PL Smart Lighting has lost algorithm to process (purposefully) under-exposed images from ISO-Invariant sensors

So this whole kerfuffle was because you wanted to inject a bit of sarcasm into the discussion??:wink::grinning: I’m still coming to terms with Smart Lighting and the new HSL Color Wheel to me there is quite a lot to learn in my retirement!


Hehe Larry; you could say so.
It’s kind of a paradox. John advised me a couple of times to use Smart Lighting but now he says Smart Lighting doesn’t work good (at least not as good as it worked in Optics Pro).
How would say native english speakers? Is ‘‘contradiction’’ the right word?

By the way; HSL tool now works like a charm. It was about time. Can they do something similar to Split toning tool?

The topic started nearly 2 years ago, and sins then we did a lot of research how certain functions worked or should work. One of those is the tool Smart lighting.
(detailes about this journey is spread around here on the topics.)
Me in person have never used DxO pre PhotoLab so no experience about SL in optic pro. So on topic i have no clue if it’s true or not.

I can relate to things that other applications are better or worse.
And music mixers are sometimes fully automated with connected knobs and sliders and sometimes you need to know yourself which knobs and sliders needs to be delt with together.
Which is better? Don’t know. One is easier/faster the other is more personal style and skils.
Me personal if something is automated i need to understand what it does for me and does it help me to save time. I want to be in control of the aidfunctionality.

For me SL isn’t that “smart” that i go higher then 50%, and aldoh it’s most of the time in my preset set as active i use it mostly as positioning the Dynamic Range, tonecurve so to speak.
Yes as the gammaslider in the tonecurvetool.
Exposure correction i use global as a floorlift where the SL is function as a flexible “bumber”
The distortion that the bumb is causing i try to correct with selective tools.

This methode i developed by extensive trail and error sins i started with DxOPL.
I used to compare me fiddling with other developerapplications or let others participate to see if it’s tool limitation or skill limitation. That’s one of the powers of this forum, from different views and skill levels are comming advises or thoughts.
My methode isn’t the best, it’s for me the best.

About the fact that SL is effecting a lot more then just the outer parts of te tiff.
In my mind SL is holding the both ends of the tonecurve and manipulate from there the "snake of tonality"curve in it’s best way to preserve it’s overal contrast balance and tonality.
That’s why it’s a enigma sometimes for me which luminance levels are effected and which not.
By using the boxes you can gain some control in that function.
So that’s why most people pownding on use the boxes.

Hi Peter;

It’s impossible to say in general which application is better. Each RAW converter is a mixed bag – it has good and not so good tools.

For my high ISO photos of dance competitions made in poorly lit gyms DxO is the best you can get.
For photos with high dynamic range (sun/shade) DxO is not the best option.

For example; I personally can’t rescue as much highlights with DxO as I can with other RAW converters. We are not talking about automation. We are talking about effectiveness of tools.

Remember my cat photo? How much work you had to do just to rescue some highlights and raise some shadows? This is not freedom to be in control. It more looks to me as limitation of tools so you had to use Control points and other tools because Smart Lighting and Selective tone sliders could not do the job.

I’m sorry but that is simply not true.

With RAW files, it is absolutely imperative that you do not exceed the dynamic range of your sensor in the highlights. Just like with film, do you know exactly by how much you can over-expose before the sensor no longer records highlight detail? With my camera (Nikon D810) it is no more than 2 stops, but preferably 1⅔ stops to guarantee detail; otherwise the detail is irretrievably lost.

  1. Have you actually measured how much you can over-expose a highlight without it blowing?
  2. Do you use manual exposure and spot-measure high dynamic range shots?

If not, then you cannot reasonably expect any post-processing software to “recover” non-existent detail in the highlights.

Here is an image, taken contra-jour which, apart from the sun flare, is about as extreme as you can get in terms of HDR in a single shot and processed entirely in DxO :

Before processing

Showing Smart Lighting zones for spot measure at medium (50)

Final image with other tonal changes including shadow and mid-range enhancement in the contrast palette


Hi Joanna;

‘’ If not, then you cannot reasonably expect any post-processing software to “recover” non-existent detail in the highlights. ‘’
Yes; I know that but that was not my point.
I don’t know if you understand me because of my limited English. I will try the best I can to explain what was my point.
Lets say I have a photo with high dynamic range (sun/shade). I put this photo in DxO Photolab and in Lightroom. I can recover (or recreate) highlights in Lightroom but I can’t recover the same amount of highlights in DxO.

