[Rondec]

How we operate our cameras hasn't changed, but how the sensor is working is not the same as film. This is a thread about the sensor performance and people keep talking about ISO like there is actually varying levels of sensitivity, and that is not the case.

---------- Post added 09-17-16 at 09:01 AM ----------



Yes the relationship is still roughly the same or hand held light meters wouldn't work with DSLRs. Some manufacturers are exposing for 12% gray and some are exposing for 18% gray. It's not that they are cheating ISO, they are simply using a slightly different standard.

Clearly the goal is to have an iso-less sensor. Sony sensors are pretty close, but the A7r II and A7s sensors do some on-sensor magic to bump dynamic range at high isos that other sensors don't do.

[normhead]

How we operate our cameras hasn't changed, but how the sensor is working is not the same as film. This is a thread about the sensor performance and people keep talking about ISO like there is actually varying levels of sensitivity, and that is not the case.

ISO is about how people and cameras deal with lower than optimal light levels. So in that sense it hasn't changed. There were ways of processing, using larger grains in your emulsion, under-exposing and pushing your developing, all methods of dealing with low light. With digital, it's all done in camera, but it's the same struggle. An electronic struggle not a chemical struggle, but not so different really in terms of overall effect.

[Rondec]

I guess the question is what DXO Mark means when they say at iso 3200 they measure the K-1 to have an iso of 3265, but the A7r II to have an iso of 2293 and the D810 to have an iso of 2367. To me, that means that if you would manually set your iso to 3200 and shutter speed and aperture the same on all three of these cameras, the A7r II and D810 will come out half a stop under exposed compared to the K-1.

If you only own one of these cameras, you probably wouldn't know the difference if, say the D810 uses iso 4000 in situations where the K-1 only uses 3200. The end result is exactly the same, the only difference is that if you over state your iso, then the dynamic range at a given iso will look better, when in reality the curves lie right on top of each other (actually the K-1 is a little better than the D810).

[biz-engineer]

If you only own one of these cameras, you probably wouldn't know the difference if, say the D810 uses iso 4000 in situations where the K-1 only uses 3200. The end result is exactly the same, the only difference is that if you over state your iso, then the dynamic range at a given iso will look better, when in reality the curves lie right on top of each other (actually the K-1 is a little better than the D810).

That is true , you won't see a difference when using Auto ISO. But in cases where you'll use a light meter, or if you fix the ISO yourself, you won't get consistent exposures with A7r or D810, while you'll still get wanted exposure with a K1. With Pentax you can still decide to overexpose or underexpose but that's you decision, while A7r and D810 already decided something without telling you. I prefer the K1 in this respect, as I don't consider sensor performance on DXO chart if it does not relate to practical use of the camera in the field.

[Winder]

I guess the question is what DXO Mark means when they say at iso 3200 they measure the K-1 to have an iso of 3265, but the A7r II to have an iso of 2293 and the D810 to have an iso of 2367. To me, that means that if you would manually set your iso to 3200 and shutter speed and aperture the same on all three of these cameras, the A7r II and D810 will come out half a stop under exposed compared to the K-1.

If you only own one of these cameras, you probably wouldn't know the difference if, say the D810 uses iso 4000 in situations where the K-1 only uses 3200. The end result is exactly the same, the only difference is that if you over state your iso, then the dynamic range at a given iso will look better, when in reality the curves lie right on top of each other (actually the K-1 is a little better than the D810).

They are probably adjusting each exposure to a sRGB brightness of 48.45% or RGB of 124,124,124 and then test for noise, DR, & color.

---------- Post added 09-17-16 at 10:18 AM ----------

With Pentax you can still decide to overexpose or underexpose but that's you decision, while A7r and D810 already decided something without telling you. I prefer the K1 in this respect,

You can under or over expose any of them and they are all deciding many things without telling you. As long as you are familiar with each camera, there is really no practical difference between them when it comes to the final image quality.

[beholder3]

I guess the question is what DXO Mark means when they say at iso 3200 they measure the K-1 to have an iso of 3265, but the A7r II to have an iso of 2293 and the D810 to have an iso of 2367. To me, that means that if you would manually set your iso to 3200 and shutter speed and aperture the same on all three of these cameras, the A7r II and D810 will come out half a stop under exposed compared to the K-1.

If you only own one of these cameras, you probably wouldn't know the difference if, say the D810 uses iso 4000 in situations where the K-1 only uses 3200. The end result is exactly the same, the only difference is that if you over state your iso, then the dynamic range at a given iso will look better, when in reality the curves lie right on top of each other (actually the K-1 is a little better than the D810).

