ETTR with Olympus OMD

Software darktable
1

I need to tweak my Olympus OMD E-M5 Mark III settings so that I would ETTR more accurately. Currently there is a pretty significant difference between the histogram shown on the live view camera monitor and that shown in darktable.

I’ve read this blog post by P. Forsgard on Exposing To The Right with Oly OMD cameras:

Apart from suggesting using the muted picture mode plus other settings, he also suggests to set the AdobeRGB color space rather than sRGB.

Why AdobeRGB would be better to get more accurate clipping information from the camera monitor?

2

ETTR is a good idea for most cameras, however some new cameras, like Fujifilm XT-3, have invariant ISO and do not benefit from ETTR. But that is a side issue since your camera most likely benefits from ETTR. Now the histogram (scope) displayed by Darktable may look very different to that displayed on the camera. I believe the camera displays the histogram of the JPG image it would produce (even if you are shooting in RAW), but I am willing to stand corrected if someone knows different. Then when you look at the histogram in DT you have numerous options for setting what the histogram displays. In the screen shot below you can see I have my histogram set to work profile (selected from the drop down menu as linear Rec2020 RGB for my computer). Experiment with the options here and see if that helps.

image

Adobe RGB is a bigger color space and I agree with his choice for the camera setting. Shooting in RAW doesn’t affect the use of muted picture style as this is not affecting the RAW file except for the embedded JPG.

BTW, when I export for end use from DT such as a JPG to share on line I export as sRGB since most monitors are limited to and expect to see this display. When I export from DT as an archival Tiff file for further editing later in programs like GIMP I export as Adobe RGB as it has a wider gamut of color. Very very few people have Adobe RGB capable monitors so most monitors are limited to sRGB.

I hope this information is helpful. Also don’t push to far to the right because recovering clipped highlights is a nightmare in any program and even if the highlights are not clipped the saturation of colors near the extreme can be reduced. But ETTR does reduce the noise for many image sensors.

5 Likes
3

Histogram displayed is what the jpg produced by the camera would like. So what you see on such mirorless camera. So it’s just normal that it’s different on darktable.

Setting Adobe RGB or sRGB is only needed for jpg produced by your camera. Raw don’t have color space. That’s why darktable have an input color profile module.

Such reading will detail that more: Why Choosing the Wrong Color Space Can Be a Disaster | Fstoppers

2 Likes
4

I shoot with an Olympus M10, so the settings are similar. I use very similar settings to what he recommends, but I use the Natural instead of muted and sRGB. I get consistent results using those settings.

I dont love the term ETTR. A better term is to capture the image with the most amount of energy without clipping. Energy is mainly governed by shuttle (duration of light reaching the sensor) and aperture (amount of light that enters thru the lens). Go as high of camera exposure/histogram as possible, but the raw clipping indicator on darktable not to show clipped areas. Try to find how the camera histogram looks when it does clip, so you can avoid it. Basically, build a correlation.

1 Like
5

Thanks for the clarification about this. It is good to know.

6

Histogram and highlight/shadow visual aids shown in OMD camera refer to the picture shown in the live view, i.e. a jpeg image that the camera makes on the fly.

1 Like
7

Yes, I see, but I think that I might tweak the jpeg settings, so that the in camera histogram as well as highlight/shadow visual aids look more similar to what I get in darktable (histogram and raw clipping indicator).

Wouldn’t color space have an influence either on what I’m after?

8

Ideally you don’t care about any resemblance between the in-camera jpeg and darktable. What you would like is a histogram reflecting the raw capture. Or at least one that reliably reflects the areas clipped in the raw data.

So probably in-camera settings with low contrast, low sharpening and perhaps the larger colour space (AdobeRGB). I’m not really sure about the importance of the colour space selection, concerning raw clipping indications. And of course, your camera jpegs won’t look nice at all with such settings (but who cares if they do what you want?)

Then again, while ETTR is very useful, in practice you can have some problems: birds don’t follow posing instructions (or any other instructions, for that matter). So one shot can be backlit against a very bright background, the next one (seconds later) against a dark background. And you often don’t have all that much time to fiddle with camera settings.

