Improved "AutoExposure" compensation estimation

Currently the only automatic modes for exposure compensation are the “highlight” protect modes (which are hit-or miss for me) and the “center weighted average” mode. I’ve posted before about the highlight modes and was told those really are only for highlight protection (although it still doesn’t make sense to me.)

When I’m doing an event shoot in a difficult venue (bouncing flash) and I have a few hundred pictures to edit, I usually select “center weighted average” exposure adjustment to give me a starting point with which to make my own adjustments. However, this mode is hit or miss, and I think DXO can improve here. They should be able to implement a multi/matrix metering mode that can be used similar to what our cameras do in autoexposure. It still won’t be perfect, but it would be better than what we have. And a raw processing tool has the benefit over a camera of already having full image.

How about it? Can we have a smarter “autoexposure” setting? It would probably help get my event shoots edited faster.

Just a question to understand this subject :
If your camera matrix metering does not give good result, how photolab could do better ?
I certainly don’t understand how you shoot and process.

Have you tried “DxO Smart Lighting”? It can work wonders - or mess up the image.
Bildschirmfoto 2022-03-20 um

1 Like

That’s very good (implied) advice, Mike.

Set the “Spot Lighting” option ON - and activate its “Tool”, which will allow you to draw rectangles on your image (do so on areas of high-contrast - - and keep the rectangle size small’ish, to start with) … then move the rectangle around until you find an appropriate starting point for further refinement.

HtH - John M

Because when using speed lights there are many many variables. The camera can’t possibly know how much light is going to fall on the subject when bouncing a flash until after the exposure is done. And TTL is hit or miss for a variety of reasons.

This is exactly what I’m trying to avoid. I’m looking for more “automatic” exposure adjustment without me having to draw rectangles on 400 images.

I am currently using some level of smart lighting with face detection combined with center weighted but I still have some faces completely blown out and overexposed by a stop or more even when the face is detected and in the center of the frame.

It simply doesn’t work like this.

Like @John-M I feel you may need to look at what you are setting your camera to and I think you might be asking for the impossible (or at least difficult) from PL, at least without some degree of AI, which is going to slow things down somewhat.

I shoot Jazz concerts and use the highlight weighted exposure mode on my Nikon D850 and still have to make Spot mode Smart Lighting adjustments, simply because there isn’t an app out there that I would trust to get things right according to my taste, rather than someone else’s idea.

Could you make one of your “awkward” images available here?

Come on Joanna… I’m expecting a reasonable metering for an image that has already been captured.
How can that be “impossible” or “difficult”. It’s certainly not more difficult than what our cameras can do.

The difference here, and perhaps you missed it, is that I’m using a speedlight… and as I said TTL is hit or miss. So I’m shooting in Manual. As I think the camera deciding the correct exposure while bouncing a flash is immensely more difficult than DXO figuring out how to adjust exposure of an image that’s already been captured.

Except PL isn’t “metering”, it is simply assessing the brightest thing in an already demosaïced image and trying to bring that within range. Do you really think that it is the job of post-processing to know how to automatically correct the mistakes we make when shooting? I have shot weddings (only a couple, that was enough) using flash with an angled reflector attached and that was using a MF film camera and no post-processing.

Nowadays, I tend to use something like 10,000 ISO and no flash, both at social events and concerts. My D850 doesn’t even have a flash built in and I have two Metz 54 MZ 4i units languishing in a cupboard for the last couple of years.

Call me old school, but I am a firm believer in getting it right in the camera rather than relying on post-processing as a rescue service.

Yes. I am asking the software to correct mistakes. The reality is my camera is iso invariant over ISO640. So I don’t worry too much about underexposed images. I just adjust in post. Rather than adjusting settings and making the partygoers wait a few extra seconds I just take the shot.

I do try to get it right in camera when I’m not shooting flash but the reality is in a dynamic situation like a dance party and lots of things changing (not to mention varying ceiling height) I shoot in real time and adjust in post.

Im simply requesting another autoexposure setting in Photolab which tries to use a little more intelligence to select an offset. You don’t need to use it. And you don’t need to vote for it.

Anyway I guess our strategies are different.

Yes, there’s certainly a lot more work involved in using Smart Lighting (with it’s associated Tool) - but I guess that supports the old adage; “No pain, little gain”.

Do they start out that way or is that the result of “Centre Weighted Average” being applied, Mike ?
If that’s how it is out-of-camera than there’s not much PL can do to recover lost detail.

I suggest you put out a new post requesting suggestions from fellow wedding photographers (esp. using speedlights) - - as they’re likely to have devised strategies to streamline exposure adjustments.

