Color temparatures, sky color, ect. and COLOR SPACE

[pailhead]

Look, no offense, but i'm not looking for poetry or philosophy, i just need an easy way to convert the RGB color of a kelvin into linear.

Right now there is a solution for vrayIES, and thats the "linear converter" that someone posted above. But it still takes a lot of copying and pasting... if you want, this is what's holding me down artistically. Imagine having to trim and assemble your own brush every time you want to make a stroke on your painting.

[pailhead]

Ok, these are my final questions, because i can't waste any more time on this if it's all in vain. Thank God that Westlund wrote their linear converter and made the job at least somewhat easier.







This is a kelvin scale that i made from the vrayIES dialog, using no gamma/LUT settings whatsoever.

Hmm, maybe I don't understand something correctly here. The Kelvin temperature defines what light spectrum is emitted by a light source, not how warm or cold this light looks to you. It is an absolute measure rather than a subjective one.

1. Is it correct to say that the scale above features a subjective measure of RGB color relating to a temperature reading in kelvins?

2. Are these readings objective and absolute when it comes to human perception of warm and cold colors? I see 400k as red, thats warm to me. I see 6000K as almost white, that seems neutral to me, while i see 8000K as blue which is cold, at least to me.

3. Is the scale above relative to something? 3200K reads as RGB above, i understand that this can be different, but wouldn't it be relativeto something else then, like a lense filter, white balance, ect.?

If you have a bulb that emits light at approximately 3200 Kelvin spectrum, you can take many different photos of this light, with different white balance settings - each of them will produce an image with different color tint, but that does not mean that the light source has somehow changed.

4. If it's so relative and cannot possibly be measured then why is there a color patch corresponding with a temperature change in the vrayIES dialog? It has to be relative... to something right? For example, why isn't 8000k in the scale above showing as red, along with an explanation "well, you can actually shoot 8000k with such a white balance and lense filter that it can actually come out as red".

5. Why is 8000K represented by R150 G205 B255 if it's actually a light spectrum value?

6. Why does 8000K become R200 G231 B255 when i start using LWF? This changes the appearance of an image without using different camera settings - the only given explanation for this appearance change in real life.


I honestly think that there is something wrong here. I am not content with the explanations given, thus i'm posting these questions. It took me a great deal of time to come up with all of these illustrations, and to write all of these explanations, so please respect that.

[rmejia]

thoughts on this

Well, this is just, like, my opinion man...

I too have had many questions while using LWF as to how far to take gamma correction, what do I need to gamma correct. Do I need to apply it to everything? The two renderings with the light colors, the one with the real color looks good, the one with the gamma corrected blue looks wrong, so I would conclude that gamma correction does not need to be applied to light colors. The things that seem most washed out when I started using the sRGB, LWF, gamma 2.2, etc ... were textures (images). So what I concluded is that the textures I use already have gamma correction applied to them. Most pictures taken with digital cameras are in sRGB color representation. Most of these images look good on my monitor (not too dark) because they are gamma corrected. Then in max, they were being gamma corrected again (twice). Since they already had gamma correction applied to them, the inverse gamma had to be used in max, so the images (textures, bitmaps, jpgs, etc..) would show as the originals, not double gamma corrected (noticeably washed out).

So in conclusion I think that only things that have been previously gamma corrected (images most of the case) need to be gamma corrected in max or else they will looked really washed out (because they will be double gamma corrected). I picked RGB color values from photoshop to use in max, and they have to be inverse gamma corrected in max, so it seems photoshop works in sRGB space.

Reflections, refractions, light colors, etc... I do not think have to be gamma corrected. It's like taking a picture with a digital cam. Gamma correction is applied to the final image, but none of the materials where changed like in a 3d application. So when a render (picture) is taken in a 3d application, none of the materials or colors have to be changed either, only the ones that have had gamma correction previously applied to them, so everything ends up in the same space.

[pailhead]

I took a normal max photometric light for comparison, and it has a completely different representation of kelvins.

On default, it maybe even less saturated than vrayIES in sRGB. When you apply 2.2 gamma, the distinction is gone, it's so pale that you can hardly even tell the difference unless it's extreme, like 3000k compared to 8000k.

Then i tried rendering one half of the image using vray-ies, and the other using max ies but its.. just... weird.

[ruud3dv]

You definitely spend a lot of time and effort explaining what you want to know. Respect. I'm not a color- or renderengine-expert, but i'll try hard to give correct answers, as far as my limited knowlegde goes on this topic. So feel free to correct me if i'm wrong.

What it comes down to is that lights themselves are not gamma-corrected in advance, the energy they emit stays the same, no matter what camera, film, exposure or whitebalance you use. Only the image, or the result on film, is gamma-corrected.

With or without LWF, if you use the Temperature-mode (8000K) for your IES-light, the outcome will/should be the same, as I've showed in earlier comparative tests. Depending on your Max-gamma settings the colorswatch representing the kelvin-colors will be lighter or darker. But if you, like you did in this case, use color-mode instead of Kelvin, and gamma-correct the representation of the kelvin-temperature, ofcourse the result will be different, in this case more blue. If you want a light to emit 8000K, choose Kelvin-mode.

If you think the first image looks too white, you can change the whitebalance of your camera to compensate for that.

1. Is it correct to say that the scale above features a subjective measure of RGB color relating to a temperature reading in kelvins?

It'll be the color representation of the chosen colortemperature, which is gamma-corrected according to your Max-gamma/LUT settings. So it's relative.

