[rangercarp]

I attended a workshop this past weekend and the instructor told us we should never use 1/3 ISO stops. He described them as "junk" and said image quality is better if you bump up to the next full stop. The argument is that the intermediate ISO's are not true ISO's but the processor simply manipulates the exposure to achieve the intermediate setting.

I have spent over an hour searching old threads and found a few that touched on this issue, but all of them had very few replies and did not come to any real conclusion. For as long as I have been using a DSLR I have used 1/3 ISO stops and never noticed a problem, but then, I never new I should be looking for one.

What is the prevailing opinion out there on 1/3 ISO stops?

[Venom3300]

That sounds like old canon sensor logic. I dont think there is any truth to that on newer sony sensors.

[rangercarp]

The instructor shoots with a D800, But said he had observed the same phenomenon with Canon. Overall he seemed to know his stuff, but this point I question.

[Sandy Hancock]

I'm too lazy to turn the dial three times as far to go from 200 to 400.
Half stops (or even thirds if you really want) are fine for aperture and shutter speed, but sensitivity? Really??

[Venom3300]

I know for sure there is a real difference between iso 80 and iso 100 on a k-5

[UncleVanya]

I use 1/3 stops and I have no idea if it works as effective as full stops or is a cheat. i have toyed with full stops on ISO and can't decide if there is much difference in the way I shoot.

[Cynog Ap Brychan]

I use half stops for everything: shutter speed, aperture and ISO. I'm not sure why, as I suspect that full stops would be just as good, given the manipulation I do afterwards in Photoshop.

[Apet-Sure]

Isn't it true that any sensor has a 'native' ISO? If that's the case, then any other ISO setting is processor controlled, and 'artificial'. To use an audio analogy, the more you turn up an amplifier, the more unwanted background noise you get. Whether you turn it up in small increments or larger increments, it doesn't change the fact that you get more noise. Wouldn't the same be true for the raw output of the image sensor?

I read an article on the development of the K-1 that discussed how to get improved results at very high ISO settings. Interestingly, reducing the inherent noise of the camera's circuitry was key, just like it would be for a 'clean' audio amp. It seems to me that whether you bump your ISO in third-stops or half stops, neither option would be inherently better for image quality. Am I wrong-headed?

[jatrax]

I use 1/3 stops on everything so that all three variables adjust at the same rate. I'm going to go with the instructor's logic being nothing more than internet myth. On film different ISO values were baked into the film so his logic makes some sort of sense that way, maybe, sorta. But on digital the ISO is just gain on the original signal, there are no stops that are any better than any other.

If this was true any number of tests would indicate a better value at some ISO's in a stepped fashion rather than the smooth curve we actually see.

[UncleVanya]

Isn't it true that any sensor has a 'native' ISO? If that's the case, then any other ISO setting is processor controlled, and 'artificial'. To use an audio analogy, the more you turn up an amplifier, the more unwanted background noise you get. Whether you turn it up in small increments or larger increments, it doesn't change the fact that you get more noise. Wouldn't the same be true for the raw output of the image sensor?

I read an article on the development of the K-1 that discussed how to get improved results at very high ISO settings. Interestingly, reducing the inherent noise of the camera's circuitry was key, just like it would be for a 'clean' audio amp. It seems to me that whether you bump your ISO in third-stops or half stops, neither option would be inherently better for image quality. Am I wrong-headed?

Some sensors are isoinvariant and essentially the ISO setting really has little impact. Some require gain prior to shooting the shot or they do not capture well. Even isoinvariant sensors have different characteristics if you shoot at native and then boost in post vs. boosting pre-shot. However the point is the sensors drive this - the K-5 is mostly isoinvarient as is the K-1 from what I have read. The K-3 is not as isoinvariant as those.

[rangercarp]

It seems to me that whether you bump your ISO in third-stops or half stops, neither option would be inherently better for image quality. Am I wrong-headed?

Your logic seems sound. To me it is counter-intuitive to suggest cranking it up a full stop in in anyway better than cranking it up by a third.

[robthebloke]

I use 1/3 stops on the k-1. To my eyes there is a difference between ISO 3200/4000/5000/6400/8000. Id agree that maybe the difference between 160 and 200 is minimal, if there is any difference at all, but the ability to have fine grained control over where the shutter speed range falls (so that the camera is generally selecting 1/80 rather than 1/60) is justification enough for me.

[todd]

I use 1/3 stops on everything so that all three variables adjust at the same rate.

Same here. It's very convenient to count clicks to make small adjustments.. I've been set this way since I got my K-5 in 2012 and it felt limiting last year I tried switching back to half and full stops (just out of curiosity)...

[Na Horuk]

The argument is that the intermediate ISO's are not true ISO's but the processor simply manipulates the exposure to achieve the intermediate setting.

Yeah, I heard that before. I'm not sure if this still applies with modern cameras, or if anyone has done tests to prove/disprove it.

I'm too lazy to turn the dial three times as far to go from 200 to 400.

For me this was a bigger deal than possible IQ issues. At the ISOs that I usually shoot, the noise levels are so low it doesn't matter much if I bump EV by 1/3 in post, or even if its done in-camera.
I think this would only show up if you shoot at relatively high ISO and apply a good amount of PP (added contrast, EV, without applying special NR). I think this is the answer for OP. Do whichever is more convenient for you, unless you shoot ISO 1600+ and need to do a lot of PP.

---------- Post added 24th May 2016 at 18:43 ----------

I use 1/3 stops on the k-1. To my eyes there is a difference between ISO 3200/4000/5000/6400/8000.

Yeah, but the question is whether shooting ISO 3200 and adding 1/3 EV in Lightroom or Faststone would bring out less noise than selecting ISO 4000 in-camera to begin with (shutter speed and aperture being equal).

[mattt]

Thumbs up for 1/3 stops when using TAv mode.

Full stops shooting legacy glass - its easier to remember settings from my film shooting.