Dehaze Slider in Lightroom CC and ACR CC

[davidedric]

This will be interesting.

There would seem to be two ways this might work.

You could produce the dehaze layer and selectively mask it. Or, produce the mask and then dehaze within it. The two products might even operate in different ways.

Dave

[Steaphany]

I realize this thread has focused on post processing to compensate for haze, but since this started out with

Scenes of distant mountains with no haze help make landscape images by Ansel Adams and other master photographers of his time and before him so attractive.

I may have missed it, but has anyone considered what ancient tools Ansel Adams may have had available to him and if such ancient tools may be possible today ?

additionally, in regard to:

That's partly because we experience so much more haze throughout the world now than the earlier photographers experienced.

Actually, Haze is a function of the molecular composition of the atmosphere and is a result of Rayleigh scattering. Regardless of what modern additions may be in the atmosphere, aldehydes, nitrogen oxides, peroxyacyl nitrates, tropospheric ozone, and various volatile organic compounds, all of which contribute to Photochemical Smog, Rayleigh scattering and Haze are natural and the atmosphere hasn't changed that much in the past 200 years.

Rayleigh scattering is what gives the world a blue sky, in fact, the spectral composition of the blue sky is Haze. If your goal is to reduce haze, then you are also wanting to darked the very sky itself.

What causes the loss in contrast and blue cast of distant mountains and scenery is the air between the distant scene elements and your camera is being illuminated through Rayleigh scattering. The greater the distance to a scene element, the more air the light has to pass through and the greater the effect of Rayleigh scattered haze.

Here are some ancient tools like Ansel Adams used which prevent haze in the first place. Filters, specifically polarizing filters and then colored filters such as the

• K2, #8 Yellow
• G, #15 Orange
• #25 or #25A Red

The reason a polarizing filter works in haze reduction is that the light from Rayleigh scattering is polarized. Just tune the filter to only pass light oriented perpendicular to the plane of the Rayleigh scattered light. ** Poof ** No Haze and no need to fix in post.

The colored filters are more oriented to B&W photography, but I have all three and had wonderful success with a K2 for full color scenes. The K2 will reduce the effects of some haze, but will still pass sufficient short wavelength spectra to allow a white balance reference to completely compensate for the yellow cast.

I have found the G to leave many blue scene elements strongly darkened, almost to black, but it is much stronger than the K2 in reducing haze.

The #25 or #25A are so strong that they are best for B&W photography, but they will yield the greatest reduction in haze.

I haven't listed them here, but if you shoot through Infrared filters, they will cut the haze, to the point of even seeing through smoke,

I also want to mention some trade-offs between a polarizing filter and using colored filters, especially if you are shooting through a wide angle lens for landscapes. Colored filters have the advantage here as they will uniformly effect the whole scene. A polarizing filter on a wide angle lens can cause a weird un-natural regional darkening of the sky. The polarizing effect can cut haze perpendicular to the filter nearly 100%, a totally black sky. But across the field of view, the orientation to the Rayleigh scattered haze will cause the sky to become tighter as the difference decreases.

Here is a frame from a time lapse I shot where the effect of the polarizing filter is obvious:

See, No Haze, but doesn't look right either

[Mike Buckley]

Fascinating information, Steaphany! Even so, I'll continue to use the term, haze, to describe the overall appearance of particles obscuring the detail of distant scenery. Even though I now understand it may not be a scientifically accurate use of the term, my use is consistent with non-scientific definitions found in standard dictionaries intended for use by laypeople.

[davidedric]

Thanks, Steaphany, that is very interesting (I am a self confessed geek and lapsed scientist).

The dehaze filter also copes with dusty air, do you have any ideas as to the mechanism?

Dave

[Steaphany]

It's OK Mike, what else can you expect from a gal who introduces her 500px profile with:

Photography = Applied Quantum Electrodynamics
Lets go play with some Photons

[Mike Buckley]

Way too funny, Steaphany!

[Steaphany]

The dehaze filter also copes with dusty air, do you have any ideas as to the mechanism?

Dave,

I am not familiar with the specifics of the algorithm Adobe employees, but I know much of the research in this field is published in the journals of Association for Computing Machinery's Special Interest Group on Computer GRAPHics, a quick search yielded a paper comparing several research groups efforts and, if you are formula inclined, some of the mathematics:

Investigating Haze-relevant Features in A Learning Framework for Image Dehazing by Ketan Tang of Hong Kong Univ. Science and Technology, Jianchao Yang and Jue Wang, the last two authors both with Adobe Research. These people are on the team that comes up with the magic that is later coded into Adobe software products. A nice feature of the paper is the authors email addresses are available if anyone wants to consult with them about their work.