2. Are these readings objective and absolute when it comes to human perception of warm and cold colors? I see 400k as red, thats warm to me. I see 6000K as almost white, that seems neutral to me, while i see 8000K as blue which is cold, at least to me.

Absolute/objective and human perception shouldn't be in the same sentence . Only the difference between them can be called more or less absolute. As was mentioned in an earlier post, you can adjust whitebalance in such a way that colder colors look warmer or vice versa. But you'll definitely notice the difference between two lights in the same environment.

3. Is the scale above relative to something? 3200K reads as RGB above, i understand that this can be different, but wouldn't it be relativeto something else then, like a lense filter, white balance, ect.?

There'll be an algorithm to determine the representative colorswatch. Not familiar with those, so can't answer.

4. If it's so relative and cannot possibly be measured then why is there a color patch corresponding with a temperature change in the vrayIES dialog? It has to be relative... to something right? For example, why isn't 8000k in the scale above showing as red, along with an explanation "well, you can actually shoot 8000k with such a white balance and lense filter that it can actually come out as red".

To easily change a tint? Or if you don't want to use color-temperature? If you choose 8000K but you want it a bit warmer, you can change to color-mode, and make a slight change.

5. Why is 8000K represented by R150 G205 B255 if it's actually a light spectrum value?

To be able to choose RGB-values instead of just temperatures

6. Why does 8000K become R200 G231 B255 when i start using LWF? This changes the appearance of an image without using different camera settings - the only given explanation for this appearance change in real life.

The visual representation of the colorswatches respond to your Max Gamma/LUT-settings. If you use Temperaturemode, the colorswatch may look different with or without Gamma/LUT enabled, but the energy emitted by the light won't change.

Hope this shed some the light on the issue, in the right colortemperature

Best regards,
Ruud

[pailhead]

When you say energy... are you referring to lumens?

The thing is, i may change my position in this discussion. Now that i've seen how standard photometric lights represent the kelvin scale, i can definitely agree that it shouldn't be so saturated.

I think the conclusion is that it's actually the people who are using the LWF method that shouldn't have a problem with incorrect color display.

But now i wonder how do you compensate for a really intense and saturated sky, and IES lights when you have no idea what LWF is.

Someone is definitely at a loss here.

[victor.nsy]

Everything is relative ... color, form, space and the viewer. The more objects you need
to consider the more complex the interaction... simplify, simplify creates rooms for
creativity.

In observation, artificial light source don't necessary emit one color(not temperature in this case).
It could be due to the reflector angle/color? I usually use a radial map to fake colors in my light as I
realise in real environment sometimes lights cast on the wall do have much difference
in color than the one with temperature settings. sometimes I would push the value by using
vraycolor and kick the multiplier.

No matter how an environment is lighted up, you would have to adjust the exposure, shutter, filter
to take a photograph, and make it look good. So the same with a 3D scene. In color we refer to
hue, value and saturation. Any of the HVS changes the color is considered different. My point is
in image production is really the artist that decide what looks right.

My opinon is that as long as you know where are the position of certain light with certain temperature
you should be able to produce something close to the real world. If all your lights are of the same
temperature then of cos it would be meaningless cos there is nothing to compare to ... . And often
even if your selected lights are of the same temperature ... introducing some variety of color makes
a scene looks better.

I believe that the colors on vrayIES are representation of the temperature ... of cos it's not necessary to be
so saturated. And of cos it will depend on the objects bounced color in your scene. It a hue representation
of the final output. A change in the light energy gives you a different tone of the color, due to the change
in intensity the color would be brighter and more saturated compare to the lower intensity light source.

As we can observe there is so many factor to a correctly lighted environment, there is no true solution that
fixed all images.

[pailhead]

Let me get this straight, if it were you victor, you'd drop the kelvin scale altogether, and leave it purely to the artist?

[victor.nsy]

understand the scale but also know that there are many factors in photography where color
don't necessary looks scientific. . It's like the disney artists they understand physics then
break it to create great animation.

[pailhead]

I think that you missed the first step, you started thinking in abstract terms without understanding the basics. Or maybe it's the otherway around and my understanding of basics is wrong.


It's just, i've set up a scene so it uses both vrayIES and standard max photometrics, they seem to have different ideas of the same Kelvin value.

[victor.nsy]

I guess you ought to use just one to the light types ... different packages
and developer have different interpretation most of the time. Just stick with
the one you are comfortable with... a difference would be unavoidable ... how
ever if you want scientific accuracy, you have to do lots of research!

[victor.nsy]

Found some time and tried the IES light ... seem to have problem with
temperature color too ... why is it so saturated? Hahah pailhead now
I know what you meant.

Either it's so wrong or I am so not used to the vray ones ... max seems
to have the temperature color I am used to, for now I will keep on using
max's.

And the light object web viewport display for the IES model is missing too?

[pailhead]

Guess that i was the one who actually had no trouble at all since i'm using LWF. Default people will have trouble with saturated lights...

[pailhead]

If i were to sum up our conclusions from this thread, and if i were to put it in layman terms... would it sound something like this?


"If you're using the 2.2 gamma thingie, the colors of your vrayIES lights and vraySKY will look good."

"If you're not using the 2.2 gamma thingie, your vray sky will seem kinda darky and oversaturated... y, while your renderings produced with vrayIES will be heavily tinted and saturated"

[JeffG2]

I don't think we should throw in the towel quite yet.

I also think something is wrong.

In my LWF tests the sky looks far too pale.