Know your gear- Digital Camera Basics.

[sk66]

To get the absolute most from your gear and to "properly apply" the exposure triangle you need to know two basic things...Your camera's base ISO (easy) and your lens's sharpest aperture (not as easy, but not hard).

Base ISO...Easy, it's the lowest ISO setting that's "native"(i.e. a number, not "Low-1" or some such)
Why is this important? Because that is the ISO the camera is designed around and optimized for. At every other ISO setting you are giving up some performance capability.
I'll use my D7000 as an example: (higher readings/numbers are better)

Dynamic range(DR): at ISO100 =14 (f)stops, at ISO1600 = 9.5 stops. That's almost five stops of loss, ~ 30%. With less DR capability highlights clip earlier and shadow detail is lost earlier.

Signal Strength over Noise (SNR): at ISO100 = 41db, at ISO1600 = 28db. That's more than 25% loss in signal strength. With lower SNR noise becomes more visible in the image.

Tonal Range: at ISO100 = 9 bits, at ISO1600 = 7 bits. Not bad, only ~2% loss

Color Sensitivity: at ISO100 = 24 bits, at ISO 1600 = 17 bits. Almost 25% loss.

You don't have to know all of the specifics, I don't. I used DXO lab test results to get the information. The only thing you have to know is that if you are not using the camera's base ISO you are giving up performance...quite a lot in some instances. The further away from base ISO, the more you are giving up.



Now onto the lens's sharpest aperture. As a general rule f/8 will work for most lenses, but it is not necessarily the sharpest for your particular lens. And by "sharpest" I mean the lowest aperture where maximum image clarity/detail/contrast/color is achieved.

I'll use some examples (Nikon Lenses)

My 24-70 and 70-200 f/2.8 lenses: best aperture ~f/4. These are very expensive "pro" lenses and they still are not at their "optimum" until stopped down a bit.

My 28-300 f/3.5-5.6: ~ f/8. This is also an "expensive" lens but for a different reason. Still, if used at f/8 I can get images comparable to the 70-200 f/2.8 which cost more than twice as much money.

The 18-200 f/3.5-5.6 I used to own: this one was a bit trickier as it's best aperture varied with the zoom setting, it was f/8 at min/max zoom and f/11 in the middle.

And now the "fabulous Nifty Fifty", 50mm f/1.4: it's best aperture is approximately f/5.6!

So why is it important to know the "best aperture" for your lens? Because at any other setting you are giving up something, either image quality or light. As you set the lens to a higher aperture number you are reducing the amount of light the lens can gather, and if you go far enough (generally above f/11) you will start degrading image quality due to diffraction. As you set the aperture lower you are giving up image sharpness/contrast/clarity/color. The losses generally begin in the corners/edges of the image and move inwards.



So, if you set your camera to base ISO and best aperture and whatever shutter speed is required for the exposure you will get the absolute highest quality image possible from your gear.... It's where the "f/8 and be there" rule of thumb came from, because f/8 "works" as one of the "best apertures" for almost every lens.

Now, you may not always be able to use the absolute best settings....maybe you need more SS. Well now you can decide "what" to trade first and why...Maybe there isn't a large DR in the image...easy, give up a couple stops of ISO. Maybe that's not enough....OK, maybe now I need to "compromise" because if I go any higher on the ISO the image will be "unusable" noisy. Easy enough, open up the aperture. Maybe the compromises are too much and this is the type of situation I'm in all the time for the type of photography I want to do....NOW it's time to buy new gear.

Sometimes you might not WANT to use the absolute best settings... Maybe I don't want to be able to see every pore in the models skin...easy enough, open up the aperture. And maybe for artistic reasons I care more about the "bokeh" in the background than I do the sharpness of my subject....well, I don't really understand this one but ok, open up the aperture more. Realistically, it becomes a balancing act. How much bokeh can I gain for the sharpness lost.....should you go to f/1.4 with that 50mm, probably not.

There's a lot more that can come into making the choices and as you learn more you will be able to make the choices more easily/intuitively. But it really is as easy as base ISO/best aperture/ matching SS. If that won't work or isn't what you want, what do you want to trade off first...Image quality in noise/color/DR then trade ISO. If it's image sharpness (again usually starting in the corners of an image) then trade off aperture.

Sometimes you can regain much of what is lost in post, but it will probably never be as good as if it were captured "correctly" to start with.

