[luftfluss]

As I've been fumbling along with a small fit of GAS, one of the driving forces for a search for an upgrade to my K-30 has been "better high ISO performance". I find that on the K-30 ISO100 to 800 to be great, 1600 to be perfectly usable for smaller prints, and 3200 acceptable (color) to pretty good (black and white).

I've contemplated such far reaching options as a 645D and a Leica M9 that top out at 1600 and 2500 respectively. So I finally decided to search my Lightroom catalog and see what I've actually been shooting to see if those systems would even be viable for me as I shoot now, ignoring a desire for higher ISO pereformance for a moment.

I have approximately 10,000 shots in my Lightroom catalog currently. Approximately 86% are from 100-800, 10% are 800-1600, and the remainder are 3200+. Going off my starred shots, the ratios remain the same but the overall shot total drops to about 1100. My fastest lenses have only been f/2.8, so there's room in my life for a couple fast primes maybe... most of the ISO1600+ shots were with my DA*50-135 so how useful a fast 50 or 30-ish would be is up for debate.

Now, I freely admit to using auto ISO and capping it at 1600 for most work, so the data isn't perfect. But it does raise a little bit of doubt in my mind: do I really need more?

My challenge to you: go through your catalog. What ISO's are you using the most on photos you actually want to keep? Knowing that, what would a better high-ISO camera realistically gain you over what you have now?

I don't know by percentage, but I do a lot of critter shooting, which frequently necessitates fast shutter speeds & ISO 800-3200. Typically, my K-5 II is set to Auto ISO 1600 and I'll push in PP when necessary.

A couple observations:

1. I see many images of critters that are soft because the shooter is unwilling to boost the ISO and thus uses too slow of a shutter speed. Ironically, sometimes this soft image actually has more obtrusive noise than a similar image shot at higher ISO / faster shutter speed because the latter photo is sharper and has more evident detail.

2. Many folks don't have a great handle on noise reduction and sharpening. Sharpening increases noise, and some folks sharpen the sh*t out of their photos, especially when there is subject movement. Over-sharpening + bunches of noise = pixely mess.

[Rupert]

I think there are so many factors involved in a shot that there is no "golden rule" to follow regarding ISO values. Some high ISO shots turn out great...some don't.....so many factors other than just the ISO setting determine the results.

Generally, with my K1 I get the best results if I stay at or below ISO 800. This is pretty easy if you can hold steady and let SR work for you at low shutter speeds.....and your subject cooperates! Color holds very well at ISO 800, but deteriorates some as you go higher.

1/160 handheld f6.3 ISO 800 @ 500mm


1/125 handheld f6.3 ISO 800 @ 500m


Depending on he subject, the light available and a lot of other factors ISO 12800 can get you a decent shot if needed.

1/80 handheld f7.1 ISO 12800 @ 240mm


I'm with Norm.....if you want the shot use what you have to use to get it. A missed shot has zero value.

Regards!

[AquaDome]

With the K-3, I stick to 100, 400, 1600. Mostly 100.

[E-man]

The concept of ISO as a fluid and adjustable factor is one I'm still wrapping my mind around as I embrace the digital medium. I'm a film guy from way back and ISO was always something dictated by the film in the camera. I did most of my film shooting at ISO 400 because I used a lot of Tri-X in my newspaper career. I need to learn to take more control of ISO (and exposure in general) on my digital cameras so I can achieve better results.

[BigMackCam]

... But my rule of thumb for ISO is always, use the lowest ISO you can. That is true in every image. Should you put a cap on that? DO you want the image or don't you? You do what you have to do to get the image, and then deal with the limitations of higher ISO later. If I'm ever in a situation that requires 128,000 ISO, I'll certainly use it. It will be better than nothing....

As I said, do you want the image or don't you?