1 Like
9

For some cameras, ISO (gain) can also play an important part of the equation - e.g. one can get much better signal-to-noise in the shadows at higher ISO, but you have to be careful not to clip if highlights are important. There’s always a compromise to be made of course, there’s no free lunch…

This camera, however, does not have dramatic shadow improvement (hence the mention of “ISO invariance” above).

10

I would really think about ‘why ettr’.
Maybe you shot some outside pictures where you underexposed a bit and STILL had raw clipping. So you want a more accurate warning .

That might be a good reason. Trying ettr just because people say so, i always advice against it and be careful.

Fact is , over exposed highlights are a pain to deal with , shadows that have a bit more noise are often not an issue at all .

So in my mind , the worst that can happen with ettr is pretty bad, and happens suddenly . While the worst that can happen without using ettr , is shadows that are a tad noisier and no problem at all.

I have my em10 mark1 said to be pretty punchy in the jpg is produces , and even set it to a pretty low resolution and have a bit of shadow boost and highlight recovery.

I do that because I use those jpgs for sharing with my mobile , so i want to have a sort of ‘ready and done look’.

With these settings , i notice on a bright sunny day i can go down to -1.5ev and be safe on the shadows , and almost never have clipped highlights (no big areas anyway ).

Your best bet might be to just set the settings as you want. And then clip something on purpose (bright lamp or something ?) . Go down in EV until you don’t clip. Then look at the camera histogram where it is. Now you have a link between raw clipping and camera histogram with YOUR settings.

Learning what your metering does i maybe more useful then hacking the jpg look into something for a histogram that will never be a raw histogram. Just my 2cts.

11

Yes, if you want to use the camera jpegs, they have to be good. And the in-camera jpegs can be more than good enough for the intended use (some professionals use them for publication-worthy material, and I’m not referring to the local newspaper).

As for learning your camera: always a good thing. But there are situations where the camera doesn’t quite react as you expect, and those usually cause important clipped areas (saturated colours are “nice” for that, it’s not always the brighest areas that clip).

And if you never use the camera jpegs as such, you may as well have the settings such that you get a maximum of useful information.

Also, none of this is “either/or”, they are all means to an end: a properly exposed raw file. What “properly exposed” is depends on the photographer, his mood and the scene… E.g. shadow noise: handheld at 300 mm on a 1.5 crop sensor means very little light to play with, so high ISO. Underexposing isn’t nice in that situation, some blown sky is less of a bother.

2 Likes
12

So-called ETTR with most any camera is a pain. Most of the available metering mechanisms are anchored to asserting a middle-gray, probably well and good for the intended subject but not so good for light energy at the upper end.

I recently reverted to my Z 6’s middle-gray-anchored matrix metering for a particular situation, forgot I’d set it and used it in an “engineering capture” session where the subject was the old railcar we’re restoring, taking pictures of various structural devices as we removed the siding. Afternoon sun making stark contrast, lost a lot of detail in sunlit portions. Here’s an example:

DSZ_1943
DSZ_1943800×532 120 KB

Yeah, coulda dialed EV down a bit, but in these situations I’m concentrating more on framing the subject and keeping from falling or hitting my head in tight spaces or tipping over in compromised postures. The project coordinator used this and a few other images in a presentation, and there was comment on the washed-out highlights, from non-photographers at that. In these situations I just want the metering to just work.

The Z 6 has a highlight-weighted metering mode, and that’s what I’ve been using by default since I got the camera. In most cases I think it sacrifices about a stop from “proper” ETTR, but dangit, it never loses the highlights I care about. And, with the FF Z 6 low-light-beast of a sensor I don’t usually have to worry about pulling up shadows, except for a loss of color contrast. I can usually restore that with a bit of color saturation, early in the toolchain.

Recent Fuji cameras have some sort of raw-oriented metering mode I’ve only seen described in third-hand literature. Would be interested in hearing others’ experiences.

Oh, don’t know about Olympus cameras. If they don’t have a highlight-preserving metering mode I’d be hard-pressed to consider purchasing one.

