How to get the best raw exposure, part 2

In general, this is true. The amount of conservatism varies with the camera and the lighting.

I shot a whole series of exposures at 1/3 stop intervals, and went through them all until I found the greatest exposure that showed no clipping in the in-camera histogram. That's where I got the 1.67 stops.

I used the default settings, figuring that that's how most people would be using the programs.

I probably should have weasel-worded that. It was looking to me like ACR was doing its thing with the artificial shoulder region. I should have said "close to blown".

Jim, I thought those two articles are fantastic. I have so many questions, but I thought I’d begin with this:

My observations with my 5D Mark IV are that clipping is not a cliff. As highlights approach clipping, the highlight regions appear to lose contrast and detail. In other words, I’d expect lots of detail and some area of solid white, but instead, detail just fades away. Have you observed this?

Good read Jim.

As a jobbing people photographer who works on film sets and general events with sometimes rapidly changing light (think outdoors in bright sunshine one minute then inside a dark indoor space the next minute) I don't really faff about too much.

That said I do understand putting maximum light on the sensor without clipping and using base ISO for maximum IQ, for other genres where time allows.

For a jobbing photographer the trick is to understand the theory, whilst coming up with a way of shooting optimally in dynamic environments where you simply haven't got time to faff about. You want the tech out of the way, not slowing you down.

I shoot Sony so this is what I do:

1. Sony have zebras for stills and you can set the zebra threshold levels. In Sony speak, if shooting raw, you set "custom value 109+".
2. Basically, I'm telling the camera I'm shooting raw and the camera knows I have more clipping headroom at this 109+ setting (which is the maximum value you can choose).
3. If shooting jpeg you'd set the zebra threshold lower than 100. 90 is a safe bet.
4. I then shoot M mode in auto ISO because I want to control the minimum shutter speed to meet my motion blur requirements (front dial) and aperture for dof requirements (rear dial).
5. I can put more light on my sensor at any time by opening up the aperture and/or slowing down the shutter speed on the fly.
6. I have apply settings to live view switched on, because I want to. Auto ISO means I can then see what is going on in the EVF.
7. My Sonys have an EV compensation dial immediately to the right of the rear dial.
8. My EV dial is set to zero by default when I'm shooting.
9. But if the shooting pace is leisurely enough I take the time to dial in some positive EV on that dial until zebras appear, then ease back by 1/3rd of a stop.
10. This approach does not put more light on the sensor, because I've already set my aperture and shutter speed, it will simply lower ISO if I dial in +EV.
11. In practice this gives me perhaps a bit more DR in post for cleaner shadows, although in reality it doesn't make a huge amount of difference to real world results.
12. I find this approach using zebras with my Sony's renders the live histogram redundant. I don't have to chimp either.
13. If I was shooting say a landscape on a tripod I'd slow down, set base ISO, and ETTR using zebras to put maximum light on the sensor for sure. Because I understand that.
14. In the studio using strobes I do not use auto ISO. I set to base to ISO100 instead. But sometimes, if I need more light than my 600w strobes can deliver, I'll switch to ISO640 on my cameras. Because they have a dual gain sensor, with ISO640 being the switchover. In other words, the second best ISO after base on my particular cameras.

That's not too different from what I did when shooting Sony. The key thing to remember is that ETTR only applies at base ISO.

I haven't used Canon gear since the aughts. However, I'd doubt if the raw files exhibit a shoulder region. My guess is that the shoulder you're observing is coming from your raw developer.

Yes I understand that, that's why I said by using EC in auto ISO all I'm doing is reducing the ISO for not much (if any) gain. It makes me feel better though :-)

But, here's a question then Jim. I don't think dual gain sensors have been discussed specifically.

You say ETTR only applies at base ISO. What about dual gain threshold ISO640 (on my A9II's) which would be ISO800 on a Z6, ISO400 on a Z7 and so on?

One difference between us that I never use Auto-ISO.

Here is the exposure strategy I use with several dual gain cameras:

blog.kasson.com/x2d/hasselblad-x2d-exposure-strategy-m-and-a-modes/

blog.kasson.com/gfx-100/fuji-gfx-100s-exposure-strategy/

Here's an example of using low-ish ISO in low light:

blog.kasson.com/nikon-z6-7/nikon-z7-exposure-strategy-an-example/

Before raw conversion or after?

Hi Jim and thanks. Your post regarding GFX-100 was best aligned to my question and makes perfect sense. For other folk reading this, the make of camera can be disregarded, if you have a dual gain sensor camera the same observations apply, subject to finding out the dual gain ISO for your particular camera.

I'd like to dig a little more here, for the benefit of beginners and the curious. I'm in the curious camp. You state that you don't use auto ISO and I understand why, but we shoot different stuff.

To dial that in a bit more, let's consider the likes of myself that shoot on film sets and events in general, in very dynamic and changing lighting environments. This might also apply to folk such as wedding photogrphers.

I've set out how I do it and why. I'm very curious as to whether you'd still use a fixed ISO for my kinda bread and butter work as opposed to my "M mode auto ISO strategy". And if yes, how and why?

Finally, if you do propose an alternative way of working would it give a dramatic gain in IQ without slowing down how you shoot in such dynamic environments?

I think I'd like to get to a point where we have clearly understood best in class strategies based on genres shot.

I observed this using Canon DPPro. If I’d pull down the highlights, they were just… blobs. No detail. They didn’t show highlight warnings in DPPro.

I’m not as scientific as you and Jim. For me, it was more like “Ouch, I burned my fingers! Don’t do that again!”

What you are observing isn't the camera behavior, it's how DPP is designed.

Wow! Thank you very much! That’s pretty surprising; I didn’t imagine it was the Canon software. I gave up on DPPro about six months ago and I’ve been using DxO PhotoLab, so maybe I won’t have this problem any more.

Here is how clipping looks like on a raw histogram:

1drv.ms/i/s!ArStsPjQ301PmTGIRYjPX5i7OTOO?e=GpypBH

will shoot some more tests today as i have a day off work.

Raw file
1drv.ms/i/s!ArStsPjQ301PmTJktgwBeeArFb21?e=gxNLgG