Regards, John M

Just finished the gallery. I would say in general “center weighted average” was overexposing well more than underexposing. It’s possible it was because the background not lit by the flash was darker than normal. I’m not suggesting center weighted average isn’t working properly. I’m just saying it has the limitation of any spot weighting mode

Anyway some of the images were overexposed originally and some were underexposed. I have enough experience with my gear to know how much is recoverable and in all cases the overexposure wasn’t to the point of clipping the raws. (My gear has about 1.5-2.0 stops of additional highlight detail available in the RAWs.) So all were recoverable and the final gallery looks great. (And my customer will be happy.)

Again I was just trying to suggest a little more intelligence could be built into PL to better calculate a better starting point for further adjustments.

Anyway perhaps my next shoot I’ll post some examples.

I’m with you, Mike. But I doubt DXO will implement an “auto” correction for exposure/levels as e.g. LR & C1 do.

I believe they would be able to do it without resorting to ‘AI’ as has been suggested. LR has been doing this for some time, I guess without ‘AI’ (i.e. ML) inputs in the first instance. There is, after all, a very long history of both analog and digital signal normalization and homologation (to various desired signal ‘models’) that I’m sure they can call on to reshape the luminance distribution. It’s simply calculus.

But “Smart Lighting” stands in the way of such a tool, I think.

I would really like DXO to deconstruct “smart lighting” into its component parts of luminance re-distribution and local-contrast changes. If they did so, they could presumably employ some signal processing to suggest a ‘better’ distribution of the luminence ‘signal’ in the RAW file i.e. “Auto” lighting and leave the user to fix the local contrast for themselves. However, the PL interface might then look similar to the LR ‘Basic’ panel with exposure, contrast, clarity, vibrance, blacks, whites etc.

So I don’t think they will do this. They would probably tell you that “smart lighting” is the auto-lighting you ask for. Unfortunately, it’s that and more.

‘Smart lighting’ using their canned algorithms is no doubt a selling point to people who want one-click solutions (although as both Joanna and Mike-M note, it is best used in a different, more experimental way). It’s a box-of-tricks that DXO probably believes gives it a marketing edge on its competitors.

[ I don’t agree. I think DXO’s unsurpassed demosaicing (and raw noise reduction that is closely linked to it) is their real asset. I would sacrifice ‘Smart Lighting’ on the altar of more flexible-but-expert user controls. But I don’t think DXO will be moved by my opinion :slight_smile: ]



Smart lighting sometimes does a decent job of leveling a scene. But it doesn’t adjust overall exposure. In general it is useful. I use it but mildly, I don’t generally overdo it. Usually I start at 25 and rarely go higher.

Again to be clear I’m rarely processing less than 150 images per shoot, and I’m typically trying to turn the galleries in 48 hours. So this kind of “auto” setting really would be helpful.

Again typically I’m shooting without flash and getting the overall exposure right in camera. But with strobes, I often “shoot now and fix it later”. And I’m trying to speed up the fix it part.

So your demand is more flash related, not exposure.
Maybe an example would help to understand the problem.


Ahah! I think I might have found your problem - are you relying on the ceiling to bounce and soften your flash? If so, this is indeed going to cause you problems.

I use a reflector/diffuser (about 5" across) attached to the up-pointed the flash head at 45°, which means I always get the same amount of diffusion and more reliable exposure.

This shot was taken at 1000 ISO, 1/250 sec @ f/11 with the Metz 54 flash and such a reflector…

Notice the beautifully diffuse lighting and lack of harsh shadows.

This request really isn’t about shooting style. As you know with bounced or direct flash with or without a diffuser inverse square law means subject distance changes also mean the amount of light hitting your subject from the flash changes greatly as distance changes.

Again. I have no interest in getting every shot just right in camera when using flash.

This topic is about lack of any kind of effective auto exposure setting in Photolab. The highlight modes only really work with overexposed images and center weighted average isn’t effective enough. I’m asking for a multi metering type option that is a little more intelligent than center weighted.

Not exactly.
By bouncing you enlarge the distance the light has to go. This means that the amount of light hitting your ‘subject(s)’ is more equal. The longer the light distance, the more equal lights on your subject(s), also in the depth.
You might have a look at Bounce flash examples - Tangents. On the right side you will find a series of articles.


I know that. Joanna was suggesting using a diffuser/reflector which suggests some of the light is direct vs all bounced.

Anyway as I mentioned the venue had variable ceiling heights so even bouncing wasn’t giving consistent results. And again my goal was to prioritize capturing the moments and not getting the exposure on the money.

If you weren’t using a reflector that’s exactly what you would get.

Here’s my camera and flash setup

For landscape shots…

For portrait shots…

The result is that all the flash is reflected, as you can see from my example shot.

Edit - oops! I turned the reflector the wrong way for the portrait setup, but you can see the principal.