It means you need to expose the same sports image for 1/500 on a Pentax and 1/350 on the Sony to achieve the same results in Raw when using same nominal ISO settings.
And: Never compare images using the same nominal/fake exposure parameters or you are doing nonsense.

[normhead]

But much of the time I have to adjust my exposures with the histogram in any case, so it ask becomes moot.

[Class A]

You have to remember, they shoot in a windowless space with nothing but one incandescent light.

Where did you read that?

From the DxOMark page on their lens measurements:

"The target is uniformly illuminated with halogen lights filtered to get a daylight color temperature of 5500K."

.

[Rondec]

But much of the time I have to adjust my exposures with the histogram in any case, so it ask becomes moot.

If you look at the SNR or dynamic range curves, it is moot, because you can see that regardless of iso inflation, the curves lie smack dab on top of each other for the K-1 and D810 -- or close enough as makes no difference.

The "problem" comes in when someone posts their Fuji photo taken at iso 6400 and claims that it has better dynamic range and less noise than even full frame cameras. But if their camera is really only shooting iso 4000 when it claims to be shooting iso 6400, then of course your image will look cleaner than another brands iso which is a lot closer to 6400.

[biz-engineer]

But if their camera is really only shooting iso 4000 when it claims to be shooting iso 6400, then of course your image will look cleaner than another brands iso which is a lot closer to 6400.

That's the point, + actual shutter speed lower on the 4000 ISO camera; so that's cheating.

[Gimbal]

If you look at the SNR or dynamic range curves, it is moot, because you can see that regardless of iso inflation, the curves lie smack dab on top of each other for the K-1 and D810 -- or close enough as makes no difference.

Actually the dynamic range curves are not on top of each other, there is almost one stop between them.
Well maybe not a full stop but at least 0,6-0,7 stops between iso100-1600.

[monochrome]

Actually the dynamic range curves are not on top of each other, there is almost one stop between them. Well maybe not a full stop but at least 0,6-0,7 stops between iso100-1600.

"The Nikon D810 just edges the K-1 for both color depth and dynamic range, but the results are very close, and the K-1 trumps the Nikon by around a third of a stop for ISO, with scores of ISO 3280 vs. ISO 2853, respectively.

"The Nikon D810’s achieves its fractionally-higher headline dynamic range (Landscape) score of 14.8 EV at its base ISO of 64. The K-1 achieves a very close 14.6 EV at ISO 100, however, and between ISO 100 to ISO 800, offers greater dynamic range than both the Nikon D810 and Sony A7R. All three 36Mp sensors offer good dynamic range of over 10 EV up to ISO 3200."

"Signal-to-Noise Ratio (SNR 18%) analysis shows that for printed results, the K-1, D810, and A7R broadly offer the same levels of noise at all sensitivities. Excellent SNRs are achieved up to ISO 400, with well-controlled noise on RAW files at ISO 1600, too. At the higher sensitivities, noise becomes more prevalent, with the Sony A7R performing slightly better at ISO 25,600 and the Pentax K-1 showing its lowest SNR of 7.1dB at ISO 204,800."

[normhead]

Where did you read that?

From the DxOMark page on their lens measurements:

"The target is uniformly illuminated with halogen lights filtered to get a daylight color temperature of 5500K."

.

I read that on the DxO website, on a page since taken down. When people point out the flaw in their methodology, they just change their story.

The light source description used to read that it was a light source similar to what you would typically find in windowless basement at a social function using off the shelf household lighting. The fact that they changed the temperature without any description of when the change was made and what is old an what is new data, also makes much of their previously done work suspect.

[Class A]

I read that on the DxO website, on a page since taken down.

Was it perhaps this sentence from the description of colour sensitivity measurements:

"Typically, with a tungsten illuminant, the blue sensitivity is very low, due to the weak sensitivity of silicon in short wavelengths and the lack of short wavelengths in tungsten light."

?

That sentence just states a fact, as opposed to describing their process.

When people point out the flaw in their methodology, they just change their story.

I cannot confirm that and unless you can backup your serious allegation (use a wayback archive or similar) I will regard your statement as a baseless contention.

We may not agree with DxOMark's idea of a single score for a complex performance (a questionable practice for sure) and/or the weightings that DxOMark uses to obtain that single number, but you surely cannot accuse them of not knowing what they are doing, or using inadequate procedures.

DxOMark's procedures are well thought out and there is no better alternative anyhow. To preempt your counterargument: No, (unscaled) test shots under varying conditions with undefined focus targets are not a substitute.

[clackers]

Where did you read that?

From the DxOMark page on their lens measurements:

"The target is uniformly illuminated with halogen lights filtered to get a daylight color temperature of 5500K."

.

?

Halogen lamps are incandescent.