underexpose landscape images by 2 to 3 stops

[Guillermo Luijk]

In order to be everyone talking about the same thing, the term "underexpose" has to be clearly defined. Underexpose, i.e. expose less, with respect to what?.

In landscape photography, where usually the sky matters, shooting RAW means only one correct exposure: ETTR. Anything below that is underexpose and anything above is overexpose.

Regards

[Bart_van_der_Wolf]

In order to be everyone talking about the same thing, the term "underexpose" has to be clearly defined. Underexpose, i.e. expose less, with respect to what?.

Exactly, couldn't agree more.

In landscape photography, where usually the sky matters, shooting RAW means only one correct exposure: ETTR. Anything below that is underexpose and anything above is overexpose.

That's my take on it as well. One additional complication may be the highlight compression that applications like Lightroom / ACR tend to apply by default. Reducing the amount of the Highlight slider there will help a lot.

Cheers,
Bart

[Tim Lookingbill]

The 2 to 3 stops is pretty much meaningless because you have to eyeball how many stops is contained in each and every landscape at the same time determine whether each stop under or over in exposure will fit that unknown scene dynamic range into the dynamic range capabilities of the camera. Some cameras can capture 8 stops, some 14. How many of you can nail it every time just by looking at the camera's LCD? I can't with my DSLR and I've taken hundreds of landscapes shooting Raw.

I always just expose for cloud highlights which will make the Raw preview in ACR/LR quite dark looking. Is that under exposure?

And because from experience shooting a lot of these daylight scenes that the full spectrum aspect of such light tends to provide very clean, noise free tree foliage shadows capturing at the lowest ISO of my camera. But I'm using an old DSLR with 8 year old sensor technology having roughly 8 stops of capture capability.

I have no idea how dark or light a stop influences the results because I have no idea what contrast ratio it will render the scene with. A stop doesn't give a crap about contrast which is what we all have to edit back into or take away from our Raw images.

[Hans Kruse]

Exactly, couldn't agree more.

That's my take on it as well. One additional complication may be the highlight compression that applications like Lightroom / ACR tend to apply by default. Reducing the amount of the Highlight slider there will help a lot.

Cheers,
Bart

About highlight compression, I find that if two shots are exposed one stop in difference and if I decrease the most exposed in Lightroom by one stop and compare to the lesser exposed the histogram is very closely the same and the look of the shot is very close. What I see is that if you increase exposure for a given shot then highlight compression really goes into action without clipping until a certain point. I have done many comparisons of shots from brackets and edited one of them and then copied the edits to the other shots in the bracket sequence and normalized for the exposure difference by alt-shift-cmd-M (on the Mac) and then they look very very closely the same, except for noise and potential clipping in highlights. So therefore when I bracket, I choose the most exposed in Lightroom that does not have clipping with zero sliders except for the sun or other highlights that will always clip. I find very very few cases where this is not the optimal shot with the optimal exposure. So underexposure does not make any sense to me.

[Hans Kruse]

Any suggestions on why this might be a good idea? (Seems counter to the advice I've seen on LuLa and elsewhere.)

Makes no sense to me. I have not read why this person would suggest that, but I would say that the best exposure is the one that is the most exposed without clipping essential highlights. The final picture may very well in extreme cases need to be blended with another exposure some stops higher to avoid shadow noise depending on which camera is used.

[MarkL]

In order to be everyone talking about the same thing, the term "underexpose" has to be clearly defined. Underexpose, i.e. expose less, with respect to what?.

In landscape photography, where usually the sky matters, shooting RAW means only one correct exposure: ETTR. Anything below that is underexpose and anything above is overexpose.

Regards

Agreed. I think in the case of the text in question in means the meter reading for the scene framed as-shot in matrix or maybe centre weighted.

[luxborealis]

Any suggestions on why this might be a good idea? (Seems counter to the advice I've seen on LuLa and elsewhere.)

Making a blanket statement like this may be fine for point and shoot photographers, but it is totally irresponsible in this day of digital imagery with the ability to see a histogram of the file - especially from a supposedly accomplished photographer writing a book.

I would assume (and perhaps incorrectly, but I don't think so), that someone who is taking the time to read a book for photographic instruction is interested in more than broad "shotgun" generalizations like the one given. He may as well have said "Just point and shoot on P mode" considering how good matrix meters are nowadays and because anyone following his original advice must not be interested in anything beyond snapshots.

Sorry for the rant, but advice like this really hit a nerve!

[Guillermo Luijk]

About highlight compression, I find that if two shots are exposed one stop in difference and if I decrease the most exposed in Lightroom by one stop and compare

My version of ACR doesn't apply any noticeable highlight compresion when the exposure slider is set above 0.0 and highlight clipping begins:



If anyone is interested to find out how their exposure (or any) slider works just post the before (0.0) and after image(s) and I can calculate the curve.