I learned to deal with DxO’s quirks, don’t have LR7 to compare nor the experience with it to compare the level of ease of use. So i take it for as you offer it.
Maybe i have less “problems” in highdynamic scene’s because i am used to a maximum result that DxO is offering me. And i have my panasonic G80 set on idyn auto.
This lowers automatic exposure in 1/3steps to -1ev on high dynamic scene’s to create a underexposed image in general. ( in oocjpeg it also alters the tonecurve by suppressing highlight and lifting shadow’s but in my rawfile you only have the ev as effective correction.

This works like a charme for me. It’s a sort off auto ETTR mechanism.

In general DxO seems to deal better with underexposure then near to blown highlight levels.
But i needed in SP7pro also a fair time practise before i could recover highlight to my liking.
And now DxO is producing for me better images then SP7pro can.

So i am positive minded about DxOPL and provide the staff with my remarks and idea’s for improvements which can help to give us a better overall quality.

It’s a moving experience and in my case it’s moving upwards to my expertations of my ideal developertool. (I am aware that using one type of software you get a form of “stockholmsyndrome” and grow custome to it’s quirks to be blinded for things and on the other hand grass is always greener on other peoples lane’s.)


Hi Peter;

I agree. If software has limited capabilities you need to adapt how you shoot your photos. I think Joanna tried to explain this to me but I didn’t understand – to underexpose photos a stop or a stop and a half because DxO can’t handle highlights well. I think this was her intention.

As you say I also discovered that DxO is not the best software out there for highlights recovery.
But as John showed us with his chandelier photo DxO is also lacking in some other areas.

On the other hand for high ISO images made in not demanding light DxO is king. It’s beyond me how Adobe can’t make something similar as Prime noise.

I do NOT agree. Here’s why: When we look at a multistep process, each step will introduce errors, no matter if you want it or not. Sacrificing quality in the first step (taking the picture) will make itself noticeable in your end result unless you put in a lot of energy to fake away the reduced quality by replacing one error by another (e.g. noise vs. sharpness).

One other thing: DPL does tend do mark highlights as blown. Try it out and use the “neutral colors, realistic tonality, gamma 2.2” selection from the color rendition tool, what you’ll get is an image that looks fairly flat and has no blown highlights, unless they were really killed with the exposure. Starting with this plain looking image, you can decide what you are willing to sacrifice in order to get the result you’re after.

I find that DPL’s standard presets tend to display blown highlights with a margin of up to one EV compared to what is actually in the raw file. Using the color rendition as mentioned above brings back some of this headroom.

1 Like

That’s not correct.

DxO can handle highlights perfectly well - as long as the highlights are not over-exposed in the first place.

Digital camera sensors are very unbalanced in how they distribute the dynamic range available.

Most sensors will only cope with 2 stops maximum of over-exposure - some less than that.

Take the example of using something like Fuji Velvia 100 transparency film…

The maximum dynamic range of Fuji Velvia 100 is around 5 stops and you do not have the luxury of digital post-processing to change that. What you have to do is to spot-meter the highlights and shadows to determine how many stops of range you have. Then you use graduated filters to bring down the highlight areas to within the 5 stops available. This doesn’t work too well unless it is something like a landscape where you have straight lines separating out the highlights from the shadows.

This technique of spot-metering requires that you use manual exposure mode - don’t try to do this in any of the (semi) automatic modes.

You can do the same thing with digital but you need to establish what is the dynamic range of your camera before doing anything else. Instead of referring to a film manufacturer’s documentation for dynamic range, you can use DXOMARK’s sensor database to find the total dynamic range of your camera’s sensor.

A camera’s spot meter will always try to return an exposure that will make the spot appear 18% grey. Thus, if you want to make a highlight area appear brighter than that, you need to adjust the reading to over-expose - and, as I said before, that usually means by anything up to 2 stops, depending on your sensor.

If you try to increase the exposure more than that, you will lose detail in the highlight area.

To repeat - do not try to use automatic metering modes apart from something like Nikon’s Highlight-weighted mode, which can help in situations like stage lighting with spotlights.

No matter what your experience may be of “highlight recovery” in other software, if there is no detail recorded because of over-exposure, the most any software can do is reduce the level of the area to be slightly grey, which may give the illusion of “recovery”, when in fact there is very little chance of any true detail.

