Lists

Terms and Conditions
Lists hosted on this site
Email the Postmaster
Tips for posting to public mailing lists

Re: Neutral grey under different lighting

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Neutral grey under different lighting

Subject: Re: Neutral grey under different lighting
From: tom lianza <email@hidden>
Date: Fri, 07 Sep 2007 08:46:32 -0400

Hi to all,

I've been reading all these terminology discussions and it becomes clear to me that some folk are spending a lot of time trying to understand color science and some of the terminology. As a researcher in this field, I thought that I would share some thoughts with you. First let me state that metamerism is highly over-rated. Metameric matches are not really common and I would argue that when discussing neutral appearance, they are probably not desirable. If an image media takes on undesirable color appearance under different types of illumination, it is an undesirable media. We don't need to argue over color science terms, it is simply unacceptable. Epson got that message very early on. Terms like "color constancy failure" or "color inconstancy" all reduce to the same term in customer speak: unacceptable. The fact that a particular neutral on one media doesn't match a gray card when viewed under arbitrary lighting doesn't and shouldn't, be a surprise if you understand what a gray card is and how it should be used.

When illuminated , a gray card is a moderate luminance, diffuse representation, of the source that illuminates it. The "neutral" aspect of a gray card comes from the fact that it has minimal impact upon the spectra that illuminates it. IT HAS NOTHING TO DO WITH IT APPEARING NEUTRAL. Naturally, if we shine red light onto the gray card, it no longer appears "gray", it looks red. As a matter of fact, if you take a nice walk around your neighborhood at night and look at all the various potential sources of illumination, (I wouldn't recommend looking through too many windows, you could get into trouble) they all seem to have a hue. As the relative illumination of the scene increases our sensation of absolute hue begins to be affected by the average level of light in that scene. This is DaVinci's first axiom of color: the color of each object in a scene is affected by the color of the light that illuminates the scene. A second, but far more interesting axiom is: Every object in a scene acts as a source for every other object in a scene and is affected by the color of the light that illuminates a scene. Axiom #2 is important to color scientists because it generally results in justification for their paycheck. As a result of that, and the basic intractability of finding a solution to that problem, we tend to avoid investigating it. Researchers like Edwin Land and his followers still look at the problem, generally for a fee, but Retinex theory is alive and well and it even works....sometimes.

Now let's take this gray-neutral card into the real world. There is a concept in color science called "adopted white point". The "adopted " white point is the color of the illuminate that would make a neutral gray card appear neutral to an observer who has "adapted" to a given scene. Read the previous sentence one more time. Note that "adopted" is NOT "adapted". It's important to note that if we are looking at a scene illuminated by a perfect D50 source, the "adopted" white point, may not be D50. How can this be? The scene contains lot's of sources. That's why the lawn appears greener than an individual blade of grass. Let's take that gray card to the beach at sunrise. As I stand with my back to the sun and look at the gray card, it does not appear neutral. Why? Because the surround is at relatively low luminance, and the illuminant has an obvious hue. Now, what is the "adopted" white point in this situation? That's not really easy. We would need to measure the gray card (spectrally), deduce the hypothetical source that would make the gray card yield a flat return spectra, calculate the CIE XYZ coordinates from this hypothetical source, and use those values in your favorite chromatic adaptation algorithm : Bradford, VonKries, wrong VonKries.....etc. That would yield an image that represented the non-neutral representation of the gray card in the scene. That is what you would need to do to build an accurate ICC profile for the camera imaging this scene. A very well known color scientist told me that chromatic adaptation technology is a good example of a "robust color science solution". Now there's an oxymoron. Alternatively, you could set the white balance in the camera to D50 (if you have such a setting) and just take the picture. That would yield a "colored" representation of the gray card which would be far more accurate than color balancing the camera to form a neutral appearing gray when reproduced.

When you balance your camera to a gray card, you "discount" the illuminant. When you compare a gray card to a printed media, you "markup" the illuminant. You may count it too much. As a photographer, you have to ask yourself if that is what you really want to do? In Cinematography, the source of illumination is often as important as the actor in the scene. Cinematographers spend a lot of time trying to capture the nature of the illuminant in the scene. The key is how the media looks when illuminated. Trained photographers and Cinema Colorists, can look at an arbitrary scene and judge color quite accurately without any gray reference. Yes they are in a controlled environment and that's far more important than a gray card.

Regards, Tom Lianza Director Cinema and Video Technology X-rite corporation. _______________________________________________ Do not post admin requests to the list. They will be ignored. Colorsync-users mailing list (email@hidden) Help/Unsubscribe/Update your Subscription: This email sent to email@hidden Prev by Date: Re: RIP for Mac Suggestions Next by Date: James R Bachman/US/DNY is out of the office. Previous by thread: Re: Neutral grey under different lighting Next by thread: Re: Neutral grey under different lighting Index(es): Date Thread