Regards

[Jack Hogan]

My version of ACR doesn't apply any noticeable highlight compresion when the exposure slider is set above 0.0 and highlight clipping beggins:

Hi Guillermo,

I assume you are showing the ratio of the luminosity channel curves.  Perhaps the tones are 'compressed' by keeping the highlights fixed and  pushing up from the bottom (the bumps in the lower left hand quadrant of your graph)?

Jack

[Guillermo Luijk]

No, they are RGB curves. But since channel differentiation didn't add anything (the three RGB curves overlapped), I just plotted the B curve.

The bumps on the left were discussed with Eric Chan and it's not clear the reason for them. Could be some non-standard 2.2 gamma curve related issue. Anyway they only affect the deep shadows (think that those curves are plotted in a 2.2 gamma corrected domain).

[Bart_van_der_Wolf]

I assume you are showing the ratio of the luminosity channel curves.  Perhaps the tones are 'compressed' by keeping the highlights fixed and  pushing up from the bottom (the bumps in the lower left hand quadrant of your graph)?

Jack's correct, and it would show easily with a stepwedge exposure. Highlight differences (equal increment/step sizes in luminosity) are compressed, but that can be changed with the 'Highlights' slider control in ACR/LR. It's an intentional effect of the process 2012 conversion engine.

Cheers,
Bart

[fdisilvestro]

My version of ACR doesn't apply any noticeable highlight compresion when the exposure slider is set above 0.0 and highlight clipping begins:



If anyone is interested to find out how their exposure (or any) slider works just post the before (0.0) and after image(s) and I can calculate the curve.

Regards

That looks like Process version 2010.

[Isaac]

… but I would say that the best exposure is the one that is the most exposed without clipping essential highlights. The final picture may very well in extreme cases need to be blended with another exposure some stops higher to avoid shadow noise depending on which camera is used.

For a high-contrast mid-day landscape picture I've just been working on: I took a matrix metered reference exposure, then I set the exposure using the live view histogram and checked the camera jpeg histogram after a test shot and bracketed 0.7EV; and as-it-happens that was -2EV +-0.7EV from the reference exposure.

Now I'd say that -2EV plus -3EV, from the matrix metered exposure, would have provided what I needed in the raw file.

Maybe I'll try to keep track of how well that works out for the scenes I photograph in the future.

[Jack Hogan]

For a high-contrast mid-day landscape picture I've just been working on: I took a matrix metered reference exposure, then I set the exposure using the live view histogram and checked the camera jpeg histogram after a test shot and bracketed 0.7EV; and as-it-happens that was -2EV +-0.7EV from the reference exposure.

Now I'd say that -2EV plus -3EV, from the matrix metered exposure, would have provided what I needed in the raw file.

Hi Isaac,

Assuming you own a recent FF Nikon ('matrix') like the D750 or D810 you may want to try a single capture in Highlight-Weighted Metering mode with ADL set to Auto and see if you achieve results comparable to the ones obtained through your machinations

Jack

[Isaac]

Assuming you own a recent FF Nikon ('matrix') like the D750 or D810 …

No, a relatively old SLT-α35.

Although Sony cameras (including mine) have also used Apical's DR technology for several years.

I would use jpeg and would use camera DRO/ADL and HDR and AE; but I don't want even the minimum-level of camera noise-reduction / sharpening / contrast / WB -- so I'm pushed into using raw without DRO.

[Justinr]

No, a relatively old SLT-α35.

Although Sony cameras (including mine) have also used Apical's DR technology for several years.

I would use jpeg and would use camera DRO/ADL and HDR and AE; but I don't want even the minimum-level of camera noise-reduction / sharpening / contrast / WB -- so I'm pushed into using raw without DRO.

Sounds exciting, what do they look like?

[AlterEgo]

Although Sony cameras (including mine) have also used Apical's DR technology for several years.

several years as in since A7s (announced on Apr 6, 2014), no ?

[Isaac]

No, several years as-in November 8th, 2006.

[AlterEgo]

No, several years as-in November 8th, 2006.

I see... my fault, I missed that it was __apical__, not __aptina__  !

[dwswager]

There is no practical reason to generally underexpose by 2-3 stops for every landscape image.

Without more reasoning, it seems he matrix meters and then to ensure highlights do not clip, he underexposes.  But this does not account for any other tones in the image, nor DR range of the camera versus the DR range of the scene.

My philosophy is to bracket when I'm in doubt.  Not only is it more likely you will get a usable file, but sometimes compositing or HDR is worth doing even when you are not going for some oversaturated look.

Because of the extended DR of my D810 an how well the shadows lift, I have played with the highlight weighted metering mode as a way to "auto" meter while protecting the highlights, but I'm not totally happy with it in some situations, so I still bracket.

Any suggestions on why this might be a good idea? (Seems counter to the advice I've seen on LuLa and elsewhere.)