So until now I’ve been thinking of dynamic range recovery as being a combination of highlight recovery and shadow recovery. But actually, I’m starting to think that if my highlights are blown, that’s maybe not a dynamic range issue, but an exposure error on my part. I’ll explain what I mean.
Anyone who’s spent any time recovering tonal range in raw files will know that the extra highlight detail available is restricted. I reckon with modern sensors you might be able to get back around 1.5EV of highlight detail, and I really don’t see much difference between all the cameras I’ve tested, reviewed and owned. Fujifilm’s extended dynamic range modes work differently, but that’s another story.
That’s still enough to pull back blue skies and bright clouds in most outdoor shots, though maybe not brightly-lit scenes viewed through windows from indoors, say, or backlit subjects. So here perhaps we’re talking about an exposure issue, which is our job, not the sensor’s.
Shadow recovery from raw files is different. Here, you can carry out huge adjustments and still get good image quality – though this does show up the difference between sensors and cameras. My Lumix G9 (Micro Four Thirds) produced a horrid magenta tint in the very deepest shadow tones, while my OM-D E-M1 III produced neutral shadow tones but more noise and less detail if you boosted them too far. My Canon EOS R8 raw files offer decent shadow recovery, while my Nikon D800 and D610, old as they are, offer quite exceptional shadow recovery.
It’s not just the sensor size. There is a broad narrative that the smaller sensors of MFT cameras cannot match the dynamic range of larger sensors, also because they capture 12-bit raw files not 14-bit. That’s not been my experience, though, both with my own images and the time I’ve spent lab-testing cameras in controlled conditions, but that’s another story.
Incidentally, my full frame Canon EOS RP had the worst shadow detail I’ve seen from any camera, with blotchy color artefacts and an almost complete loss of detail. It’s not just about the sensor size.
Are blown highlights a metering issue not a dynamic range issue?
I’m starting to think of it this way. If the highlights in a scene are important and they’ve been lost, then perhaps it’s my fault for not checking the histogram to make sure they’re in range. Digital sensors aren’t good with highlights, but we know that and it’s been that way for years, so I should by now have figured out how to adapt my exposure technique, and I guess it’s my lookout if I haven’t.
Is it easier to think of dynamic range simply as shadow recovery?
Maybe so. I’ve always thought that a rather narrow idea, since it seems to me that blown highlights are also a dynamic range issue, and a serious one.
But if I think of highlights as an exposure issue instead, that leaves shadow recovery as the way to restore brightness in other areas of the picture and a measure of the dynamic range available.
I just need to remember to expose for the highlights, and maybe allow around 1.5EV maximum leeway when shooting raw. So far so good.
Testing is not the real world
Testing is designed to find the limits of performance, not necessarily to reflect real-world situations. It’s very easy concoct a test that reveals a technical weakness in a camera or sensor, but not so easy to find real-world shooting situations where that applies. I regularly see dynamic range tests that increase exposure by 4EV or even 5EV to see how the shadow detail responds. And yet after looking through dozens of high-contrast raw files to find an image for this article that needed this kind of correction, I couldn’t find any. There were a few that benefited from a +3EV adjustment for the shadows, but that’s all.
With a 3EV adjustment, my MFT images were a little noisier than those from my full frame cameras but still quite usable. If you gave me the choice I would certainly choose a camera with 12EV of dynamic range over one with 10EV.
(But we’ve all heard of exposure bracketing, right? Lightroom and Capture One are excellent at merging bracketed shots into high-quality, natural looking DNG files. You don’t have to do everything with a single raw file.)
Whether I would ever shoot any scene that needed the extra dynamic range is a different question. Besides, there are other issues with turning high brightness ranges into effective images.
You can get all the tones but none of the contrast or ‘sparkle’
This is a problem for high dynamic range photography in general. You need a lot of tonal compression to squash a wide brightness range into the contrast range of a print or the average monitor. Some software will apply tone mapping to get a punchier result – Affinity Photo’s Tone Mapping Persona is excellent at this. What you don’t want is the oversaturated, localized contrast HDR ‘looks’ that were all the rage a few years back.
Extended dynamic range for local adjustments – the real advantage of raw
Most raw processors will offer highlight and shadow recovery tools to selectively enhance very bright and dark regions and still produce a reasonably contrasty and punchy image. This is a kind of automated local adjustment.
But if you do it yourself, you can really exploit the extended dynamic range of raw files, and this is where shadow recovery, and the quality of the recovered areas, becomes paramount. You wouldn’t increase the exposure of a whole image by +3EV, but you might do it for objects or areas in very deep shadow.
And it’s true that in these situations the superior dynamic range of my Nikon D610, say, will produce much better results than my OM System cameras. But the adjustment has to be pretty extreme for the difference to become obvious and the fact is that doesn’t happen all that often.
Even here, though, the greedier you get with your tonal adjustments, the harder it is to avoid artefacts and unnatural tonal transitions. Lightroom’s AI masking is brilliant, but produces edge artefacts that soon show up as bright outlines if you push your luck too far.
So what am I trying to say with all of this?
- I think it might be an error to treat blown highlights as a dynamic range problem when perhaps it’s actually an exposure problem
- Shadow recovery is perhaps the best way to judge dynamic range
- More dynamic range will always seem better, though the number of times you need 12EV rather than 10EV (for example) may be limited
- Lab tests will reveal weaknesses in performance but often with exposure adjustments you would never apply in real life
- Capturing a high dynamic range is one thing, converting it into an effective image is quite another