Even if software has unlimited capabilities, you still need to adapt how you shoot your photos, not to suit the software but to suit the sensor of your camera. It’s called getting the exposure right rather than hoping some magic tool will make up for your laziness in the first place.

Believe me, when you’ve wasted a few 5" x 4" sheets of transparency film in a large format camera, you soon learn how to get it right in the camera. Don’t forget, with film, you don’t have the luxury of looking at the taken image on the back of the camera like you have with digital. If it wasn’t right when you took it, it will cost you around £10 for the sheet of film plus processing - then the cost of a return trip to the same location to do it all over again.


{Big sigh} - - Yes, it can have those meanings - but I meant it in the sense of being witty.
" to say something in answer to a remark, typically in a sharp, angry, or witty manner".

I assumed, by assuring you (genuinely) that I bore no hard feelings, that you would understand that.

John M

Did you talked about idyn?
This is a kind of dynamic range mechanism.
It’s originally for ooc jpegs but it’s side efect is -1/3, -2/3,-3/3EV correction on the metering value.
This way it’s prevent blown highlight at exposure level. Til -1 stop.
It only works when needed so normal scenery wil not be effected by this setting.:slight_smile:

Exactement !

Thanks all for your answers.

I appreciate your explanation. Not that I don’t know all that about exposure but it’s nice to refresh those things from time to time.
To say it simple;
at the moment I write this I have THE SAME RAW FILE opened in DxO Photolab 3 and in Lightroom. Lightroom can bring back all those details in highlights and DxO can not. Conclusion: DxO has limited capabilities.

No; no hard feelings. It was just a misunderstanding.
I’m glad someone else finds Smart Lighting not as capable as they say it is.

Different tools for different jobs!

I find that SL creates results that I like with some of my images and that I don’t like what I get with all other images. I prefer to work with the tone curve, it is not smarter than I am :wink:

i think i can learn from you about the use of tonecurve tools.
i only use it when i am stuck and my more normal toolset doesn’t cut it.
the basics i understand but the fine workings as gamma’s and RGB decentred tuning goes more out the corner then trough… :thinking::sweat_smile:

if you have some time to spent freely i would appreciate a “tutorial” in DxO tonecurvetool handling.
i meant by tutorial some examples how you handle things like that “clearview” trick (was that) you? showed to adjust blacklevel for getting water clear.

Are you using it from step one or at the end of the image development?

From the manual

“First, generally speaking, DxO Smart Lighting changes bright images only slightly, but has a stronger effect on darker images. It has little effect on highlights, unlike Exposure Compensation. Second, you should stick with the three automatic correction modes as much aspossible, as they can cope with most situations, and then fine-tune with the Intensity slider afterwards. If you need to do further corrections,use the Selective tone palette or the Tone Curve.”
Bolded is mine.

1 Like

I recognize that recovering blown highlights isn’t as easy with PhotoLab as it is with the single adjustment in Lightroom. Nevertheless, I’m skeptical that PhotoLab is incapable of recovering highlights just as well. I’ve seen that claim refuted to my satisfaction and suggest ruling out the color rendering or profile used in PhotoLab as the cause. It might also be that some RAWs, but not all, really do render better in Lightroom than in PhotoLab. This might be because of sensor characteristics and working color space (Lr uses ProPhoto, PL uses Adobe - how ironic :smile:) and possibly also algorithms that interpolate or extrapolate details rather than recover details that have truly been lost in the capture of an image. Lots of possibilities which vary from situation to situation.


From what I can tell, idyn, etc, is a Lumix thing for jpegs and doesn’t touch RAW. Since I don’t have a Lumix and don’t use jpeg, I’m only really interested in getting the best out of RAW images.

From what I can tell, DxO does a much better job of handling RAW, simply because RAW files contain a whole load more data.

I’m sorry to say this but I think you’ll find it’s more that you haven’t yet mastered DxO’s capabilities.

Why not post that file here so that we can play with it?

1 Like

That’s why i use SL in boxes mode to define the bright parts which i want to “pin” down before i push exposure up.or a shadow place before i lower exposure.
I have some examples here on dxo feedback but i can’t find t easy other then in the EA section.
I am pritty sure that they are on this side also but where i don’t know.
found one old one to use SL box as “pin” on a bright spot so you can lift shadows without raising the highlights.

1 Like