I am embarrassed to ask this question because it is so elementary. But since I don’t really know the answer …
How does a modern camera, and in particular, a mirrorless camera, determine the appropriate exposure on an nonmanual mode. Chatty Bing tells me the camera has a built in light meter separate from the sensor. Is this correct or does the camera make use of the signal from the sensor? In the early days of digital I “learned” that exposure was based on the jpeg. Well it is certainly true that any after the fact judgement of the exposure is based on the jpeg. But what about before the fact?
And perhaps a related question. Does the scene mode chosen – natural, muted, vivid, monotone, etc. – in anyway effect the camera’s choice of exposure?
DSLRs have those, but all mirrorless (AFAIK) just use various algorithms on the image feed, the same data you’re seeing on the live view while you take the photo.
Not sure… some cameras do do various tricks to get more dynamic range, such as underexposing the sensor, then in the JPEG engine increasing the brightness of the shadows, but leaving the highlights.
FWIW, I think of those as color modes, or jpeg profiles… I don’t think so, but stand to be corrected.
Hope this helps… it’s actually quite an interesting subject once you start really looking into it.
Cameras essentially check the average brightness in the live-view video feed, and accordingly adjust the aperture, shutter-speed, and ISO. Of course, it’s probably a bit more intricate than just an average. There’s an adjustable center-weighting to consider, and perhaps it tries to avoid blowing out too many pixels.
DSLRs use a dedicated sensor, whereas mirrorless use the main imaging sensor.
Then, as @bastibe says, it adjusts the exposure so that the average luminance is equal to 18% grey. Exactly how that average is calculated depends on the chosen metering mode. Some cameras might take the JPEG profile into account, but I don’t think that’s common, if it happens at all. However, any live histogram is almost always going to be based on the JPEG.
Thank you all for your responses; they have helped me clarify my thinking. I still have some uncertainty with respect to some of the subtleties but I’ll set those aside for the moment.
I did a quick and dirty experiment to test whether the scene mode (jpeg mode) in any way affects things. I mounted my camera on a tripod and shot a scene five times, each time using a different scene mode. And indeed the camera chose the same exposure (iso – I shot in manual with auto iso) for all five modes. So it seems that the scene mode does does not affect the measured exposure – at least within a 1/3 stop precision. However when I opened the 5 raw files with RawTherapee and applied a neutral profile to each frame, I am noticing there are subtle differences in the various histograms. (You can notice the differences using Fast Raw Viewer as well). So now I don’t know whether my experiment was just insufficiently controlled or whether the scene mode actually does affect the raw data. My plan is to do a more controlled experiment, but I won’t get to it immediately.
Sometimes the cameras have a high-dynamic-range mode, which does change the physical exposure. Fuji calls it DR, Nikon calls it Active-D, other manufacturers have their own names.
Using the main image sensor doesn’t mean ‘using the jpeg’ and I would be very surprised if that happens on any ‘respectable’ camera.
The main sensor are basically loads of small light sensors , that give measurements . Those measurements are what are stored in your raw file. Basically :).
Creating an image from those measurements comes after that, so I think they use the main camera sensor just fine.
But… I’m just guessing what seems logical to me, and I’ve been surprised by that before :).
While this is true, most cameras apply their “jpeg style” in real time on the preview showed on the screen/ovf, if configured of course. Just being pedantic, as at least with fuji, the jpeg style chosen has no effect on the metering of course.
All of the reported iso were the same. However I also wonder if – and based on zero evidence whatsoever – perhaps the iso varies continuously and it is only the reporting that varies 1/3 stop increments (or whatever increment you have set in your preferences).
I agree that the main sensor is just “basically loads of small light sensors”. So that certainly the exposure can be set based on the output of those sensors. (How the output of all those sensors are averaged together is determined by the metering mode.) But in principle, it could also be determined based on the jpeg – expose the sensor / compute a jpeg / determine whether the jpeg is over or under exposed according to the metering mode / increase or decrease exposure / repeat. Providing you have the computational horsepower, such a feedback loop should be quite capable of setting the desired exposure.
Furthermore, the display is showing a continuous display that at least mimics the jpeg. (This is also necessary to compute the real time histogram). So the camera has to do a lot of real time picture processing anyway. (It can probably skip the data compression that producing a jpeg entails.) So the question becomes does the camera make judgements about exposure before or after the image processing? I think in principle it could be either.
If the exposure determination is made before the image processing, the raw output should be independent of the scene mode chosen. If the exposure determination is made after the image processing, then perhaps it might affect the raw output.
The other possibility would be minuscule changes in the lighting of the scene. Haven’t seen your shots, if they’re indoors with constant light then I’m out of ideas…
