How to determine the best exposure for raw files

I will accept that it's the way it goes for you in real life. Using studio strobes, I never use anything but base ISO, and almost all the time, I have to dial back the power on the strobes.

thats funny ive never seen anyone actually show a side by side comparison between a field monitor and a raw histogram ,care to post a link ? may be all the professional videographers have no idea what they are doing.

"200ws" battery powered one that might not deliver enough light on a group of people to fully saturate those silicone buckets ... it is a party, I do not haul more gear than let me be a part of it vs mind the strobes from being destroyed :-)

agree shooing at base iso all the time is just paranoia. we may as well get into the debate as whether its better to use ND filters instead of high speed sync 😁

my camera blinkies shows highlight clipping exactly to the scale of your raw histogram. my suggestion for you is buy a better camera 😁

One example, and without even a JPEG histogram / blinkies scan/shot from the back of the camera, proves nothing. But you know that.

My idea of a better camera may be totally different from yours :)))

From what I learned at dpreview the best exposure* for raw files is when you have allowed the most light as possible to hit the sensor within your DOF and motion blur constraints without clipping important highlights.

The easiest way I know to do that is set the widest aperture (smallest f-number) that gives the desired dof and the slowest shutter speed that meets blur requirements. Then let the camera set iso where it likes as long as important highlights are not clipped.

If the camera sets base iso and highlights are clipped then you need to compromise on either dof or blur requirements assuming you cannot lower the light from the scene.

If there is no motion in the scene then the decision is easy - set a faster shutter speed.

The final image lightness is set in post.

exposure* - amount of light that struck the sensor per unit area while the shutter was open.

Im away camping so limited resource's, and no field monitor. the rear ldc is tiny and hasn't the resolution. but the blinkies dont care about resolution. and i knew i had already posted a raw histogram so that was easy to download.

Is that at variance with what I said in the article?

I would hope not because it would mean I understand the concept of setting the best exposure for raw files incorrectly.

What I posted is just the quick and simple short summary in layman's terms I use.

You are constantly changing your tune. First you claimed that "all cameras histograms are calibrated on the safe side of clipping thats all". Remember what I've asked? Now you are dragging blinkies into it, and not posting any proof, not even a shred of it, that histograms are somehow calibrated to be useful for a raw shooter or that blinkies can always be trusted. Don't bother, unless you are presenting proper facts.

Why is setting the ISO part of the best exposure? If I have time, I'd rather set the ISO myself considering the point where dual-conversion gain kicks in and how much highlight headroom I want.

If you are at base ISO and highlights are clipped, then you need to increase the shutter speed, no compromise of blur requirements is necessary. If the shutter speed limit is reached, then you need to increase DOF.

Unless you want to blur the motion, it is always safe to increase the shutter speed if there is motion in the scene.

They're invariant enough (at least my Fuji/Sony sensors are) that, if there is a ton of dynamic range to deal with, you can set an ISO well below the dual-gain threshold and still have sufficiently low enough read noise to produce a clean result. Is the sensor as invariant as above the dual-gain threshold? No, it is not, there is a price to pay in read noise, but is a very minimal one. Sensors aren't perfectly invariant above the dual gain threshold either, there are no absolutes here. As I'm mostly shooting APS-C, after maximizing my exposure, I generally shoot with whatever ISO will keep my important highlight detail just below clipping to keep read noise at an absolute minimum, and only intentionally "underISO" in dynamic/unpredictable low light/high DR situations and, even then, not usually more than a stop or two.

This was shot at base ISO (200) instead of ISO 3200 to maintain sufficient highlight headroom for the very bright signage. Would the shadows have been cleaner if shot above the dual-gain threshold (ISO 800 in this case)? Yes, but not by much, and some highlight detail would have been lost. A low light shot with an old Canon sensor (which I consider to be non-invariant) would have been a mess with a 3 stop push in the shadows at any ISO.

Thank you Jim, the article is brilliant and easy to understand for people like me (with math, computer science and a bit of physics in the background), although could be perceived as a bit technical for those without such education.

The only comment I have to make is about the term 'visible noise'. The amount of visible noise will depend on mapping from raw data to jpeg (the whole processing chain), e.g. you can process it to 'low key' and have very little visible noise that'll be lost in the deep shadows, despite high noise in raw data - and that's even before the noise reduction. Does 'visible noise' imply the default in-camera jpeg engine? But then the article specifically mentions it's about producing the best raws, not jpegs.

That is, the term 'visible noise' is not very well defined without mentioning the specific processing chain.

True enough, but I was trying to keep it simple, and, all else equal, what counts is SNR. You can of course reduce shadow noise visibility by crushing the blacks, and increase it with shadow pushes

The basic question that arises here is: Are you a Troll or a Moron ?

Nowhere have I said that it is. I was just commenting that I leave the camera to set ISO where it likes after I have set the exposure*

Not necessarily true because if you read my post you would have seen I said

Setting a faster shutter speed might eliminate wanted blur in the image.

You are just effectively repeating what I already posted.

Again, you are repeating what I posted earlier

That's fine. Just do what works best for you.

* exposure - amount of light that struck the sensor per unit area while the shutter was open
** optimal exposure - the maximum exposure* within dof and motion blur requirements without clipping important highlights.
*** under exposed - more exposure* could have been added with the DOF and blur constraints still being met without clipping important highlights.

A sweeping statement, indeed. Any references that back it up for "all cameras histograms"?

It's RawDigger ... and asking people to compare an in-camera RGB histogram to a raw data histogram is straining your credibility, IMNSHO.

Excellent ... as usual.