^^^ This. Assuming we've used the lowest shutter speed and widest aperture possible to get the intended creative effect, we're stuck with the ISO we're given (or we push in post, which in turn brings up the noise). The ISO - to coin an over-used phrase - "is what it is"

I will say this, though... I have a number of photos where I've used a higher shutter speed than necessary - either because that's what I was shooting at moments before, or because I didn't fully understand the required shutter speed to freeze what I was shooting. So, I ended up with ISO 800 or 1600 shots instead of maybe ISO 400. But the performance of modern sensors is so good that I still got very decent photos. This is a good and bad thing; on the one hand, I got some great photos... on the other, I can't help but feel I've been a little lazy at times with managing my shutter speed setting.

[Edgar_in_Indy]

1. I see many images of critters that are soft because the shooter is unwilling to boost the ISO and thus uses too slow of a shutter speed. Ironically, sometimes this soft image actually has more obtrusive noise than a similar image shot at higher ISO / faster shutter speed because the latter photo is sharper and has more evident detail.

This is a very good point, and it was a lessen I had to learn the hard way, over time, as I started shooting at my daughters' dance events.

I've had too many photos that would have been GREAT because the timing, composition, and lighting worked out perfect, except that I was using too low of a shutter speed, so the subject was soft or blurred. It's heartbreaking whenever that happens, so now I don't hesitate to crank the ISO up in those situations. As you said, the noise is less of a problem than your subject being soft or blurred.

[luftfluss]

This is a very good point, and it was a lessen I had to learn the hard way, over time, as I started shooting at my daughters' dance events.

I've had too many photos that would have been GREAT because the timing, composition, and lighting worked out perfect, except that I was using too low of a shutter speed, so the subject was soft or blurred. It's heartbreaking whenever that happens, so now I don't hesitate to crank the ISO up in those situations. As you said, the noise is less of a problem than your subject being soft or blurred.

It's funny how long it takes some of us to learn this lesson. And sometimes I still find myself trying to push the envelope with slow shutter speeds in poor lighting. But I'm not as bad as I once was.

A couple weeks ago I saw a garter snake poking his little head out of the grass to take in the rain. I immediately fired off a burst of shots at 1/350s - a fairly safe speed to ensure I captured a sharp photo. I then grabbed my tripod and went for the gold with ISO 100 and 1/10s.

[Kerrowdown]

ISO - what do you REALLY use?

Certainly for me... a lot higher these days with the K1s, if my "old school" brain will allow settings beyond the 100 ASA level.

[photoptimist]

Ultimately equivalence is codswallop unless everything behind the sensor is exactly the same - pixel size and pitch, pixel design, presence and exact nature of AA or IR filters, digital signal processing hardware and especially firmware - everything.

Back in the film days, when the same basic recording medium (down to the same chemical batch of the same emulsion) could be put behind anything from 110 to 8" by 10" in any camera you cared to name, perhaps broader comparisons could be drawn. Now they cannot. I'm not sure you can do an apples to apples comparison between what happens when the light hits the sensor of a 645D vs. a K-1 or Nikon D810, for example. This is why I get wary when I see people arguing that such and such a format is "worth an extra stop". It's not automatically so - there are too many other variables coming into play, and most of them relate to what happens between when the shutter finally closes and the customer gets the print they paid for.

Although estimating the effect of equivalence might be easier with perfectly matched big and small format system, both statistical analysis of cameras and physics suffices to prove the point. Multivariate analysis can readily handle all the variables to estimate the effect of formate size on image quality.

And simple physics proves that a 75 mm lens at f/4 & 1/1000 shutter speed on an FF camera gathers more than twice as many photons (that's the signal) as a 50 mm lens at f/4 & 1/1000 shutter speed on an APS-C camera. Your film emulsion example proves the validity of equivalence in that the "noise" in a final printed image is much much lower with an 8x10 negative than with a 110-format negative.

Even if digital cameras have shown vast improvements in their noise levels (analogous to devleoping batter and better emulsions), the point remains that a larger format camera gathers more signal than a smaller format one. Two digital cameras of fromt he same generation of technological refinement, but different format sizes will show the equivalence effect just as two film cameras do.

