[dansamy]

Well aren't all manufacturers' iso claims an approximation or emulation of the film? How far off they are from actual performance just means they really suck at guessing!

[kosmoejtg]

I thought there might be some standards. It is easy to see why a manufacturer would fudge the numbers. It would be really hard for a user to prove it was off. Not like when Ford fudged the MPG for their hybrids, it was obvious to the customer.

[jhux1972]

I just looked up the X100 results to see if Fuji "has form" on cooking the ISO books.

Fuji claimed ISO1600 = ISO1000 (vs 1477 for the K-3)
Fuji claimed ISO3200 = ISO1012 (vs 2922 for the K-3)
Fuji claimed ISO6400 = ISO1076 (vs 6057 for the K-3)

LOL, that makes me a tad suspicious

WOW. This may explain those test shots and then some.

[dansamy]

WOW. This may explain those test shots and then some.

Easy for your high ISO shot to look good when it's only marginally higher iso than a stop or two down.

[DSims]

I didn't realize that ISO could be so different between manufacturers. I read a thread earlier bashing DXO, I guess they do serve a purpose if they expose this like this.

I'm one who's been critical of DXO. But they definitely serve a purpose.

The problem's more that some people quote them as an absolute authority, rather than simply using them as a resource which does useful, but limited testing.

Obviously, their results need to be applied properly. Their data should be used to predict, or perhaps explain, the results we'll get when using the equipment. And their "actual ISO" measurements can help make sure we're comparing apples to apples. But their numbers can't replace the actual results we get using the equipment.


The problem with both numbers and observation (quantitative and qualitative) is that they both require understanding the context - you have to pay close attention.

The pitfall with numbers is that they're prone to the illusion that they're easy and absolutely reliable. We can forget that they're still completely dependent on the test conditions, sample variations, and most importantly the fact that the tester has to think of all possible relevant variables, along with ways to deal with them. In some cases he may not be able to measure or adjust for certain variables - even if he's fortunate enough to identify them in the first place.

[Rondec]

I'm one who's been critical of DXO. But they definitely serve a purpose.

The problem's more that some people quote them as an absolute authority, rather than simply using them as a resource which does useful, but limited testing.

Obviously, their results need to be applied properly. Their data should be used to predict, or perhaps explain, the results we'll get when using the equipment. And their "actual ISO" measurements can help make sure we're comparing apples to apples. But their numbers can't replace the actual results we get using the equipment.


The problem with both numbers and observation (quantitative and qualitative) is that they both require understanding the context - you have to pay close attention.

The pitfall with numbers is that they're prone to the illusion that they're easy and absolutely reliable. We can forget that they're still completely dependent on the test conditions, sample variations, and most importantly the fact that the tester has to think of all possible relevant variables, along with ways to deal with them. In some cases he may not be able to measure or adjust for certain variables - even if he's fortunate enough to identify them in the first place.

I like their sensor compare function, not a big fan of their lens scoring. But, people are all too willing to write off cameras for 0.5 EV difference in dynamic range or a sports iso that is 80 points lower than another camera. A 5 point difference in the total sensor score may be visible in real life shooting and editing -- but even then, it will only be in relatively extreme situations. It would be nice if the Fuji was tested by DXO Mark, because I believe that it would end up scoring about the same as a K5 series camera -- probably not significantly better or worse. But I guess that won't happen because of Fuji's proprietary sensor.

[MJSfoto1956]

my take with DxO is just the opposite -- I find their system score (i.e. body + lens) to be very useful to understanding how to compare one system to another. What one can interpolate from the data is that Pentax DX lenses "tend" to deliver better performance/value than CaNikon DX lenses (and it is clear which are the exceptions). The unknowns will likely forever be unknown to the general public: for example, how does a Leica R lens converted by Leitax perform with the K-3 as a "system"? If one is adventurous, there are lots of great options out there no matter what body you are using.

Michael

[tuco]

I just looked up the X100 results to see if Fuji "has form" on cooking the ISO books.

Fuji claimed ISO1600 = ISO1000 (vs 1477 for the K-3)
Fuji claimed ISO3200 = ISO1012 (vs 2922 for the K-3)
Fuji claimed ISO6400 = ISO1076 (vs 6057 for the K-3)

LOL, that makes me a tad suspicious

It's not the actual measure ISO that should be a concern, I suspect. If one camera can get a correct exposure at a lower actual ISO then the other, all the better. That is, if both cameras produce a good exposure at, say, a setting of ISO 1600 on the camera and one actually tests to be a lower ISO I'd say all the better, no? It can capture a good exposure at a lower actual ISO while the other needs it cranked up more.

[geomez]

It's not the actual measure ISO that should be a concern, I suspect. If one camera can get a correct exposure at a lower actual ISO then the other, all the better. That is, if both cameras produce a good exposure at, say, a setting of ISO 1600 on the camera and one actually tests to be a lower ISO I'd say all the better, no? It can capture a good exposure at a lower actual ISO while the other needs it cranked up more.

The problem is that it's disingenuous. If the true ISO is far lower than the advertised ISO and a reviewer just takes the advertised ISO at face value, they may come to the conclusion that the camera does better than it actually does at a higher ISO or that it does better than a camera with a more true to life ISO at the same ISO. Fudging the ISO is using a lie to sell a camera.

[Rondec]

It's not the actual measure ISO that should be a concern, I suspect. If one camera can get a correct exposure at a lower actual ISO then the other, all the better. That is, if both cameras produce a good exposure at, say, a setting of ISO 1600 on the camera and one actually tests to be a lower ISO I'd say all the better, no? It can capture a good exposure at a lower actual ISO while the other needs it cranked up more.

I have read on this and the best I can come up with is The Online Photographer: Why ISO Isn't ISO, which doesn't really explain their measurements. It makes no sense to me that (if you are shooting raw) the actual iso can change from 1000 to 1100, your shutter speed can double and you have a decent exposure. The whole equation depends on iso being something meaningful. One stop difference in aperture, iso, or shutter speed has to be the same or you won't get accurate exposures.

[john5100]

K3 at high ISO

I shoot high ISO all of the time as a concert photographer. I did a show for a friend's son where the light was ridiculously low and I was shooting at ISO 10,000 and then pushed it 2 stops to get an image that worked. Yes it was graining but at least I got the shot.

This show was shot at 1/160, F2.8, ISO 10,000 pushed to ISO 40,000 because it was so dark. I told the lighting guy to turn up the lights but of course he didn't.


Here is an actual show at ISO 10,000

[jhux1972]

Good stuff John. I'll actually be able to try this myself as my stepson has a band. They actually had a concert this weekend. What lens(es) do you usually use? Guessing the 1.4 wide open?

[john5100]

Good stuff John. I'll actually be able to try this myself as my stepson has a band. They actually had a concert this weekend. What lens(es) do you usually use? Guessing the 1.4 wide open?

Have fun with it. Lighting for smaller shows is always difficult. I used the DA* 16-50mm at F2.8.

[jbinpg]

That isn't what happens though, that's the whole point. One camera can't get a correct exposure at a lower ISO without a wider aperture (or better T stop) and/or a slower shutter speed than the other camera.

Unless the image is being pushed in camera as in the ISO-less camera?

Jack