Ever wonder why cameras with the same ISO can look so different?
Yeah, me too. There’s no easy answer as to why.
When I started out as a young camera assistant I remember working on a low budget feature film that shot on Fuji film stock to save money. (Fuji was popular partially due to its look, which tended to be much more neutral in color compared to Kodak warmth, but was mostly because it was less expensive.) The high speed stock, 8514, was considered almost insanely fast at ASA 500. When viewed on the big screen, however, it was a snowstorm of grain. Because of this most DPs never rated it faster than ASA 320.
At the time I remember thinking, “Why is this ASA 500 film stock so much grainier than Kodak’s ASA 400 film stock? It looks like Kodak’s stock pushed by a stop or two. How can their ASA ratings be so close?” As I worked with more DPs I learned that they almost never rated film stocks at the manufacturer’s suggested rating. Most of them overexposed film stocks a little, particularly high speed stocks, and printed them down for richer blacks and more saturated colors, while some underexposed and printed up to reduce color saturation and contrast.
Eventually Kodak dumped the ASA standard completely and created an EI rating, which stood for “exposure index.” This was their way of saying, “We don’t care about what number we get from the ASA rating method. We think our stocks look better at a different number. Set your meter to this instead.”
Such honesty from a manufacturer was refreshing.
>Over time, ASA (American Standards Association) transitioned to ISO (International Standards Organization), and the same thing that happened to ASA has now happened to ISO: It has been “genericized.” While ISO ratings are based on some sort of testing, there’s clearly enough room for error that manufacturers can skew the numbers however they like. The ISO trademark is now to film speed as the word Kleenex is to facial tissue in the U.S.: it no longer refers to a specific process or product, but represents an generic concept.
While the technical requirements that determine a film’s speed is different between ASA and ISO, and the ISO standard for determining a digital camera’s speed is different still, the goal is always the same: if you set your light meter to a specific value you should get predictable exposure results out of a camera. Specifically, the aperture and shutter information given by the meter should result in an 18% gray card placed in front of the camera appearing 18% gray on a calibrated Rec 709 monitor. How that happens varies. There are standards for determining ISO settings for still photography (and, if you want to pay for the privilege of reading them you can do so here but it’s clear that they don’t mean much).
The evidence for this is built into every camera in use today.
Each one has a “native” ISO determined by the manufacturer. It also has settings for numerous other ISOs, both higher and lower than the “native” ISO. If ISO is a standard, and every sensor can be measured in such a way that an objective ISO speed rating can be assigned… why are the other speed options labeled as ISO as well? Shouldn’t every digital camera have one objective ISO rating and a variety of exposure index (EI) options? After all, exposure index simply says, “Set your meter this way and you’ll get predictable results.” It doesn’t say anything about the actual noise characteristics of the camera, which is the limiting factor in preserving highlight latitude and determining base sensitivity. That’s supposed to be the domain of the ISO rating… but clearly it’s not.
ISO seems to have become a recommended value, not an objective measurement.
Where does this leave us? As always, when it comes to new camera equipment… suspicious.
It really is fascinating to see how different cameras can show such different noise characteristics at the same ISO rating. I find almost all cameras to be just a little too noisy for my taste at their defaults, but I’m probably pickier than most people: I shoot a lot of visual effects, which require very clean images for compositing and manipulation, and I also shoot a lot for the web, where less noise means better compression and more shadow detail. (Codecs throw away a lot of information in shadows, and if the shadows are moving due to noise then the codec is going to throw away a LOT more detail. The cleaner the shadows are the more detail will be retained.) My rule of thumb for visual effects is to divide the camera ISO by 2 for best results, and often I’ll do the same for my regular work. Many of the lesser codecs used in production (Sony’s XAVC comes to mind) are tremendously useful for many things, but noise does stress them a bit—probably less so with intra-frame codecs (where all compression is done within a single frame) but definitely more so with inter-frame codecs (where compression occurs across a sequence of frames).
In my next article I’ll touch on how I go about determining the proper ISO (really EI) of a camera. For now, just know that ISO values don’t seem to have any objective real world meaning. They are recommendations that are driven partially by technology but largely by marketing. I would be much happier if all newly-released cameras sported EI values instead of ISO: ISO is supposed to say “This is an objective speed rating for a camera” but clearly doesn’t, while EI says “This is where we think our camera looks best,” which is at least more honest.