Question For a Roger Regarding Flare

[willie_901]

On your web site's article on backlighting you (or Francis) discussed factors influencing flare and wrote, "Over-exposure. You might expect exposure to march in step with density (of negatives) but it does not usually work that way: flare almost always builds faster than image density." This statement seems to apply to both ghosting flare and veiling flare.

Do you think this holds for digital media as well? The flare intensity would be proportional to exposure. But would the flare amplitude increase more than direct-light amplitude, especially with gratuitous, gross overexposure? I can't think of a reason why light inducing chemical reactions would be different then light creating electrical charge in semiconductor wells.

Clipped analog signals at the sensor when the shutter is open and clipped analog signals in the ADC when ISO is greater than base ISO cause problems. But flare (reflections) are no different than any other light source in these events. So I an not thinking about clipping.

Sensor bloom is efficiently suppressed in contemporary, sensor-semiconductor technology. I recently read over exposure artifacts in digital imaging is almost always mis-attributed to sensor bloom (excess charge propagating to nearby sensor wells) rather than lens independent flare effects associated with microlenses and sensor cover glass. If these effects are in fact responsible for artifacts in overexposed images, then they could be relevant to my question. They would also be highly dependent on camera design and hardware differences.

So my primary concern is veiling flare contributions from the lens itself along with reflections arising from the lens-sensor geometry.

Sorry to ask annoying questions, but my research on this topic is frustrated by conflicting, contradicting articles.

 

[Roger Hicks]

Dear Willie,

I'm not sure it's a meaningful question, except perhaps in philosophical terms. After all, we are talking about a negative (exposure keyed to the shadows) and a positive (exposure keyed to the highlights). Once the highlights are blown, they are blown: they cannot get any more blown. A negative can however always get denser, up to the shoulder of the characteristic curve.

Having said this, my suspicion -- and it's no more than a suspicion -- is that yes, (flat) digital sensors and (deep) film would be very different because of halation and irradiation in the film.

I could however be wrong. Sorry I can't be more help.

Cheers,

R.

 

[Sejanus.Aelianus]

I'm a jpeg rather than a raw user and I've noticed that newer chips write more highlight information than gets displayed on the default viewer apps. Using Apple's Preview, for instance, I can recover approximately 2 stops more highlight information, simply by using the highlight control on the colour tool, than is displayed by default. The older the chip, the less highlight detail I can recover.

I'm guessing this is not news to most users?

 

[willie_901]

Thanks for your insights.

Since my post I found an article by Erwin Puts. He concludes:

"Over-and underexposure also have a major impact on the results."

"The most realistic version of the series of photographs is the one with the correct exposure. The best photograph for visually grasping the effects of the flare is the one with one stop over exposure. I will present both pictures. The comparison images presented here show a severe amount of flare that is unlikely to be encountered in normal photographic conditions. "

So it seems overexposure makes the problem worse because the effects away from the overexposed region(s) affect on the shadow regions is at least more obvious when there is gratuitous overexposure.

 

[Roger Hicks]

Thanks for your insights.

Since my post I found an article by Erwin Puts. He concludes:

"Over-and underexposure also have a major impact on the results."

"The most realistic version of the series of photographs is the one with the correct exposure. The best photograph for visually grasping the effects of the flare is the one with one stop over exposure. I will present both pictures. The comparison images presented here show a severe amount of flare that is unlikely to be encountered in normal photographic conditions. "

So it seems overexposure makes the problem worse because the effects away from the overexposed region(s) affect on the shadow regions is at least more obvious when there is gratuitous overexposure.

Dea Willie,

Ah, yes: a different starting point. I'm sure he's right: I was considering the highlights too much. I fear one needs D/log E curves to illustrate it well.

Cheers,

R.