Overexposure vs Underexposure

[weepete]

I use the blinkies too mate, they can be really useful. I also have a peek at the histogram too mind you

 

[gsgary]

I dont have a screen to look at, but know it will be ok because i know how to use sunny 16 rule

 

[kathyt]

I use my blinkies and histogram. I will go about 1/3 to 1 stop above, but I spot meter with that. I still find myself bumping the exposure in post most of the time. It is very rare that I overexpose SOOC.

 

[gsgary]

I use my blinkies and histogram. I will go about 1/3 to 1 stop above, but I spot meter with that. I still find myself bumping the exposure in post most of the time. It is very rare that I overexpose SOOC.

Get yourself a good handheld meter

 

[SCraig]

...I think of the Highlights vs Histogram much like an analogue clock vs a digital. If I walk past an analogue, I can tell you the time of day straight away. If I walk past a digital clock, I have to read the time, which may or may not tell me what I want to know as I cruise by.

Believe it or not this is why many race cars still have analog instruments. You have to READ a digital readout whereas with analog all you have to do is GLANCE at it to see where the needle is pointed.

 

[kundalini]

It is very rare that I overexpose.

Which has been the detriment to many among the male (and possible female) TPF membership.

 

[weepete]

I dont have a screen to look at, but know it will be ok because i know how to use sunny 16 rule

Yeah mate, but you've left a lot of what you do out of that post too, it's not just knowledge about the sunny 16 rule but how you set up your shots too. Ultimatley its a tool like anything else, feel free not to use it if you wish.

 

[cynicaster]

Thanks Scott. I'm going to have to chew on that a little bit to understand it. I can admit that it is WAY over my head.

If the technical nitty-gritty of analog/digital conversion doesn't sink in, I wouldn't sweat it. Just try it for yourself--expose to the left and right, then adjust each one in post to get roughly the same result, and compare the quality.

It takes very little increasing of exposure in post to coax out the fugly noise in the shadows, but you can decrease it all day, with the only potential consequence being loss of detail in the darker areas if you go too far.

The first time I learned this concept and saw it at play was one of the biggest eureka moments in my hobby thus far.

 

[Gavjenks]

I think the over or under issue depends on a couple things. First off--the "new-sensor" Nikons have very wide dynamic range that they can bridge, and they also have amazing highlight recovery ability in their RAW files. So...with those kinds of cameras (D3-series, D700, D800, D7000 and other new APS-C Exmor-generation sensors) I think in RAW capture mode, the best thing to do is to overexpose a bit, and then "pull" the highlights back down in post processing.

With older cameras, like my Nikon D2x or Canon 5D classic, the dynamic range the sensor can handle is significantly, noticeably much LESS. With those cameras, and with cameras of that older era, blowing highlights was a terrible No-No, and often resulted in images that could not be saved in blown highlight areas.

I've argued with a few people who dismiss this point of view, but I own three d-slrs, which I have used within the last year, to greater or lesser degree: a Nikon D2x, a Canon 5D, and a Nikon D3x. The best way to expose with these three cameras is as described above; over-exposure with the D2x or 5D is often disastrous; with the D3x, the sensor performance is so amazing that overexposure is not an issue. If you own a 5D and think it can handle the same DR as a newer-generation Nikon, you're full of ***+. Period.

If you consistently have to raise the exposure .36 EV or whatever, maybe it would be good to dial in + .3 EV on the exposure compensation, or if your camera allows for it, make an "exposure offset", or whatever your camera maker/model calls it. THis will calibrate, or offset, the meter, so that a 0.0 =/- match needle reading is already biased the .33 EV it needs to be.

A couple of points of comment here:

1) Jpegs only have 8 stops of dynamic range by definition, and recent technology Canon and Nikons (and presumably other brands) all exceed this already by several stops at lower ISOs. So as long as you are shooting RAW, even if one brand is one or two stops better in dynamic range than another, it will only come up as an issue if you over or under-exposed part of your image by MULTIPLE stops and need to compensate later by multiple stops, which in most cases, you shouldn't be doing if you're paying attention at all. Or if you are shooting at very high ISOs, where dynamic range is lower (see point #2 though).

2) Nikons indeed have superior dynamic range at low ISOs that you would normally use in full lighting, like daytime photography or studio work. Dynamic range is not a static value, though, and in fact represents as a curve across ISO values. Other cameras end up having better dynamic range at very high ISOs like 800+ For example, compare the 6D with the D800 on DxO Mark: DxOMark - Compare cameras side by side <---Click on the "measurements" tab, and then go to "dynamic range" to see the full curves. At the lowest ISO, the Nikon is 2 stops ahead, but at high ISOs, the Canon is 1 stop ahead. Thus, the choices you make about how to aim your exposure depend on what model and what ISO you're shooting at. There is no catch-all solution. It will depend on studio versus night shots, brand, etc.

3) Keep in mind that if you want the full advantage of 14 stops of dynamic range on a recent generation full frame Nikon, you have to choose the 14 bit RAW format option. If you do 12 bit, then your dynamic range is 12 EVs max, no matter how good the sensor is.


But rarely does any of this matter, IMO, unless you shoot JPEG alone, or routinely shoot very dark scenes at 6400-12800+ ISO or so. Then you need to worry a lot about this stuff (if JPEG, then you can't adjust much in post, and with high ISOs, your sensor's dynamic range approaches that of a jpeg, so RAW doesn't help a whole lot). If you shoot RAW, though, at low-ish to medium ISOs, then you'll be fine as long as you don't miss your exposure by something like over 2 whole stops in either direction, with just about any recent generation camera.

