EXPOSURE SKIN TONE

[lucianosantos]

Hi Roger, I’m new to the forum, I’m Brazilian and I admire your work! Please excuse any errors in my English.

I have a question regarding exposure for fair skin outdoors during the day in strong sunlight. What is the maximum IRE (Index of Relative Exposure) that this skin tone can withstand without losing color information?

 

Thank you!

Luciano Santos

Topic

[Roger Deakins]

Well, I wish I could give you a number. It seems to me that you loose saturation immediately you overexpose a surface, whether it be skin or not. The effect of color loss might only become noticeable once you get above a certain level, perhaps a stop or two, but anything above key appears less saturated than at key. Just look at your hand being lit by the light from a window. Take your hand closer to the window and the highlights appear less and less saturated. I’m certain someone will have a more scientific answer.  Maybe the magic number!

[dmullenasc]

I don’t have a scientific answer but skin is a pastel color and as it gets lighter, like any color, it gets less saturated by eye.

A similar thing happens in colored lighting — 5600K light on a 3200K camera looks pale blue when overexposed and deep blue when underexposed.

As to how overexposed a face can get and still be corrected back to normal, that depends on the camera and the recording format, the skill of the colorist — and the face.

[lucianosantos]

Thank you, Roger and Dmullen, for your collaboration!
I asked this question thinking there might be some standardization in the industry; however, I see it as an aesthetic choice aligned with the narrative.

Please excuse any mistakes in my English.

Thank you!
Luciano Santos

[carloscfrias]

Hi! Maybe it’s a little late but I’d like to add that, when working on digital cameras and scene-referred color spaces (log, linear RAW…), we must differenciate between recording the color information and visualizing it.

With any profesional digital camera, you can retain the color information up until the end of its dynamic range (that’s why ISO changes the dynamic range, you can record all the information if you are careful not to clip anything and then use the ISO to set it on the level you want).

Then, as Roger described, perceptually the color gets less saturated as it gets closer to white, but this happens on the display stage of the pipeline. This will also depend on the LUT you are using to visualize the image: more contrast will result in the skin getting closer to white when overexposed a bit and different choices in color response may desaturate colors in certain regions for aesthetic reasons.

Because every log curve sets the luminance on different IRE levels and because it ultimately depends on the dynamic range of the sensor used, there’s no IRE reference level for this on the recorded state. Since every display transform will render the luminance on different IRE levels and will have different color response, there’s no IRE reference level for this on the display state.

Hope this helps!

[Author]

Replies