The color of the Kelvin temperatur with the photometric lights has changed with the versions Max 9 and Max 2008, and is still persistent without any legacy option. Something must have been changed with the internal calculation of the rgb values out of the Kelvin temperatur. I have made some test with the scanline and MR renderer. I ´ve let MR rendered .mi files and inspected the color values of the same Light, which are for the example Max 9: 1 | 0.882938 | 0.77311 | 1 and for the Max 2009: 1.41342 | 0.924217 | 0.532697 | 1. Can anybody tell what has been changed and how to convert one value to another. The image and Max file I provide are made by Sascha Selent who had discovered the bug before.

	Attachment

	Attachment
11528_ZxrFaAF9BCO96t9ldCEx.zip

Okay, I had some conversation with one of Autodesk´s Doc-Team. Here it is:

“That’s really a question for the mr team but I’ll give you my point of view.

I believe photometric lights were re-written at some point after max9. However, I feel they have improved but they do need to be used with a few pointers in mind.

First, Photometric lights are not meant to be used with Scanline renderer, I advise you to use them only with the mr renderer. They are meant to be physically accurate and the scanline renderer is anything but. I noticed that your file was set to Scanline when I opened it.

With that out of the way, you need to take a look at the Kelvin temperature scale. In fact, 4000K is closer to yellow/orange than it is to white, although 4000 is referred to as “Cool White” sometimes. Google “Kelvin Temperatures” and you’ll get quite a few hits with a chart like this one:”

	Attachment

"This said, your renders are far too orange but this is because you’re not using the mr renderer and the proper exposure. Here’s a render of your scene that I opened in max2009 and rendered using your light (didn’t touch its specs) and the mr renderer, FG and with exposure set to mr Photographic Exposure.”

	Attachment

"As you can see, it’s yellowish like the chart but not as orange as your render. (see below)

Also, the mr Photographic exposure has some additional controls to fine-tune the color output.”

	Attachment

"If you want a pure white color, you ought to be closer to a 6500 Kelvin value. That’ll get you closer to the max 9 render you got.

Hope this helps,”

Okay, these are the suggestions from the Autodesk-Doc-Team-member so far. Here is my reply:

“Okay, here some more questions:

“I advise you to use them only with the mr renderer”
I have to be very flexible with renderers - so far I usually used finalRender, but now I use VRay - both quite physically accurate. I never use MR - I do not get the results I desire… ;( - so using MR only is no solution for me…

“In fact, 4000K is closer to yellow/orange than it is to white”
Yes, that´s true - and it shall be. Like the renderings before Max2008 show, the light´s color is quiet exactly what the chart shows. But it´s not that crazy orange.

“Also, the mr Photographic exposure has some additional controls to fine-tune the color output.”
I also never use exposure, because it washes out the colors in a hard to control way.

“As you can see, it’s yellowish like the chart but not as orange as your render. ”
That´s true, but it forces me to use MR… And without tweaking with exposure the result is exactly the same as with scanline and every other renderer - different from prior releases.

“If you want a pure white color, you ought to be closer to a 6500 Kelvin value. ”
I´ll check this. In Max2008 I found a workaround for the problem: I grab the color of the Kelvin-temperature and put this in the Filter-Color for the light. Now I set the light to the “D65White"-preset - which is pure white. This way I can exactly reproduce the Max9 behavior. In Max2009 the “D65White"-preset is NOT pure white, it´s somewhat reddish - so I can never reproduce the look of pre-Max2008-behavior. I checked the 6500 Kelvin value. It´s not possible to make a pure white photometric light in Max2009. The difference is significant.

I cannot rely on MR, I have to use my plugin-renderers.

Since Max2008 I cannot rely on the Kelvin-values for photometric lights. They are wrong. They are heavy too saturated and far away from the chart you sent.

Since Max2009 I cannot reproduce the old behavior - therefore I cannot reopen an old project, make a change and rerender this scene without significant color-changes.

I HAVE to use photometric lights with other renderers than MR, especially because of the feature “WEB-distribution”.

I love improvements, but i think it is critical to do changes that doesn´t allow you to exactly reproduce results from prior releases.

Isn´t it possible to implement a legacy-option for the use of photometric lights, so that old-styles can be reproduced?”