[BuddhaPi]

Thanx sk66! this was simple and easy to understand, even for a dunce such as myself. I'm still kinda waiting for the light bulb to go off in my little head, where it all makes sense, this was super helpful!

[RLucas]

Good read Steve! Thanks for taking the time to write this out.

[JFSanders]

Well done Grumpy. One small nitpick is that while a 2% loss seems trivial with regard to tonal range I suspect that the effect is more logarithmic than linear. And as such it is more noticeable to the viewer. Although they may not be able to quantify it.

Jim

[sk66]

Thanks!


Jim,
Actually the tonal range spec. threw me....I don't exactly know what it means. I would have expected it to be a part of DR or Color sensitivity. It's obviously related and apparently the least significant loss, but *I* couldn't quantify it on it's own. I do know what tonal range in an image is....

I did a bit of quick research and couldn't find it defined as a separate sensor measurement...But I'm guessing it might be the "gradation steps" sensed w/in one level of DR/color

As I said, I didn't/don't know all of the specs for my sensor....I don't really care, it's the one I've got to work with at the moment. All I care about is how to best utilize it.

[JFSanders]

In my research it came down to how the sensor's data is interpreted by the rendering engine in the camera. But in general the higher the ISO (photosite sensitivity algorithm) the harder it is for the camera to differentiate between the color information created by the Bayer array. And to decrease banding, random and hot spot noise the bit level is dropped to 7 from nine. This leads to a decrease in rendering shadows as they actually are.

Jim

[Tito87]

Thanks for doing it. It was a helpful read.

[sk66]

You mean the bit losses in color range and tonality for every ISO increase are actually programmed in? Wow, I would have thought it was simply a side effect.

I think the noticeable effect for the viewer would be primarily influenced by
1. having seen the original scene
2. having given up something I shouldn't have.

i.e. if there isn't a large DR, or variety of vibrant colors in the scene, then I can trade ISO all the way to the signal noise floor with very little noticeable effect.

[JFSanders]

Basically it comes down to how much computer you can put into a camera. That is why programs such as noise ninja etc. can make improvements in images. They use algorithms tailored to the scene and data sets.

As for what the viewer perceives to be a true image, well that gets complicated quickly. Suffice it to say that a young person with little experience is more likely to believe than a person with more life experience.

i.e. if there isn't a large DR, or variety of vibrant colors in the scene, then I can trade ISO all the way to the signal noise floor with very little noticeable effect.

Making the wrong trade off will be readily apparent to some people but those folks will have a trained eye as opposed to the average viewer as you know.

Jim

[ravncat]

I agree whole heartedly - learn your camera, know how the sensor responds and you're absolutely giving up more and more image quality the higher you push your iso from base.

I just have a couple small side notes;

1. An interesting tidbit The Native Iso of a sensor often has slightly better performance than the Base iso (or sometimes referred to as Base, or true base iso. This value varies from sensor to sensor but is often not the lowest non boosted value. It's often quite close to the base iso. The difference is small enough that Base ISO is simplest - if you're curious about your camera - some web searching will find it. For example the Nikon D300 has a native iso somewhere around (i.e. +/-5) 160.

Here's one thread that talks a little about it Camera Labs :: View topic - Know your Base (or Native) ISO Iso gets tricky when we see how flexible the standard is. Native Iso should be where the dynamic range/color depth etc is the greatest.


2. Dxo's Method of measuring dynamic range uses a very engineering definition of dynamic range. To quote Luminous Landscape "The standard definition is, as mentioned above, how many F stops above and below middle gray can be recorded while delivering full texture and detail. The DXO definition, according to their web site, is the range between zero signal to noise and full saturation of the sensor."

I think doesn't represent what photographers are finding/seeing as a usable dynamic range for their cameras. Dx0 always reports a larger dynamic range than I have seen with any of my cameras at one time.So, for example I get about 9-10 stops of dynamic range with my d700 at base iso, DxO reports 12.2. My figure is in the ballpark of many other photographers. I'd recommend taking photographs of a scene and adjusting your exposure and finding how many stops above and below middle grey can be recorded with full texture and detail.

3. 9bits vs 7bits may be a 2%difference in bits, but the data values produced are at a much greater than 2%difference. 7bit vs 9bit is 128 vs 512 values of data. which is quite significant.