 

[Derrel]

But rarely does any of this matter,>

Sorry, but that is utter nonsense.Period.

The new-sensor Nikons offer the best performance in term sof dynamic range that the world has ever seen. Canon cannot even put a camera within the Top 10 on DxO Marfk's sensor performance tests.

The actual imaging performance of the D3x versus the Canon 5D is WORLD'S better for the Nikon, with its better sensor. In everyday, real, regulkar, normal photo situations, in low light, when shooting in backlight, or whenever shooting in daylight. Even when both cameras are shooting in 12-bit RAW capture. How do I know??? I actually own the two cameras...

You can theorize alllll you want...but when the camera go into the real world, the superiority of SONY's Exmor sensors is easily apparent.

If you want to see head to head how far Canon has fallen behind SONY in sensor tech and real-world imaging quality, look at Fred Miranda, Canon 5D-III vs Nikon D800 + head to head comparison.

Google it. Canon's best sensor falls flat. IN DAYLIGHT and at its base ISO value. At 100 ISO, the Canon 5D-III sensor has terrible noise in the shadows, and pattern banding!!!!

All your theorizing falls away http://www.fredmiranda.com/5DIII-D800/index_controlled-tests.html

 

[Gavjenks]

Um, I posted the link to results from DxO mark. That's where I got my numbers from... The single score is fairly useless, as it seems to refer merely to the maximum dynamic range at any ISO, not an average or anything that covers the full range of actual use cases. Thus, I suggested looking at the full curves instead.

DxO mark CURVES show that:

D800 vs. 6D (low end full frames):
ISO 100: Comparable models, Nikon is about 2 stops better dynamic range than Canons
ISO 800: About equal.
ISO 6400: Canon is about 1 stop better than Nikon.

D4 vs. 1Dx (high end full frames):
ISO 100-800: Nikon is about 1 stop better than Canon
ISO 1600+: They are the same, or Canon is 1/2 a stop better, or so.

D7000 vs. 7D (high end crop frames. No numbers yet for 70D):
ISO 100: Nikon 2 stops better.
ISO 200: Nikon 1 stop better.
ISO 400+: Nikon plateaus at about 1/2 stop better from here on out.


In all cases, at low ISOs, all cameras are many stops better than a jpeg, meaning it won't matter really if you shoot RAW, unless you majorly blow your exposure or are shooting extremely contrasty scenes at rock bottom ISOs (which is very rare. Backlit sunsets of people is about all I can think of, but only if shot at ISO 100, which I personally probably wouldn't do... Doorways would qualify too, but you could easily just do an HDR for a scene of non moving objects like that anyway, for an effective dynamic range of up to 20+ EV easily with any camera). At low ISOs, the brands get closer to one another, but also closer to the range of a jpeg, making the differences matter a little bit more.

But meh/shrug. Not much practical utility difference either way in most real life situations. And when there is, it depends what you want to shoot. If you do a lot of sunsets at low ISOs, then Nikons are better for you. If you shoot a lot of things at ISO 6400+, then Canons are better for low end full frame (not enough to really make a purchasing decision), or about equal for other types of bodies.

In the vast majority of non-extreme situations in between where a typical person will take 98% of their shots, it simply doesn't matter, because you won't need or really utilize the difference anyway in your final jpeg, since your camera will already be very capable of pulling in extra highlights and shadows no matter what.




I think that similar to megapixels, dynamic range has already been more or less maxxed out in usefulness for most normal situations. Actually useful technological advances in the future are IMO going to be in physically larger sensors for less money, and in faster data transfer speeds (allowing cheaper high FPS and supporting the feasibility of larger sensors).

Also better sensor-based AF. With sufficiently high data transfer speed + better sensor based AF, we could also do away with mirrors and shutters (since the live view would update at near light speed, AF just as well, and the shutter could effectively be electronic only). Solid state cameras won't break nearly as easily (lifetimes of millions of shots rather than 100,000), would be cheaper and easier to manufacture, weigh less, and have much cheaper and higher quality wide angle lenses, without any of the relative disadvantages that mirrorless cameras currently have versus DSLRs.

And possibly exotic new technologies like fresnel lenses, multi sensor arrays, light field cameras, etc.

 

[Johndow]

In response to the OP's original post, my default tends to be +1/3 over exposed.

The majority of my photos are of my family, 95% of which would be outside in good (bright) weather. I find that my camera will under expose on bright days. This wouldn't be such an issue if i were just taking landscapes, but faces tend to always be underexposed. Even at +1/3 i sometimes have to increase in PP by 1/3 if i had the camera pointing up and took in a lot of sky.

However, i have found that with LR4, i don't have to increase the shadows, but may have to reduce the highlights sometimes. In high ISO photos it helps to overexpose, because recovering detail from an underexposed photo produces awful noise.

If i was indoors or in low light i would leave the compensation at +1/3 unless I really needed the extra shutter speed to counter blur.

As my subjects are my family and any sky is just background, large dynamic range is not a requirement. If the sky is well exposed it's a bonus, but if it's blown out it doesn't matter, as long as the exposure on the main subject is spot on.

 

[RacePhoto]

Depends on what you are shooting. If light subjects, you may want to under expose, intentionally. If normal or dark, I have the camera set EV to 1/3rd over All The Time! I'd prefer to lose some highlights and not have confetti and spots in the shadows. It's that simple.

When you have to raise the exposure, (in processing) you raise the noise. When you lower it, you lower the noise.