Want to share the following surprise observation. I don't think I saw this discussed before.

Shot a bracket last evening on a 5d mark 1, AV priority, 0/-2/+2 stops.

Develop: Begin with defaults, normalize white balance, adjust exposure 0/+2/-2 such that all images look the the reasonably close on screen. Same for the histograms.

Next I develop for the shadows but guess what: I need black values of 23/6/100 to get similarly looking images.

Doesn't this imply a couple things

1. The raw process chain applies the black slider before is applies exposure. I always intuitively assumed the opposite.

2. I want those extra levels in ETTR because the black slider is very coarse on the dark exposures shadows.

3. I don't want too much ETTR because I may run out of black slider runway on light exposure.

Really 2 and 3 are a consequence of 1 I think.

My own "right" answer to the above is to expose reasonably and not push extremes, but the apparent fact that the black slider behaves dramatically different on a histogram that looks similar was a surprise to me.

Thanks

Richard

P.S. I run Lightroom 2.3/CR 5.3 64 bit.

Hi Richard, good observation. The histograms are useful, but provide limited information. They should not be used as an indicator of how a given slider will behave in CR/LR. For example, two images -- one in focus, one slightly out of focus, but otherwise equivalent -- will appear to have similar histograms. When you examine the images themselves, of course, they will appear to be rather different, one being sharp, the other not. Consequently, the Detail tab's sharpening controls will behave differently on the two images. This is another example of how two images with similar histograms would respond differently to a given slider.

You are correct that at the lower levels, between 0 and say 10 or so, each single unit adjustments has far too much adjustment...once you get over 10 in the Blacks, you can go further with each increase effectually less impact than the same adjustments under 10 would have.

If you need to reduce the Blacks from it's default black, the odds are pretty good you've under exposed or are shooting a scene whose contrast range is far beyond the dynamic range of the sensor.

Also, in a bracket, normalizing a series of exposures implies you WILL need to hit the Blacks on the + exposure. That's not at all unusual...

A +1 exposure needs about double the blacks for similar shadows (and more fine-grain shadow adjustments for the +1 exposure).

Never thought of near zero blacks as a sign of poor exposure (because I always assumed the exposure slider would take care of my poor exposure ;-)) but now it makes a lot of sense.

Thanks Eric and Jeff.

Richard

Interesting.
Seems I can reproduce the effect.
The higher the real exposure, the higher the Blacks setting needed,
even though exposure was "normalized" via the Exposure slider (before setting the Blacks).

Peter

--

Remember, results in CR/LR depend only on your slider values, not the order in which you set them. For example, doesn't matter whether you set Exposure to +1 first or Blacks to 8 first. Regardless of what order you set them in, you get the same results. Same is true for all of the other settings.

Eric, - yes.
However, that is most likely not the point.

Here’s another example:
Scene of moderate DR, two shots with 0 and +2/3 EV
Exposure setting in CR to see some first "white clipping": +1.40 and +0.70
So far, the numbers are logical, but then:
Blacks setting needed to stretch the histogram over the full length and to see some first "black clipping": 25 and 55.

Empirically seen, the results seem to support conclusion # 1:
the raw process chain applies the Blacks slider before it applies Exposure.

Peter

--

No, what you are seeing is both of those shots were, arguably, under exposed...

I'm not sure Eric is in the position of giving a definitive answer but, seriously, trying to figure out what order Camera Raw/Lightroom is doing things is pretty much navel gazing...even if you successfully determined the order (which I do actually agree with you) so what? The point is to try to optimize your image to the best of your ability in Camera Raw/Lightroom–regardless of what the processing pipeline is doing...really, what does the image look like? Is there anything that Camera Raw/Lightroom is doing to KEEP you from getting an optimal result? If so, tell all, if not, go out and shoot something otherwise you're just gonna sprain your brain...

(really, I have a _LOT_ of experience talking to software engineers and I'm here to tell ya sometimes "stuff" really doesn't matter)

Yes, it is probably a good example why cameras should have an ETTR exposure mode. With a scene of limited DR, current auto-exposure tends to fail (as far as I can tell) because the histogram is stretched by the usual tone curve applied. In-camera jpg looks "normal". De facto the image is underexposed. For above example and the purpose of illustration, the relevant tone curve controls were set to zero /linear in CR.


Actually I was interested if the OP or anyone else would develop above conclusion # 3. IMO a valid point, also with regard to color integrity. But then I tend to conclude that the assumed sequence and execution of "blackpoint before whitepoint setting" (under the hood) is finally the better choice.


Yes, different subject, for example but not limited: "True Recovery" by blending differently processed Raw images.


OK. Always a good advice.

Peter

--

Peter, regarding #3: I do not think you can have too much ETTR being negatively affected by the Blacks in CR/LR. If you find Blacks are acting too aggressively, switch to the Tone Curve for a softer, more gradual rolloff of the shadows.

Regarding the "True Recovery" thread -- are you asking for an ability like this built into CR/LR to simplify the workflow?

Eric, - yes. Definitively.
Off topic though:

From what I can tell, such blending in Photoshop – based on one single differently processed Raw image – can easily outperform the Recovery slider in CR/LR as far as highlight recovery is concerned. FWIW, I think that such procedure could be adequately parameterisized. We are of course aware of Fill Light which seems to be much better construed compared to Recovery, but it appears to start at the wrong end for the given task.

Shouldn't such method of "HDR blending" - whether from multiple exposures or a one single capture - be a core competency of Raw conversion software,
while doing everything and having access to all the "non-treated" linear data.

Peter

--

Hi Peter, I would say this really falls under the broader topic of tone mapping (or "tone management" or "tone expression") -- at least, those are the terms I would use to describe these controls -- and yes, I agree these are generally useful to have in raw conversion software, particularly since the input dynamic range is known.