And here the final reply of Doc-Team-member:

“Yes I can see your dilemma. Unfortunately, as I don’t use 3rd-party renderers, I’m afraid I won’t be able to help much more. I’m sure there are ways to use photometric lights with Final Render or VRay but although I know how good these plugins are, I’m not familiar with their workflow.

All your points are quite valid but they really need to be discussed with the development team. As I’m part of the Docs team, I’m afraid my input doesn’t carry much weight as to how the software is implemented.”

----------

I´ll stay in touch with the problem and I´ll try to contact some people of the development-team. If I get some more news I´ll post it.

Agan I´m suprised that me and “maxell” seem to be the only ones who recognize that problem. Does no-one use Kelvin-lights?

Sascha

First, I´m quite puzzled about the statement that the scanline renderer is not physically accurate. I´ve used it a lot with the radiosity solutions for lighting simulations, what it is explicitly meant for. This combination is the successor of the famous lightscape program, which is all about simulating physically correct lighting situations.
Also, the rendering with the scanline and the MR renderer (without exposure control) are looking identical. So imo it can not be a valid advice not to use the Photometric lights especially because they are the only ones with which .ies files can be used.

Another point that I came across while dealing with this problem is, that the pure white Kelvin color of 6500 are only visually correct if one has adjusted the monitor to the same. On Windows PC the sRGB color space is meant to be used. It relies on the assumption that one is viewing in an environment with the 65000K color temperature.

It seems to me that the Light colors have been incorrect until the release of Max 2008. What really is missing is a legacy option to convert older files to the new settings or vice versa.

maxell 27 September 2008 09:02 AM

First, I´m quite puzzled…

Yip, I agree.

maxell 27 September 2008 09:02 AM

Another point that I came across while dealing with this problem is, that the pure white Kelvin color of 6500 are only visually correct if one has adjusted the monitor to the same. On Windows PC the sRGB color space is meant to be used.

Does this has any effect on the final RGB-values in the rendering? I think it has not, doesn´t it? It has only effect on the way the colors are displayed on the screen, right? So watching the image on a different screen shows different colors? And what about the non-Kelvin-lights? Are they shifted in color this way?

Does this has any effect on the final RGB-values in the rendering? I think it has not, doesn´t it? It has only effect on the way the colors are displayed on the screen, right?

That´s right! It does not influence the color values but

So watching the image on a different screen shows different colors?

it should prevent colors from looking different on various screens. That is what colormanagement is all about, and calibrating the monitors to a common color space is part of it.
And if you have a monitor that is not calibrated to sRGB than you will pick wrong colors (e.g. the light filter color) because the color picker swatch is displayed in a distorted way. The presets of the “Kelvin lights” are using fixed RGB values which can not be altered by the user. So you have to adjust your monitor accordingly to get the visually correct renderings.

maxell 30 September 2008 07:12 AM

That is what colormanagement is all about, and calibrating the monitors to a common color space is part of it.
And if you have a monitor that is not calibrated to sRGB than you will pick wrong colors…

This is the reason why I do not use color-management in non-print-enviroment. So this way my colors are not distorted in any way. This is the reason why I can reproduce the exact look of Max9 in Max2008. If I rely on a calibrated screen I will get wrong results on a non-calibrated screen for sure - and most Users do not have calibrated screens. The reaction of most clients to the term color-management is “color-what?”.

maxell 30 September 2008 07:12 AM

And if you have a monitor that is not calibrated to sRGB than you will pick wrong colors…

Is that true? I would think that picking the color of a screen that is under influence of a color-profile would give me a changed - and therefore a wrong - color. I didn´t check this, but I think a color-profile for the screen may change the way colors are displayed on the monitor but not the rgb-values themselves. I think a non-calibrated color 128, 128, 128 may look different under influence of a different color-profile, but if you pick this color it is still 128, 128, 128. But I´ll check this.

Okay, let me know what you know about color-management - I hate this topic, it sucks! I suggest it is a invention of DTP-industry to sell hellish overpaid trainings an equipment. I do print as well and my experience is: if possible sh** on color-management. It´s no guarantee that the printed result is as it shall be - even if you do everything right. The only way to guarantee best-possible results is experience, a printing-shop you can trust in 100% and proofing, proofing, proofing. I do not want to know how many footage has been shifted in colors by accident because people do not really know what they do in color-management.

By the way: no contact to the dev-team yet, but I´ll stay in touch…