How to determine when the midtones are properly exposed using the exposure module?

Using the scene-referred workflow and following the documentation, the exposure module should be used to set the midtones (before adjusting the highlights and shadows with filmic rgb):

Use the exposure slider to adjust the midtones in the image to an appropriate brightness level. At this stage, don’t worry about highlights and shadows – these will be handled later.

How can I determine when the midtones are at an appropriate brightness level? I am using the darktable-elegant-grey theme and have been trying to match the midtones to the roughly 18% gray border around the picture, but I find it is still hard to tell if the photo matches the border or not. For example, here’s an image before I start making any adjustments:

And here are a couple of exposure adjustments:

How can I tell when the midtones are bright enough (and not too bright)? I realize that it is tricky since I am looking at a screen emitting light which will make it look brighter than a print (which is only reflecting light) so I tend to err on the side of making the midtones a bit brighter than I think they probably should be. Does anyone have guidance on how to properly expose the midtones with exposure?

One suggestion someone posted here was to use the channel mixer to quickly convert the photo to black and white to make it easier to compare to the 18% gray border, however even then I find it hard to pick which of the grays in the image I should be comparing to the border:

Thanks for the help!

The very literal answer to this is to drop a color picker on what you want to be the midtones, set the picker to the LAB color space, and L=50 is the middle.

The better answer is to edit more and more photos and you’ll eventually end up with a sense of where you want your tones to be either while you’re at the scene or while you edit.

1 Like

Your eyes will tell you. That’s an artistic decision and an arbitrary choice from the author. Color science stops here, you are the guy making the image.

It’s like playing violin. How can you tell if you are off-key ? Well, you have to train your ears, and then you will know.

3 Likes

And as in playing the violin, as long as the intervals are correct (and you are playing alone), it doesn’t matter if you are a bit off on the base note (unless someone around has an absolute ear/eye)

2 Likes

Thanks for the guidance. Would you recommend exposing it a little brighter than what I think looks best (if intending to make prints from the photo) since it will appear brighter on a monitor than a print?

Timely question for me, I’ve been considering just this of late…

A few years ago, I took on ETTR as an active part of my shooting, looking for a way to keep the highlights of my images in the resolvable dynamic range of my camera. The quality of my captures improved significantly, but recently in looking back at the renditions I’ve noticed a tendency for me to develop ETTR images too darkly (darkly? Is that a word??).

So recently, I’ve picked at that tendency in my developing, and what I’ve found to work for my predilections is to concentrate on the subject of the image, getting it into the right tonal range for the image to draw attention. Then, I try to shape the tone curve to help the shadows and highlights as best I can with a filmic curve. Now, knowing the image’s subject can be a bit vague, as some of my images tend to be about groups of things in a composition, rather than just one thing. But then, I back up to consider that composition and what tonality best suits it. Seems to be working for bear-of-little-brain here… :laughing:

5 Likes

That makes sense to me - just try to focus on the subject and the composition and that will ensure that the exposure is correct.

Speaking of ETTR and dynamic range, someone in another thread recently suggested that I look at HDRmerge and use AEB. I’ve been trying this as an alternative to ETTR in situations where it makes sense (still subject, enough light for the shutter speed to be fast enough) and it is an intriguing idea because it eliminates the guesswork associated with ETTR and you don’t have to sacrifice any of the highlights. Previously when doing HDR work with Luminance HDR or Photomatix, I was using one of those tools to do the tonemapping (which often resulted in poorer results than when I was using my normal workflow with darktable). With the HDRmerge plugin, I can keep everything within darktable and use my normal post processing workflow on the photos, so it seems like a nice way to add more data to the shadows and highlights for scenes that work with AEB.

Thanks for the question I have been pondering about the same issue. The main issue I struggle with is how do you get the overexposed area’s back when you push the exposure to the right?

So you ‘optically’ evaluate your image. You decide it’s too dark. You increase the exposure. By doing so you push your highlights over the edge… Like your 2nd and 3rd image.

I’m talking about highlight information, not clipped channels or totally burn out highlight info.

How do you get them back? By lowering the white point in filmic?
If I do so, the colours start to turn mushy and faded…

It’s fairly normal that the whites end up over-exposed when you set the gray point. In filmic you recover them by pushing the slider for the white point to the right (increasing the value). The desaturation is a normal side effect.

To recover the saturation, you can add saturation ("middle tone saturation"under the “look” tab); iirc, the default is ~10%. Also, changing the “preserve chrominance” option (under “options”) can have an effect. The “best” option depends on the image, so you may want to try several.

1 Like

I have discoverd that you can use the tone equalizer to bring the highlights “back in check” by lowering the nodes on the right side (highlights) and leaving the filmic module as is…

IMHO bracketing is a good tool to capture a high dynamic range scene if the subject allows. What I’m finding with my shooting is 1) certain subjects like moving railway trains just don’t work with it and 2) the still lifes I’ve recently taken up may have significant shadow components but i usually end up just pushing them down into dark oblivion.

