How to get the best raw exposure, part 2

There are modes in which that does happen. In the kind of cameras we're talking about, there are modes where it doesn't. It's important to keep the discussions of exposure and ISO separate, or we'll just go round in circles.

I basically failed English at school. and so be it, it didnt stop me raising an academic daughter. but i also get tired of the anal correct terminology discussions.

When you say you use JPEG and sRGB, do you mean that you have the camera write the JPEG files? Then my article doesn't apply to you.

Or are you saying your final form is an sRGB JPEG, and you shoot raw? In that case, what tone curve is applied in postproduction is entirely up to you.

Im going to ask my daughter whether that is true or not . should be interesting

Well that was interesting Bob. my daughter said whoever wrote that is a wanker, insinuating the other person is in her words Dumb. in fact she said you made so may mistakes and didnt really use Precise correctly.
sorry cant reply as ive been called to a job.

I thought that you were objecting to the principles that I espoused, as amplified by Iliah. If I'm wrong, I apologize. But you must see that there is no reason why you have to deal with a jet black viewfinder if you do what I have proposed.

That's only if you use 'exposure simulation' (how it's called in Canon cameras). I think the exact technique hugely depends on the brand and even concrete camera model.
When I use ISO-invariant range with my Canon R5, I'm forced to disable the 'exposure simulation' for two reasons: a) better view in the EVF b) the camera struggles with AF if EVF is too dark.

When I ETTR, I'm forced to enable the exposure simulation' because otherwise the camera doesn't show live histogram.

Not an experiment and not funny - just plagiarism.

You copied just the bottom half of my post.

Above your quote I said

"does not even come close to proving your opinion"

I was saying DonaldB's comment doesn’t prove his opinion of Fast Raw Viewer is true.

If you're worried about really deep shadows, it's worthwhile to avoid the ISO setting a third of a stop below the first ISO that has the high conversion gain.

But once you've had to raise ISO because your chosen exposure won't give you ETTR at base ISO, then there is little advantage, and possibly a sizeable disadvantage, in trying to find an ISO setting that pushes the histogram aggressively to the right.

Sorry, I was having a little bit of fun.

If you disable exposure simulation your histogram and blinkies/zebras will be useless. I wish we had cameras that would allow for correct histograms even with exposure simulation turned off.
How do you handle ISO in the invariant range that avoids clipping?

No apology needed. But equally, surely you can see why event, wedding photographers etc don't feel the need to go there in the first place?

We understand what you are saying and we don't disagree with the facts you are stating.

We are simply saying that we don't want or need it!

10% more complexity, 10% slowdown in fluidity of working for a zero increase in real world deliverable IQ for our clients.

Surely you can see this Jim?

The more we can dial our cameras in so that the minute we switch them on we are ready to shoot the better.

The absolute worst place to be would forever changing our settings and then having to remember to change them again and risk being in totally the wrong place.

When I switch my camera on it defaults to 1/250th, f5.6 auto ISO. I'm immediately ready to grab a shot without thinking. That's priceless. I can dial it in from there as needed. But getting the shot ALWAYS comes first, second and last, I'm not thinking about an optimum raw file, aside from dialling in my settings as the shoot progresses.

I'm deliberately using auto ISO, as using a fixed base or 640 involves faffing about with other settings to maintain my EVF view, which I then have to remember to change next time.

I aint going there, nor would anybody else I know in the world I inhabit.

Let's take a closer look.

With a linear y-axis, there are no clipped pixels to be seen. We can tell there are some from the numbers, but the count has to be low. With a log y-axis, you can see some green channel clipping.

But the X2D doesn't have a log y-axis option, so we can't make that comparison.

Didn't I already say that without understanding your constraints and goals -- and probably even after I understand them -- that I'm not going to tell you how to do your job? If you've got something that works for you, carry on. And if your clients don't care about the extra IQ that you might be able to get if you change your methods, then that extra IQ is worthless to you. Again, carry on.

Thanks for your reply.

I have the camera write both raw and JPEG, but only use the JPEG from the camera as a quick preview. I always edit the raw file. The final form is sRGB JPEG. As a raw converter, I use Canon DPP or libraw/dcraw_emu or rawtherapee and in case it matters I have not used any Adobe software for many years. I gave a Canon DPP example and a libraw/dcraw_emu example. Any printing is done by a service (fineartamerica) that only accepts JPEG, but the prints seem acceptable to me and when others buy them satisfaction is guaranteed by the print service.

So far as I am able to understand, the extreme values for sRGB pixels are vulnerable to loss of detail and color shifts if one color channel is saturated. So far, I try to avoid saturation or clipping of values in one color channel if it contains detail that is important to me. For the sea gull image I used as an example, the reddish coloring that was clipped in the out of camera JPEG is used to identify the species. Changing the color by clipping the red channel would obscure that information. By taking the raw and expanding the dynamic range enough so that it was not clipped when I produced an sRBG JPEG I preserved that information. If I had used a histogram, I would never have known that the red channel was clipped in the important parts of the bird because it is so few pixels.

Extreme changes in the shape of the linear to gamma curve result in artifacts that I find unpleasant, so it is not practical to use any curve I might choose.

Creation of a gamma curve from linear raw data results in further loss of information especially at the bright and dark ends of the curve. How would ETTR preserve that information? Applying the gamma curve involves integer to floating point to integer conversion with a result of quantization error which I would sometimes like to minimize.

The out of camera JPEG for my camera always discards some highlight data which may be recovered when processing a raw file, but it often is not enough when the important part of the photo is a bright bird or flower. To preserve that information, it is often necessary to tell the camera EV -2/3 to get even more detail into the raw file, yet I cannot see this in the histogram.

In the two histograms that I include in my post, the second one might lead me to believe that I could use EV +1 or EV +2 with auto exposure even though that would mean loss of important detail.

I should have used a photo of a flower in sunlight or a Northern Cardinal in sunlight for my example since that would show it better. Instead I used a recent photo. A quick switch from perching bird in shade to flying bird does not leave much time to second guess the auto exposure chosen by the camera.

Thanks again.