In the past couple of weeks and months, I’ve repeatedly been asked about “linear camera profiles” because some (more or less influential) websites and photographers announced it as the “new best way” to develop your raw files — promising much improved quality, better highlight recovery, claiming that they’re much easier to work with, sometimes combined with damning Adobe for “chopping off” data, and whatnot.

If you’ve seen similar videos and articles and are wondering whether it’s really worth it, the best is of course to experiment. Which is what I did, and in this article, I’d like to offer a counterpoint from someone who’s been working with single exposures and developing photos in Lightroom for nearly 15 years now.

TL;DR: you do not need a linear profile for better highlight recovery, better image quality, or whatever else. The linear profile is simply a different starting point. I do not find them easier or more logical to work with — on the contrary. More about that below.

Normal Profiles

Without going into too much detail: every “normal” camera profile gives the mid-tones in a raw file quite a boost, via a tone curve. That’s not Adobe-specific: all raw converters work this way, and so does any camera’s JPEG engine. This mid-tone boost leads to a compression of the highlights and a lift of the shadows — both are usually rather desirable! Here’s a screenshot from the Adobe DNG Profile Editor that illustrates this nicely:

The red curve is the “normal” tone curve, in this case of the (old) “Adobe Standard” profile, where an input of 0 equals an output 0 (pure black). An input of 128 (middle grey, in the center) is boosted to an output of ~208. Also clearly visible in the screenshot is that an input of 256 (pure white) equals an output of 256 in the Adobe Standard profile’s curve: at the top right, the two curves merge again.

Now when you read “highlight compression” remember that this is an interpretation of the raw data. No data is lost in this process, all the highlight information is still there, in the raw file (this would not be the case if you’d capture JPEG, straight out of the camera).

Better Highlight Recovery?!

Naturally, the claims for better highlight recovery with linear profiles piqued my curiosity — I mean, just looking at the above screenshot, how could it be? It’s just a curve that interprets the raw data. To find out, I created linear profiles, both from the “Adobe Standard” profile, and from the “camera matching” profiles that I prefer (see below), for a comparison.

The expected result, of course: there is NO difference in recoverable highlights between using “normal” and linear profiles. When you have blown out highlights from overexposure in a photo, you can’t bring them back. A normal profile is absolutely NOT “chopping off” highlights and leading to data loss (as one video that I saw claimed).

Here’s the proof, in a 200% crop. This is the sun, at the edge of the Panamint range in Death Valley, photographed from Badwater Basin:

On the left, you see the image using the Adobe Standard profile. On the right, the same image, using the same profile with a linear tone curve. Since the linear curve doesn’t compress the highlights, it is overall darker already. For that reason, the Adobe Standard rendition has its Exposure in LR set to -4.0 and the linear profile only to -3.0 — but you can see that the area with complete loss of highlights (no detail & no color information) is identical.

So, as stated above: linear profiles do NOT allow more highlight recovery. At all. What’s gone on the sensor level is gone and no profile will ever change that (this is not a surprise). It’s not possible to change physics with a camera profile — and for the same reason, linear profiles don’t allow more shadow recovery either.

Why Are Profiles That Way?

Now, what Adobe and camera makers (when you use JPEG) achieve with this “mid boost” is actually a couple of things: first, our human vision isn’t linear, so this is a gamma correction. It also “sort of” emulates film, and tries to mimic film’s more gentle roll-off into over-exposure. This and more is thoroughly explained in the white paper “Raw Capture, Linear Gamma, and Exposure” by Adobe’s own Bruce Fraser.

Let me digress for a moment: where film and digital fundamentally differ is where the reserves lie as this really good video shows. Film has more latitude for highlight recovery — a lot more: in the linked video, the author concludes that the recoverable exposure range of film is from -2 to +10 stops (meaning that essentially, you can’t really overexpose film). Digital sensors on the other hand have more latitude for shadow recovery (usable range -6 to +2 stops).

So while you should “expose to the right” with digital sensors to maximize the signal-to-noise ratio, in high-contrast situations it is always better to underexpose if you need to preserve critical highlight detail. If you’re uncertain, exposure bracketing is your best friend! As shown above, what’s lost to over-exposure is truly lost — but you can pull the shadow detail up a lot, at the cost of a little more noise.

Linear Profiles

A linear camera profile on the other hand does not boost the mid-tones. The highlights are undeniably better defined — because the curve doesn’t compress them! This may indeed be very desirable if you’re using exposure blending, in particular with luminosity (luminance) masks.

But uncompressed highlights mean that they are overly pronounced, at the expensive of the mid tones. The rest of the image looks way too dark, and doesn’t match how we perceive the world, with our human eyes. But, even if we leave human vision aside and focus purely on data, it makes getting a pleasing image out of a single exposure much more of a struggle than in has to be, and in particular if it is one of a high contrast situation (which is where digital photographers find using raw data most useful).

Using the “Auto” feature in Lightroom* with a linear profile led to severe loss of highlight detail in some of the photos that I tested (example; a 200% crop from this image). Manual intervention was badly needed. Clearly, this algorithm is fine-tuned for normal, and not linear profiles.

If linear profiles are a struggle to work with for single exposures but have better highlight definition, the question inevitable arises: how do you “un-compress” the highlights of a standard curve, when using a normal profile? That can be done either with an exposure adjustment (and thus, pulling the highlights out of the most compressed range), or with the “Highlights” adjustment in Lightroom: pull the slider to the left and you’ll get better highlight separation. This is called tone-mapping and I think that, combined with local adjustments, it’s a faster and easier path to getting pleasing images.

A Better Starting Point

It’s completely legit if some people like a purely linear interpretation as a starting point, of course, but I would rather call it a matter of personal preference. When using exposure blending with luminance masks, exposure fusion, and high dynamic range imaging, linear profiles can be very useful, because these processes benefit from a clearer highlight definition in a “component” image (in other words, you do not need to prepare your “dark” exposure for the best possible highlight representation).

My critique goes towards the unreflected praise that these profiles received in some articles and videos. As a friend of single exposures, I find it awfully difficult to get a rich, lively rendition of an image when I start with a linear profile — because the most critical part, the mid-tones, are so poorly rendered when using a linear profile. Trying to bring them into a better range (which is exactly what normal curves do!) leads to an intense struggle with the over-pronounced highlights.

I do appreciate that this discussion draws attention to different camera profiles: my starting point has long been a “camera matching” profile in Lightroom, namely “Camera Neutral”, because it gives me a flatter initial rendition of the image and also, the color interpretation of the “camera matching” profiles in general is much more accurate for my Nikon camera. I calibrated my monitor for accurate colors, it makes sense to use profiles that better match the specific color rendition of my camera.

If you’re “fighting the sliders” in Lightroom or Camera Raw, it may simply be because Adobe’s current default profile, “Adobe Color”, already has much more contrast and saturation than what you’d like to begin with. If that’s the case, then I would suggest to switch to “Adobe Neutral” or, depending on how good this renders colors from you sensor, “Camera Neutral” as your starting point, instead of using a linear profile. I’m getting much more usable results that way, faster, and without any of the other shortcomings of using a linear profile.

*) I do not use the “Auto” feature in Lightroom in general — it does the weirdest things to my photos! — but I read this recommendation in an article about linear profiles by Tony Kuypers.