbs:twothumbs:twothumbs

 

[souptree]

I confess I get lost every time I read one of these color threads. :thinking: This is NOT helping with my sense of self-importance.

Oh look, a bunny!! :nana:

 

[Patriot]

I confess I get lost every time I read one of these color threads. :thinking: This is NOT helping with my sense of self-importance.

Oh look, a bunny!! :nana:

You're not alone!

I know this is going to be helpful to me in some way but after the first read through it's mostly above my head. There are obviously one or two base concepts that I'm unable to wrap my brain around. For example I'm looking at the charts and they're telling me what white is but yet I look at the actual web page where the text is and it looks "whiter" to me than any of the chart examples. I'm going to have to go through this all again.

Thanks for the thread E.G. It looks like you've put a lot of thought and work into it.

 

[jtr1962]

I just came across this thread and I'll agree it's going to take a while to wrap my head around exactly what is being done here. My interpretation (and correct me if I'm wrong) is that a light is shined at a neutral background (meaning it returns all colors equally), and the camera's Kelvin setting is adjusted in 100K increments until the color of this neutral background appears gray. The Kelvin setting where the background appears gray (grey for those on the other side of the pond) is the CCT of the light source. The best way to determine when the background truly appears neutral would not be by eye, but rather when the red, green, and blue components are equal. In practice this may not always be possible if the light source lies outside the black body locus. In that case, the closest shade of gray must be determined by eyeballing it.

I really need to take some time to reread this entire thread and digest it before commenting further. I just wanted to make sure I'm interpreting the basic methodology correctly.

 

[Bullzeyebill]

As much as I decry subjective information, and long for good objective info here on CPF, I am in a quandry about this whole issue about tint/color. I am now thinking that what I see and what others report about what they see, relative to color and tint reflected by an LED or and incan light, is more important than so called objective info as offered by the OP or others. In effect, let my eyes, or your eyes, do the telling. For simple old me, I can go with LEDs in the daytime and incans at night, and so what about CRI and CCT.

Bill

 

[ElectronGuru]

Thanks for the thread E.G. It looks like you've put a lot of thought and work into it.

I thought about it in my head for nearly a year, then developed the actual process over two months. So if you want to understand the theory/process (optional), its going to take a bit to catch up, but I think its worth the trip.

My interpretation (and correct me if I'm wrong) is that a light is shined at a neutral background (meaning it returns all colors equally), and the camera's Kelvin setting is adjusted in 100K increments until the color of this neutral background appears gray.

Don't overthink this, its simpler than it looks. The 100K increments are not used in creating the finished samples, that is only explaining the the process behind the results, sort of like "showing my work" in a math class. At no point is reflection taking place. At no point is the color of any object coming into play with these measurements. Here's a view of the physical steps:

ONE, setup the camera as a target for the flashlight

TWO, place the expodisc in front of the lens (this is the thing that blurs the light

THREE, shine the light source at the target (camera)

FOUR, take the picture

​

The light and color of the light is recorded directly by the camera. The disc does not modify the color in any way, it only smoothes it out (makes all parts of the image the same light/dark'ness) so that the sample is easier to view.

Think of a camera set to 4300K as a pure representation of our eyes perception of color. Once set, anything that looks yellow to the eye will look yellow to the camera and anything that looks blue to the eye will look just that much blue to the camera - and in the resulting photograph. Point the configured camera at light that looks green to you, the photo will be green. The complicated part was getting the camera to mimic the eye. Now that its there, the results *should* be nearly intuitive. We've just got all these :hairpull: numbers confusing everyone.

I am now thinking that what I see and what others report about what they see, relative to color and tint reflected by an LED or and incan light, is more important than so called objective info as offered by the OP or others.

Our eyes weren't build to measure, they were built to perceive. The camera and photos it produces are accurate perceptions - its the K #'s themselves that are limited to black body representations. I wish I had left Kelvin values off my samples, instead of helping folks relate to what they already knew, the numbers have only confused and obscured what I want everyone to see:

The color in a sample picture is not constructed. Its what the camera actually sees/records and is all you need to look at to intuitively understand the color of the light that a given light source is producing.​

Anyway, thank you guys for having the courage to speak up. This gives everyone a chance to learn more.

 

[ElectronGuru]

I'm looking at the charts and they're telling me what white is but yet I look at the actual web page where the text is and it looks "whiter" to me than any of the chart examples.

White (as used in English) has two characteristics, being the most intense (bright white) and having the least color (along with black). In this guide, I use gray to represent white (the color) because of a specific limitation in our computer displays. In reality, if light gets more intense, we see it as brighter. But displays are limited to a single uniform max brightness, so its simply not possible to represent correct brightness and correct color at the same time.

Think of it like listening to recorded music, where the song suddenly gets twice as loud. The system playing the music can only accurately produce the volume increase if the first part of the song was playing at half of whatever the system was capable of producing. Its the same with recorded images, just with the intensity of light.

Borrowing a graphic from post 22, the left half is how color is represented here. The right half is what happens to those colors when we try to get white to show up as white. As you can see, everything non-white also turns white. Colors wash out as they approach brightness equal to the intensity of white, killing the usefulness of the samples:

 

[The Coach]

I just found this. Some really good info. Needs to be brought up again. :thumbsup:

 

[benthiccracker]

I just found this. Some really good info. Needs to be brought up again. :thumbsup:

+1

 

[5S8Zh5]

Wish those pics were still active. A great thread nonetheless. Thanks.