How to get the best raw exposure, part 2

Just published:

www.lensrentals.com/blog/2023/05/how-to-expose-raw-files-part-2/

I'll take questions and comments here.

Jim

Ding Ding. Round 2!

<withdrawn>

This is interesting stuff. Yet I wish there was some way to get more useful histogram data in-camera instead of having to wait until I can view the result on my PC.

Is UniWB do-it-yourself really the only way?
According to Fujifilm-X the 'Natural Live View" (NLV) should do the trick. However, I found a blogpost of yours where you report this to fail big time. The comments even rumour on the NLV uses an internal DR of 400%.
The only thing left is then selecting a low contrast JPEG film simulation (like Pro Neg Std.), setting DR 100% and turning down Highlight Tone and Shadow Tone to -2.

Thanks, Jim! I could follow all that. I had come to the same conclusions pragmatically and from what you wrote in part one. But where is all the mathematics you threatened? :) :) :)

David

At the end of the article. If that's too little math for you, I can fix that. 😊

Depends on how chromatic the highlights are and how accurate a solution you want. Some people don't like the green look that the EVF takes on with UniWB.

Not a single integral, div, grad, or curl! 😁😂

David

Bracketing is the other option. Use RawDigger or Fast Raw Viewer to pick the best one.

@SrMi That's indeed what I do in high-contrast situations, either using them to make an HDR or pick 1 as the best starting point for postprocessing.

Magenta filters (strobes / daylight).

and then some raw converters might need "you" to create your own profiles as canned might not work "well" ( granted that depends of course how strong the filter is CC10M for slight correction vs CC40M+ for example )

that is only if you camera can't switch WB effect off ... most I bet can - unless you want UniWB + blinkies/zebra in real time - not post shot review

That's what I want. Chimping to find out what the histogram is seems so old-fashioned to me now. Not to mention slow.

in some camera models blinkies work in post shot review too though (no need to use a histogram) + in some camera models you can assign WB/effects preview in EVF/LCD on/off to a button that helps too ... but of course a lot of people do not bother with all that all, so it is just a fine print and not much relevant

Some comments on Part 2. Perhaps nitpicky but intended to be constructive suggestions that might inspire some tweaking to this article or discussion in a "Part 3":

1. I don't understand why you say, "Looks like the blue channel is blown, doesn't it?" with respect to the ACR histogram of the hillside image. The highlight warning is not lit up and the rightmost slope of the blue channel is not vertical. It's slanted and doesn't touch the edge. Given that fact, I'm confused by your observation about the ACR histogram.
2. The LR histogram for the same shot also indicates no clipping. It also might have been better to avoid the apparent difference in the histograms between LR and ACR by setting ACR to output to ProPhoto instead of sRGB. As it is, simply stating that the apparent differences "are due to the different color spaces used by the two programs for histograms"dredges up another tricky, but otherwise unexplored, topic.
3. The hillside shot was perhaps a less than ideal sample image to use here or else a less than visually optimal Y-axis scaling choice in Rawdigger because you can't see any values/peaks beyond the main drops at around 6000 or so. For someone not familiar with Rawdigger, it looks like ALL of the highlights are "about a stop and a half from clipping," not just "most of the highlights" as you wrote. In my experience, the Min and Max values indicated to the right of the histograms in Rawdigger tend to be ignored or not understood by viewers only familiar with typical post-conversion histogram displays.
4. I'm having difficulty with your claim that, for the Hasselblad shot, you'd have to reduce Ev by 1 2/3 stops to bring the green channel below clipping in the in-camera histogram. Really? We know from Rawdigger that the green channel isn't clipped. And we know from the LR histogram that it isn't clipped post-conversion (presumably at default LR settings). And we know from the Phocus histogram that it sure looks like we're dealing with a really minor (and totally harmless) highlights adjustment. Yet, somehow the in-camera is still indicating clipping at as much as -1.5 EV from the ETTR image? I'd normally attribute the long highlight tail to specular highlights, but something isn't adding up here given what we actually know from Rawdigger.
5. To me, the Hasselblad shot showed that even a shot that is over-ETTR'd in the green channel according to the in-camera histogram is going to be nicely optimized for SNR purposes, provided that the clipping is in neutral highlights. A single blown channel in the neutral critical highlights as indicated in-camera is not a problem for most raw shooters using any raw converter with competent highlight recovery (not that recovery was actually even needed in this instance since it wasn't really blown).

Sony a74

Not accurate !

my zebras had a tiny black dots flashing.

Fast raw histogram

camera histogram

ACR
[]
(/a/tlLHm6J8lsER3dMtWH8Gqkrg0gBryY7XRjF6QE8AQ8UATJdr1q1tzWDFDN8kM6eS/7531/?shva=1)
faststone


field monitor, the black dots are the zebras

Post the raw file. Post FastRawViewer statistics screenshot. It's possible that your zebras are not accurate.