Thing is, any sort of highlight-preserving exposure strategy is going to require attention in post, as the range of lighting just differs scene-to-scene. I’ve had highlight-preserving exposures that required no “look” tone curve (e.g., filmic, sigmoid, and their ilk) at all, just the TRC in the display/export transform. But, those situations I can count on the fingers of one hand, most require some sort of lift. And, that is the burden of highlight-preserving, manual attention to the rest of the image. I can’t posit any algorithm that substitutes for that attention, and I’m not so sure an AI approach will suffice in all situations.

I’m working through our Christmas snapshots right now, most shot with picture window lighting filled with on-camera flash pointed at the ceiling. The metering mode is the highlight-weighted matrix, and what I do is take the first image of similarly-lit sequence, edit it to suit, the delete the other proofs of that sequence and re-process them with the same toolchain. rawproc supports this operation quite nicely with the batch tool, so I don’t have to give each image singular attention. Definitely anathema to SOOC JPEG folk, but I like it.

13

Me too, as Fuji’s metering has caused me nothing but pain. I find Nikon’s metering to be much more refined. I use the matrix metering mode and adjust manually to ettr while using the flat jpeg profile and I find it to be very good.

14

But be careful with the exposure settings, the amount of light does still influence the image quality even in ISO invariant cameras.

2 Likes
15

Okay, I did some digging and found this:

Ha, In the TL;DR it specifically says, “Using Natural Live View to evaluate exposure is not recommended, even though it seems like a perfect use case to bypass JPEG processing. Natural Live View doesn’t reliably indicate clipping when using Live View Highlight Alert or the histogram.” Okay, so is that recommendation for all of us, or just the folk who won’t understand the implications… ??

Thanks anyway, Fuji; Nikon’s HLW matrix metering works well enough for me…

16

I enabled that mode when I first got my X-T3 because the Fuji manual makes it sound like exactly what I wanted. But it is not and I ended up clipping the highlights really bad on a few photos I really liked otherwise. I share the WTF-ness of this mode… WHY

Metering is the main reason I’m moving away from Fuji and just refreshed my Nikon gear. The Z7ii is only slightly larger then the X-T3 as well.

2 Likes
17

Cool; if you use the HLW metering mode, I’ll be interested in your take on it…

18

Nah, just matrix metering. Shutter adjustment on the front wheel, aperature on the back wheel. ISO at 100. ETTR. That’s pretty much it. I mostly be that soft light so I rarely need to bracket.

2 Likes
19

I’ll add my bit to all the good avice posted so far. You’ve got to get to know your camera and apply ETTR judiciously and according to the situation.

I’ve found that my Olympus E-PL5 very often produces under-exposed raw files, leaving one to two stops of light on the table in (I guess) an effort not to clip (it does not have a huge dynamic range).

So, I’ve learned through experience when I can safely add 1/3, 2/3, 1 or even 1 1/3 stops to my exposure. This gives me raw files that require less exposure compensation and are easier to work with. This is not hard-core ETTR but works quite well for me.

As a counter-example, I often walk around with my old Canon S90 point and shoot. On that camera, I often have to underexpose, because it will very happily clip all over the place. Its dynamic range is tiny. However, when shooting raw I don’t have to underexpose so much; its internal raw to jpg conversion is quite suboptimal in how it handles highlights.

2 Likes
20

Having been a professional photographer I really wanted my in camera JPGs to look great so at the end of the shoot the job was done and no time wasted on editing images. However, by shooting in RAW and JPG I ensured I had the best file for editing if required.

When I go travelling I mostly shoot with exposure bracketing set to +/- 1.0 EV. It covers my arse and I usually find the best exposure is the brightest one that DOES NOT HAVE CLIPPING. I can not stress how much I hate trying to fix clipped highlights. I will take noise over clipped highlights any day. Also, with my holiday photos I really enjoy spending the time developing them in DT so RAW is my go to, but I still shoot in RAW plus JPG because just very occasionally the jpg looks better than what I can edit (very rare but it does happen).

2 Likes