I’ve gradually shifted my thinking about exposure at capture from the midtones to the highlights. I now work to anchor the exposure to preserve highlights and work the rest of the image in post. That isn’t usually ETTR, as I currently lean heavily on my camera’s Highlight-Weighted Matrix metering, which is JPEG-based. With the Nikon Z 6, I’ve found there’s enough dynamic range to let me get away with that in most cases; with the old D7000, not so much. A lazy approach, no doubt, but I grew rather weary of the ‘stupid pet tricks’ required to do balls-to-the-wall ETTR…

The Z 6 also has a two-exposure HDR mode, lets you capture two successive images and it makes the merged JPEG for you. Decent performance, and it’ll let you retain the two captures as NEFs to be played with in post with HDRMerge or the like. The two captures are very close to each other in time, which makes a disciplined hand-hold viable in a lot of situations.

Some one did an analysis recently of a few newer camera’s and the summation was that the effort put in to tweaking ETTR was not longer worth it in newer cameras. The noise so much better controlled and the DR so large that gains were really minimal both visually and statistically. I will try to find the article but it seemed like a reasonable breakdown and again perhaps only true to a point or with certain cameras and lens combinations…

1 Like

I found this article to make sense in most cases and I think the most important comment was around composition etc…if you can manage all aspects and get perfect ETTR then I guess great there are some possible gains but if focussing on the ETTR doesn’t allow for other aspects like lighting composition and focus to be given as much attension as they should then perfect ETTR on an out of focus subject won’t be much good…between improvements in sensor quality and the auto bracketing that goes on in HDR modes now on phones and likely newer cameras of which I don’t have a good example …there may at least be diminishing returns to ETTR unless you know your camera so well you can quickly dial in the right amount of extra EV from experience…https://photographylife.com/exposing-to-the-right-explained#:~:text=In%20theory%2C%20ETTR%20works%20with,ETTR%20is%20all%20but%20useless.

How about submitting it to the Play Raw section and getting people’s interpretations?

Some people prefer warmer images, some saturated, some contrasty, some not… a lot of variation. Images should be under a CC license. There’s also a lot of variation based on people’s screens and whether they are using “Night Shift” or “Night Light” :wink:

I have been “doing” photography since the late 70s (well, since the early 60s if you want to count my early days with a Kodak Instamatic), and I sheepishly admit that I had to look up “ETTR”. Here is a pretty good explanation, if anyone else needs it:

Exposing to the Right

It is the opposite of what I have been trying to do for the past ten years or so. I have apparently been trying to ETTL, because the auto exposure settings on my DSLR often result in blown highlights, combined with people explaining that it is easy to bring up underexposure, while it is impossible to truly recover from blown highlights.

So, which method do or should “we” prefer?

If shooting RAW for later processing use ETTR, if working for Reuters use Jpegs and expose for the subject ignoring over exposed and under exposed areas (though with ten stops of dynamic range having detail in undesirable areas can be an issue).

Weddings are an interesting scenario where the photographer comes under a lot of pressure to show the pictures to the bride earlier, and yet what is on the rear of the camera as a Jpeg may be underexposed in a safety margin to retain the bride’s dress detail.

1 Like

@Tim ETTR makes sense when we talk about digital photography, because we can generally assume that the photometric response of the sensor is linear until we hit cipping. In such situations, we want to maximise the signal-to-noise ratio by increasing the signal across a more-or-less constant noise floor. We can then adjust the gain digitally in post-production and apply a saturation curve as appropriate to the scene.

In the case of film, the characteristics of the film emulsion are fixed, and so it becomes much more important to set the exposure in-camera so that the key midtone aspects of the scene lie within the latitude of the film. In this case, we still want to ensure adequate exposure to avoid having to “push” the film too much during development, but ETTR itself doesn’t make as much sense.

2 Likes

Agreed that it doesn’t make sense in all situations, though I have found that reducing noise in the shadows (even if the scene doesn’t have that high of a dynamic range) can still be valuable since I often end up wanting to brighten them at least a little. Not having to try and calculate values for ETTR and “just shoot” the AEB is a nice feature to this approach.

That is a great feature. Does it let you take the successive images at different ISO values, or just different shutter speeds like AEB?

Thanks for the link - I agree that shooting at base ISO when doing ETTR makes sense (aside from noise, dynamic range is greatest at base ISO too). It seems hard to quantify how much new cameras can pull back the shadows with less noise, but from what I’ve read that definitely seems to be something that has improved in recent years.

Actually this one that I submitted recently is an example of this, and some of the variations that others posted did indeed look too dark to me on my screen.