[davidedric]

Now I am way out of my depth. A little while ago I went back to my PhD thesis. I know we used augmented Legendre polynomials to represent metallurgical properties, but I was completely baffled

Thanks for the links,

Dave

[Mike Buckley]

I just now learned by working on an image that the Dehaze function also helps minimize the effects of flare. Notice that I used the term, mininimize, rather than eliminate.

[Timar]

AFAIK Adobe's dehaze algorithm employs the dark channel prior method similarly to the one published by a research team working for Microsoft. Maybe they even licenced those algorithms from Microsoft.

Steaphanie's technical explanations are spon-on (that's why I noticed that the sky is literally the limit for the dehaze filter, remember?), but I disagree with her about the usefulness of colored filters in digital photography regarding haze removal. Contrary to polarizing or neutral densitiy filters, there is no purpose for colored filters in digital photography - at least when you shoot raw. You can easily remove the color cast from Reyleigh scattering in post processing, as you have probably done a thousand times before - simply by adjusting the color balance, without loosing light transmission and valuable information in the luminance channel as it is the case when using a filter. Moreover, this approch only makes sense when the whole image is about equally affected by haze, i.e. when taking a tele photo of a distant scene without foreground objects that are less affected by haze - in such a case, most modern cameras actually do a pretty good job in correcting the color balance on their own. Color filters make a lot of sense when shooting film, as the color balance is a property of the film emulsion and can only be corrected in development to a limited extend and by going into a lot of effort. They are indispensable when shooting monochromatic film, as they are the only way to influence the spectal response of the film. They make no sense at all in digital raw photography, be it for haze removal or any other purpose.

It is the very nature of haze, however, to affect objects proportionally to their distance from the observer. That is why He et al. were able to extract such an impressively accurate depth channel with their algorithms. This information is then used as a depth mask to remove haze proportionally to its calculated occurrance in the scene, primarily by shifting the color balance to the right, increasing contrast and saturation and lowering the black point. All of those adjustments can be done manually, but it really is the depth mask where the magic takes place when using the dehaze filters.

[Mike Buckley]

[Color filters] make no sense at all in digital raw photography, be it for haze removal or any other purpose.

Never say never. Try doing some photography involving one or more color filters over a strobe that lights only part of a scene. Try using several strobes, each with a different color filter attached, to light a scene. Then try to create the exact same look without using the color filters. I'm reasonably confident you'll be convinced that, indeed, there are many uses of color filters in digital photography, whether or not a raw file format is being used.

[Manfred M]

Never say never. Try doing some photography involving one or more color filters over a strobe that lights only part of a scene. Try using several strobes, each with a different color filter attached, to light a scene. Then try to create the exact same look without using the color filters. I'm reasonably confident you'll be convinced that, indeed, there are many uses of color filters in digital photography, whether or not a raw file format is being used.

Mike - while I "technically" agree with what you are saying, the language use in photography generally does not refer to the coloured filters placed over light sources as filters. Filters go in front of the lens and gels are what we place in front of light sources.

We gel lights to either correct the colour temperature of our light sources (CTO to emulate tungsten light, CTB to emulate daylight) or just plain coloured gels to give us funky colours, like we see in theatrical lighting. Both of these uses means we are introducing mixed colour temperature light sources into our shoot and are looking for these measures to enhance our images.

[Mike Buckley]

Lee, Roscoe and ProGel (the first three manufacturers I looked up) call them filters. That makes sense to me because gels are in fact filters. Example: "The Lee Color Effects Lighting Filter Pack contains a select assortment of 10 x 12" precut sheets of gel lighting filters."

[Manfred M]

Lee, Roscoe and ProGel (the first three manufacturers I looked up) call them filters. That makes sense to me because gels are in fact filters. Example: "The Lee Color Effects Lighting Filter Pack contains a select assortment of 10 x 12" precut sheets of gel lighting filters."

Again Mike, I agree that these are filters, as the do modify the light. In my 40+ years in photography, I have never been asked to put a filter over a light source. On the other hand, I have been asked to gel a light many times.

[Timar]

Well, obviously I was refering to lens filters. Lighting filters are of course an entirely different beast

[Mike Buckley]

I just now post-processed a photo of a waterfall that had lots of mist rising in the air from the water. Unfortunately, the result was decreased dynamic range and mid-tone contrast yet lacking the appealing mood often associated with mist. Not yet being used to automatically thinking of using Lightroom's Dehaze capability, I used my other software to deal with the situation as I have been doing for years. After spending a bit of time on it, I still hadn't achieved satisfactory results. That was when I suddenly remembered that I can now use not just the Dehaze capability but that I can also use it selectively (a very recent innovation in Lightroom). The change I had in mind was achieved in just a few seconds.

It has been years since I have personally experienced such a transformative innovation in post-processing software beyond the realm of cloning.