Recently I spent some time reading explanations of dynamic range, on various photography web sites. It was heavy going.
Some seem to be PhD dissertations. They hit you with logarithms, bits, ‘D’ values, f-stops, EV, brightness/contrast ratios, powers of 2, curves, graphs and charts. Next, the biology of the eye, how it adapts to light and dark as our brain assembles an image. Followed by the history of film, with fine points about the ‘density’ of platinum prints. Then on to sensor technology…
I didn’t actually read much of it, because my brain had shut down at “logarithms.” I just wanted a better idea of camera dynamic range, and how it translates to prints sold online via a POD. What’s the DR of a real world scene? How much does a camera capture, how much makes it to a paper print, and where does the rest get lost? What follows is my highly condensed, non-expert and grossly oversimplified DR cheat sheet.
“Dynamic range” can be expressed by logarithmic brightness scales, “density”, EV , f-stops… they’re really just different ways of quantifying the same thing and you can convert one to another. I just think in terms of f-stops.
An outdoor scene on a sunny day might contain a range of 12 or more stops; a rainy day, as little as 3.
The eye doesn’t work like a camera; it adapts to brightness levels moment-by-moment, it’s always moving, and the brain somehow puts it all together. Experts don’t agree on its DR but estimate between 10-14 stop at any one time. So the human eye has evolved to take in the real world.
Ordinary digital cameras can capture 8-12 stops and today the best ones approach 15, but only if you’re shooting raw. Shoot in-camera 8-bit JPGs and it’s down to 8 stops because that’s all 8 bits can hold. (And yes, it’s basically 1 stop per bit because an f-stop is a doubling of light, and a bit is a power of 2…) A 14-bit raw file can potentially contain 14 stops of DR. So if I have a good sensor and get the exposure reasonably right, I should pretty much have it all in one raw image.
A JPG can have a Gamma value associated with it. Gamma is a way to compress dynamic range, by dropping out values where they (supposedly) don’t matter; and if that’s done you have, in a sense, 11 bits of DR. But, not really; you can’t pull in any highlights and shadows in post, and you can’t expand the areas that are compressed. That Gamma curve is baked in.
Film? It really only had 8 stops at its best. It’s complicated. Not going there.
Say have a 14 bit raw file of a good exposure including highlights – can I see it all on my display? No. The best LCD displays today claim 10 stops under “ideal conditions”, whatever that means. People looking at photos online probably have 8 stops on their monitors. (iPhones have 10 on their Super Retina screens.)
With a 14 bit image on a 10 stop display, if the DR of the image really fills those 14 bits I have to either give up seeing some of its highs and lows, or pull in the endpoints of the histogram and compress the image’s DR into fewer stops. In Capture One that means using the Blacks and Whites sliders in the “HDR” tool; detail emerges in the brightest highlights and darkest shadows, and everything in between gets squeezed together.
By doing this I lose some of the original DR, and that’s ok because it has no place to go, if the photo is headed to a web site or an inkjet print. Web images are limited to the 8 stops of DR that an 8-bit JPG can hold. An inkjet print on good paper can display maybe 7 stops from its deepest black to its whitest white. So processing that image inevitably means compressing its original 14 stops of tonal range into 8 – or just letting parts of it go all the way to white or black. Unless, of course, the subject fits into 8 stops or less to begin with.
Photography today is limited by the fact that it’s still an 8 bit world with regard to output media, i.e. LCD displays and paper prints. Displays with higher DR are coming, but it will be a long time before most people have them. I’m not aware of any printing technology breakthrough that could put significantly more dynamic range in a frame on the wall.
The part of 8 stops in JPEG isn’t correct. JPEG data has gamma curve applied which compresses a wider DR into the 8 bits.
With sRGB gamma 2.2 the file can contain 11 stops of DR.
Raw is not an acronym, it should be written ’raw’, because it is an adjective meaning ’uncooked’.
Peter, you’re right that “raw” isn’t an acronym and shouldn’t be capitalized but I think you’re fighting a losing battle on this one – that train left the station long ago. Today if you refer to a “raw file” many readers will blink and wonder what you’re talking about. But, I made the change anyway.
I also added something about gamma. Gamma only works if everything in the view chain – including video card and monitor – is properly set up and calibrated, and that’s hardly ever the case for anyone except a photographer. So it’s true – in a sense – that a perfectly exposed in-camera JPG can contain 11 stops, depending on what the camera did to create it from the raw data. There’s a cost, and you’re almost sure to lose some of that range in processing and display. The illusion of a true 11 bits evaporates if you try to use the Highlights and Shadows sliders in post-processing – there’s nothing there to recover. And by the time someone sees your final image in a web browser, on a POD site, things can be even worse if the wrong gamma is applied by their system. So yes JPEG is more than just an 8 bit format but only if it’s handled with care, and if the gamma curve works with your image.