[Edgar_in_Indy]

And simple physics proves that a 75 mm lens at f/4 & 1/1000 shutter speed on an FF camera gathers more than twice as many photons (that's the signal) as a 50 mm lens at f/4 & 1/1000 shutter speed on an APS-C camera. Your film emulsion example proves the validity of equivalence in that the "noise" in a final printed image is much much lower with an 8x10 negative than with a 110-format negative.

I'm surprised that anybody even wants to argue this point, but I've seen people do it many times, and vehemently.

[mcgregni]

We have all been constrained by ISO performance limitations with our cameras .... my K7 probably places more constraints on me than most here would now suffer!. But it all depends on the context of the image .... lighting, handheld or tripod, the steadiness required and movement capture needs.


There's no doubt that a K1 would considerably lesson my constraints! But I've often thought the ongoing mantra we've heard for so long about needing better ISO performance is so much more understandable if placed into the context of specific photographic situations ....


So for me, outdoors, family occasions, perhaps in dull light, I want to get sharp shots of the kids playing around, and with the K7 I might place a limit of ISO 800 on that. That would mean I could be shooting at ISO 800, F5.6, 1/60th .... but I need to zoom in close to 150-200mm, and that's getting shaky! Now with a K1, I believe I could expect at least around 3 - 4 stops of improved ISO performance equivalence, so just as good looking shots at ISO 6400 or 12800 as the K7 at 800 .... (that sound about right? .... maybe even more?)


So straight away with a K1 I'm now shooting those same photos with ISO 12800, F5.6, 1/1000th ..... wow! that's going to make a fantastic difference to my action shots and steadiness.


Indoors with flash I often find I am reaching the Dynamic Range limitations of my K7, as the contrast in a dark room with flash can be very high, and shadows even on skin tones can be set deep. I often find I am shooting at ISO 200 or 400 in order to improve battery life and recycling (as less flash power is needed), but even at those low ISOs I can still be getting colour noise in the shadows that needs attention in post ....


Now, if I moved to a K1 and shot those same images at ISO 400, or even ISO 800-1600, saving myself loads of flash batteries, I doubt I will have any concerns about shadow colour noise. Its looking like money well worth spending, if available!

[Edgar_in_Indy]

We have all been constrained by ISO performance limitations with our cameras .... my K7 probably places more constraints on me than most here would now suffer!.

Indoors with flash I often find I am reaching the Dynamic Range limitations of my K7, as the contrast in a dark room with flash can be very high, and shadows even on skin tones can be set deep. I often find I am shooting at ISO 200 or 400 in order to improve battery life and recycling (as less flash power is needed), but even at those low ISOs I can still be getting colour noise in the shadows that needs attention in post ....


Now, if I moved to a K1 and shot those same images at ISO 400, or even ISO 800-1600, saving myself loads of flash batteries, I doubt I will have any concerns about shadow colour noise. Its looking like money well worth spending, if available!

From what I recall, the K7 was a notoriously poor performer at high ISO, and was easily bested by my old K-x, which was Pentax's entry-level DSLR. You could upgrade to practically *any* Pentax DSLR camera from the last several years and get a dramatic increase in high ISO performance.

[skierd]

The theory is fine, but what does this have to do with proper exposure? Nothing. If I'm shooting with my K-30 at 1/1000 and f4 at ISO1600, I'd need to set a 645D at the same to get the same exposure. That's the whole point of aperture and ISO as it's the same across formats.

Now, noise at enlargement. An APS-C image enlarged to 8x10 is enlarged a lot more than a 645 image. So of course it's going to show more noise, if both are enlarged to the same size image. Enlarge the 645 image the same amount as the aps and you'll
see the same amount of noise, but have a much larger print.

Although estimating the effect of equivalence might be easier with perfectly matched big and small format system, both statistical analysis of cameras and physics suffices to prove the point. Multivariate analysis can readily handle all the variables to estimate the effect of formate size on image quality.

And simple physics proves that a 75 mm lens at f/4 & 1/1000 shutter speed on an FF camera gathers more than twice as many photons (that's the signal) as a 50 mm lens at f/4 & 1/1000 shutter speed on an APS-C camera. Your film emulsion example proves the validity of equivalence in that the "noise" in a final printed image is much much lower with an 8x10 negative than with a 110-format negative.

