[Cannikin]

Two photos, same exposure:
Fuji @ 1/50s, f/4, ISO 100
Pentax @ 1/100s, f/4, ISO 100

Yes, one of the basic principles of exposure is if you use a slower shutter speed, you increase the exposure, with aperture and ISO constant.

In your example, the Fuji should be overexposed (not underexposed as you suggest) by 1 stop compared to the Pentax. But it is not, because the actual ISO is lower than reported.

[audiobomber]

Yes, one of the basic principles of exposure is if you use a slower shutter speed, you increase the exposure, with aperture and ISO constant.

In your example, the Fuji should be overexposed (not underexposed as you suggest) by 1 stop compared to the Pentax. But it is not, because the actual ISO is lower than reported.

Sorry, you're right, my brain was on vacation. I corrected my post above to minimize the confusion.

[rrstuff]

;it is just as if car manufacturers posted inaccurate speed readings.

on paper the car goes 0-60 in 4 seconds, but it really is 0-50. Gas consumption is lower, car is more silent, police never gives you speeding ticket.

then consumer reports does a review of a car and says exactly that: the car is quick, silent and economical.

next thing you know it is 0-30 that is measured instead of 0-60. That's what happened to ISO settings and dpreview encourages this kind of inflation.

[IchabodCrane]

I agree. If you are testing, you definitely should shoot manual and then comment after the fact that shooting at iso 1600 causes under exposure by 1/2 stop or, whatever.

It's hard to do that because you first need a universal standard condition. I suppose it's possible with a very high quality light meter but even then you can't be certain.

This will always be judgment-based, IMHO. What one person claims is properly exposed, another may claim as underexposed. etc. Also, it's a judgment exercise to make two photographs equal in perceived exposure.

[K David]

Honesty vs. marketing.

Honesty vs. Dpreview

[fgaudet]

I'm not saying that the ISO fudging isn't there but it may not be as bad as it looks. A nikon lens at f/4 may have a transmittance of t/5.6 (for example) and a pentax lens at f/4 may have a t/4.5, allowing more light thru... F-stop is the same but the amount of light hitting the sensor favours the Pentax (in this example).

I didn't look what lenses DP are using on the various camera. The t-stop of the lens should also be listed since it does affect exposure.

But then again, it could swing both ways... If for some reason the nikon lens from the previous example had a better light transmittance than the Pentax, the ISO fudging would be even worse...

The whole thing is sad... bad press for a great camera (and brand) based on not-so accurate measurements... But hey, that's how it is and not only in the photography world.

[fgaudet]

ISO fudging does not refer to the t-stop value of a lens

Really? I was under the impression that light was affecting exposure and that ISO was part of it too, sorry, my bad.

Two photos, same exposure:
Fuji @ 1/50s, f/4, ISO 100
Pentax @ 1/100s, f/4, ISO 100

In this example, if the Pentax lens is about one t-stop faster, it would explain the shutter speed difference.

I'm not saying Nikon, Canon, Sony, Olympus and Fuji are not generous with their ISO rating because they are. I'm only saying that if we're going to start comparing apples to apples, the lenses are better be "equal". A Canon 24-70 f/2.8L USM @ f/5.6 will allow more light through than a Pentax DAL 18-55 at the same aperture... The aperture size will be the same but the better glass will transmit more light and more light = faster shutter. And on this topic too (aperture), manufacturer are being very generous with their measurements.

[Aristophanes]

;it is just as if car manufacturers posted inaccurate speed readings.

More like this:

http://www.nytimes.com/2013/12/24/business/hyundai-expands-choices-in-gas-mi...ment.html?_r=0

For years Hyundai fudged its mileage statements.

---------- Post added 04-02-14 at 09:48 PM ----------

T-stops do affect exposure, but ISO fudging refers to cameras ISO values not corresponding to real ISO's being used. Both affect exposure but they are not the same thing.

ISO sensitivity - DxOMark

You can buy the standard here:

ISO 12232:2006 - Photography -- Digital still cameras -- Determination of exposure index, ISO speed ratings, standard output sensitivity, and recommended exposure index

[IchabodCrane]

T-stops do affect exposure, but ISO fudging refers to cameras ISO values not corresponding to real ISO's being used. Both affect exposure but they are not the same thing.

When a Fuji camera is claiming ISO3200, this is likely to be a real ISO1400 or so. When a Pentax camera claims ISO3200 its going to be ISO3000 or so. That's where the difference is. Fuji is heavy into deception with their high ISO's.

But you can't compare the ISO reliability without a lens on the camera and hence the T-stop is quite relevant. Also, I assume your post was contained a typo. Did you mean ISO 2400 for what Fuji claims is ISO 3200?

[Cannikin]

It's not that difficult to factor the lens out of ISO comparisons (at least for cameras of the same sensor size class): use the same third party lens, say Samyang, which produces optically equivalent lenses for virtually every mount out there, including mirrorless.

Fix the shutter speed and the aperture, and let the ISO float. Compare the reported ISO on camera for a given exposure to that on an independent light meter.

[derekkite]

So the test would have to be in standard conditions, at a given shutter speed and aperture wide open. Assuming shutter speeds are accurate.

The most instructive would be two tests; constant aperture through the stops of shutter speed with iso adjusting to expose properly, then constant shutter speed through the stops of aperture.

Let the body set exposure, choose a lens focal length and aperture that is reasonably common to all manufacturers.

The noise characteristics would be different on both tests at a given iso. Choice of lens makes a difference as well; some lenses have noise characteristics that are uncorrectable.

---------- Post added 04-02-14 at 07:49 PM ----------

I'd add two more tests. Actual vs indicated shutter speed. And measured depth of field at indicated fstop. The second would be more of a lens test.

[rrstuff]

I would just be happy if they normalized for different ISO, as this is a simple tweak in firmware. I am pretty sure there is a lot of inconsistencies in focal length, f-stops and maybe in shutter speed too. But ISO is the most obvious one.

[IchabodCrane]

From DXO...

Fuji X100: Claimed ISO3200 = Measured 1,012
Pentax K-3: Claimed ISO3200 = Measured 2,933
Olympus EM1: Claimed ISO3200 = Measured 1,900

I think it's useful to read the fine print:

"A manufacturer has two options for increasing ISO sensitivity; either to order the sensor to increase its analog sensitivity, or after a picture is taken to tell the camera to amplify the image signal via algorithms at the RAW conversion stage.
Fujifilm decided to implement the second choice with the X100 and this processing action is normally invisible to the user.
As long as the algorithms used for the treatment are performing well, there is no real problem for the end user. As far as RAW measurements are concerned, the sensor’s maximum sensitivity is ISO 1000. That’s why none of our metrics goes further."


Hasn't it been proposed by some here that the Sony sensor really only goes to something like ISO 1600 (if that)? Everything past that is just from gaining up the signal. Doesn't seem DXO has tested any Fujifilm sensor in the past three years so we can only speculate the results are the same for later models.