Metering accuracy - histogram analysis

Adey

Adey

Messages
734
Name
Adrian
Edit My Images
No
Hello good people!

I have been a bit iffy about my 400d's metering since i've got it, and because of my recent-ish move to full manual, i think its something i need to address.

Ok i've been finding that almost every picture i take is underexposed to some degree. Part of this has been my reliance on the LCD screen as exposure assessment :nono: Plus, what i think is an inaccurate metering system (though not really because its all relative anyway isn't it.

But heres the thing. If i meter off a plain, even surface at a mid-ish tone (filling the whole frame to negate which metering mode im using), and adjust settings so that the needle is right bang in the middle. Should the resulting picture not show a narrow band right in the middle of the histogram? I find that its about 1/5th to the left, or about one stop (i think).

So obviously all i've got to do is do my own calibration, i.e either find something one stop below mid tone, and set it at 12 o'clock or just find a mid tone and stick it at 1 o'clock.

Is this a common thing with the XXXD's?

Still love the little blighter though, amazing how much high ISO noise can be removed by slightly over exposing.

:thumbs:
 
I understand where you're coming from, my 5D2 is exactly the same. But exposure metering is not an exact science at all. It depends.

For example, digital is not like film, and slide film is different to negative film; digital cameras are not always calibrated to 18% grey - 12% or 15% seems more likely, but nobody is saying because digital doesn't have an inherant 'speed' like film, it has sensor gain, which is cloaked in smoke and mirrors; the histogram is not linear; there is no such thing as 'correct' exposure - more like optimum exposure depending on the subject, what you want, whether you're outputting straight to JPEG from the camera or post processing, and so it goes on.

If you want to get the most out of the sensor in terms of exposure, then check out Expose To The Right (of the histogram) technique. And if you do that, then you'll find that you need to add at least one stop to the suggested exposure, possibly two or even more. See here http://www.luminous-landscape.com/tutorials/expose-right.shtml Also try your plain grey test shot and adjust the in-camera contrast setting to max, and then to min - see how the histogram changes!

You best guide to exposure is the histogram, and I find the highlight over-exposure warning invaluable - known as blinkies - in the menu. Good guide to histograms here http://www.sekonic.com/images/files/HistogramsLightmetersWorkTogether.pdf

Ultimately, the meter is just a tool to help you read the light and interpret that in the way which gets the result you want. The camera will tend to err on the side of caution (ie under-exposure so you don't blow anything important) but if it gets you within a stop of optimum straight off the bat, then all things considered it's doing pretty well. The fine tuning is up to you ;)
 
Yeah i see what you mean Hoppy. I've been trying to take more control over exposure, and its meant i've had to really experiment to fully understand how my sensor behaves. But i do feel like i understand better now, i guess each sensor/camera is different and its a case of getting to know your equipment, its much more than just learning what buttons do what.

I've found since exposing to the right, noise is reduced dramatically. I even did a tripod test yesterday because i though my high ISO pics were soft, but they are not, with right exposure they were sharp and relatively noise free. Makes me glad i hacked my 400d to get iso 3200 and spot metering :D

Cheers:thumbs:
 
Last edited:
fwiw, both my canon bodies underexposed in evaluative metering mode (500d and 5d2) i find spot or partial and then locking the exposure on something mid-tone to be pretty accurate and more predictable, but it's not 100% though. it might just be that the metering zones don't quite line up with the focus point and this it want leads to sometimes unexpected results.

i just think that evaluative metering is easily confused and it tries it's best to not clip anything.

i find dialling in a little exposure compensation fixes the problem. i take a couple of test shots, check the histogram and then add as much exposure compensation as necessary. usually 1/3 stop is about right from memory.

edit: although looking through lightroom, i notice some shots in tricky conditions (ie, somebody wearing bright coat) have had +2/3 dialed in, so experiment and see what works for you and the particular conditions, but 1/3 is dialed in frequently on my camera.
 
Last edited:
cuthbert said:
fwiw, both my canon bodies underexposed in evaluative metering mode (500d and 5d2) i find spot or partial and then locking the exposure on something mid-tone to be pretty accurate and more predictable, but it's not 100% though. it might just be that the metering zones don't quite line up with the focus point and this it want leads to sometimes unexpected results.

i just think that evaluative metering is easily confused and it tries it's best to not clip anything.

i find dialling in a little exposure compensation fixes the problem. i take a couple of test shots, check the histogram and then add as much exposure compensation as necessary. usually 1/3 stop is about right from memory.

edit: although looking through lightroom, i notice some shots in tricky conditions (ie, somebody wearing bright coat) have had +2/3 dialed in, so experiment and see what works for you and the particular conditions, but 1/3 is dialed in frequently on my camera.
Click to expand...

Well i think if im using Tv or Av i will compensate the exposure. But when in manual i'll set up till im happy with histogram reading, then find something for reference so i can take quick readings and make quick adjustments as i go.
 
You must log in or register to reply here.
Back
Top