Even if digital cameras have shown vast improvements in their noise levels (analogous to devleoping batter and better emulsions), the point remains that a larger format camera gathers more signal than a smaller format one. Two digital cameras of fromt he same generation of technological refinement, but different format sizes will show the equivalence effect just as two film cameras do.

[photoptimist]

The theory is fine, but what does this have to do with proper exposure? Nothing. If I'm shooting with my K-30 at 1/1000 and f4 at ISO1600, I'd need to set a 645D at the same to get the same exposure. That's the whole point of aperture and ISO as it's the same across formats.

Now, noise at enlargement. An APS-C image enlarged to 8x10 is enlarged a lot more than a 645 image. So of course it's going to show more noise, if both are enlarged to the same size image. Enlarge the 645 image the same amount as the aps and you'll
see the same amount of noise, but have a much larger print.

Excellent points! Yes, the entire rationale for normalizing the physical diameter of the lens' aperture by the focal length (hence apertures are written as f/4 which means the physical diameter of the lens aperture is the focal length divided by 4) is to normalize the photon flux at the image plane in different optical systems. That definition of normalized aperture size lets one use an stand-alone light meter that does not know the focal length of the lens or format of the camera and whose reading can be used to set any lens on any camera.

But there are a multitude of possible "proper exposures" in the exposure triangle associated with a given EV reading of the light meter. For example, one can dial in: (1/1000 and f/4 at ISO1600), (1/1000 and f/8 at ISO6400), or (1/4000 and f/4 at ISO6400) and get the same 18% gray object being rendered at the same percent gray in the image. Yet even if the overall tonality is the same for those three exposure settings, the noise levels will be different for different ISO settings and for different format sizes. The APS-C user might be reluctant to pick ISO 6400 due to noise but the 645 user might more readily chose higher ISO. As such, the 645 user might say that his camera can handle lower-light situations as if he had a much faster lens than is actually on the camera even though he really doesn't. (Of course, that issue brings up the DoF differences of different lens-camera combinations which is another hairball!)

And you are right about noise at enlargement. In that regard, "equivalence" is an image quality phenomenon, not a pixel quality phenomenon. That is, it's something noticed in comparing two final prints taken with different formats rather than pixel peeping files created with different formats.

[normhead]

And simple physics proves that a 75 mm lens at f/4 & 1/1000 shutter speed on an FF camera gathers more than twice as many photons (that's the signal) as a 50 mm lens at f/4 & 1/1000 shutter speed on an APS-C camera.

On A Nikon D800 the Depth of field 75mm at 10 feet is from at ƒ5.6 in 9'2" to 11'0 or 22 inches. (As per Depth of Field Table)
On Pentax K-30 at 10 feet and ƒ4 9'1.5" to 11'1 inches9" or 23.5 inches.

To come close to matching the DoF of the K-30 with the D800 (or K-1) you need to close your aperture a stop, and it still doesn't quite make it.

So for the same image ( which would be the image with the same DoF you need to shut down the K-1 another stop, and the total light for the same image is the same.

I have done this so many times, I'm getting sick of it. Someone should make a sticky.

Listen, if you don't believe me, pick your own test go to the DoF chart and do it yourself. I've done it at least 5 times and it's always the same. A 50mm lens on APS-c has more Depth of Field than a 75mm lens at the same distance on a K-1. I've even tested this myself out in the parking lot.. That's why small sensors excel at keeping everything in focus. From the same distance with an equivalent lens, the you get more and more DoF at the same aperture, the smaller the sensor is.

Only when you cheat on the DoF do you get more total light. Same image, same distance, same DoF, same total light. Your "simple physics proves" point actually proves you can prove anything you want with simplel physics by leaving out relevant parameters (clearly over simplified) , and hoping your readers are too ignorant to pick up the mistake.

The question remains, why did you need to do that? What axe are you grinding?

This was the constant argument of all the "Pentax doesn't have an FF" people before the K-1 came out. People who were constantly advocating switching to Canon or Nikon, to take advantage of "total light".