[chd]

Hi all,

I learned something new recently... evidently the ISO settings in most cameras are implemented in two different ways. For lower ISO numbers, this is an analog amplification stage applied to voltage read from the sensor. (This makes sense, and is useful.) At higher ISO settings though, the ISO increase is implemented digitally, as a simple multiplication of values after the analog-to-digital conversion. This is useless and can be done just as effectively in Photoshop or the application of your choice.

So, the question is, what is your camera's maximum analog ISO? It doesn't appear that anyone publishes this; does anyone know what it is for the K-1? (Or other cameras too, but the K-1 is what I use.)

Thanks!

[ne!]

Very interesting, I'll be fascinated to learn more.

[Adam]

Great discussion on that here:
ISO invariance for Pentax DSLR cameras? - PentaxForums.com

[stevebrot]

This has been a topic of endless discussion with little hard information. The most prevalent theories are based on inference from image noise analysis and have led to some interesting exposure practices.

For anyone interested in standards associated with ISO sensitivity assignment* and measurement, the Wikipedia article is a good starting place.

Film speed - Wikipedia | Digital Camera ISO and Exposure Index

Pentax cameras are CIPA certified using the SOS (Standard Output Sensitivity) method with reference to the 18% gray standard. Cameras from other makers are often certified using the REI (Recommended Exposure Index) where the maker recommends metering to the set index for best results and approximately linear response. The usefulness of a hand-held light meter may depend on how the camera maker defines ISO.


Steve

* The term ISO is a bit misleading in that it infers equivalence to ISO sensitivity ratings for photographic film. The two have similarities, but are dissimilar in terms of how the ratings are assigned and expected results. A better terminology might be to refer to sensor sensitivity by the long-standing generic term, Exposure Index (EI).

---

Last edited by stevebrot; 02-22-2019 at 09:07 PM.

[RGlasel]

what is your camera's maximum analog ISO? It doesn't appear that anyone publishes this

As Stevebrot posted, the confusion between ISO setting on a digital camera and the ISO sensitivity of film gives birth to a lot of misleading and incorrect information. ISO settings are just a number, which can be used to calculate aperture and shutter speed for the measured light intensity. The sensors used in most digital cameras have adjustable sensitivity (as you noted), which is adjusted by the computer in the camera according to how the camera is programmed (to give the type of peformance the camera is engineered for), not according to user settings. Not only is the sensor's sensitivity adjustable, but the camera can be programmed to compensate for non-linear sensitivity, so what might be considered maximum analog ISO can be well outside the range of light intensities where the sensor performs in a linear fashion (and from the standpoint of the user, it doesn't really matter). At some point, no amount of compensation can overcome the noise, but everyone has different noise tolerances. And because everyone is different, it wouldn't mean anything if manufacturers did publish "maximum analog ISO."

I've noticed that many serious camera users try to outsmart the camera designers by finding "mistakes" like ISO settings that the user feels are illegitimate. The people who design cameras have access to technical data that will never be public, because how the camera is engineered to get the desired performance from its sensor is a confidential business strategy, which can confer a competitive advantage. It's not a bug or a defect, it is a feature.

[rawr]

It doesn't appear that anyone publishes this; does anyone know what it is for the K-1? (Or other cameras too, but the K-1 is what I use.)

There are a few sites to look at with data that you may find useful, like Photons To Photos (even though I find his research often poorly described and presented). You can also play with DxOMark's charts for the K-1, but these don't give away much about 'real' vs 'amplified' camera ISO.

[biz-engineer]

As Stevebrot posted, the confusion between ISO setting on a digital camera and the ISO sensitivity of film gives birth to a lot of misleading and incorrect information.

There is no confusion. The value of ISO (gain of the signal chain), exposure time and lens aperture trio do not lie. Digital multiplication is part of the sensitivity of the imaging pipeline, which defined that relationship between output image and incoming light. Whether gain is achieved in analog or digital is done in order to deliver the best image results. So, there is no trick to improve image quality, the best trick is to use the camera normally.

[stevebrot]

The value of ISO (gain of the signal chain), exposure time and lens aperture trio do not lie.

Unless, of course, they do lie.

The proof is when one uses a hand-held meter and the indicated LV (ISO/shutter/aperture) for an unambiguous subject* results in significant over or under exposure over all or part of the camera's supported sensitivity range. Often enough, the internal meter is designed to compensate for such quirks, but that is not helpful for situations where using the internal meter is ill-advised or not possible. Pentax is pretty good in that regard.

So, there is no trick to improve image quality, the best trick is to use the camera normally.

Absolutely! The camera is designed for exactly that...normal use, and with upper end models, normal use is quite broadly defined. Trust the tech.


Steve

* Specifically, an evenly-lit white wall. One would expect the same or very similar LV from both the in-camera meter and a hand-held meter (with some allowance for light loss with the TTL reading). The histogram from an exposure using the hand-held meter reading should be centered and symmetrical.

[biz-engineer]

The proof is when one uses a hand-held meter and the indicated LV (ISO/shutter/aperture) for an unambiguous subject* results in significant over or under exposure over all or part of the camera's supported sensitivity range.

Pentax is pretty good in that regard.

I have a standalone light meter, and I compared with the K3 and K1, the difference is within +-0.3ev. Pentax ISOs seem to be close to the ISO spec., unless the values of shutter time displayed by the camera are different from the actual time of the mechanical shutter. For Fuji, I've seen that the ISO measured by DXO rather different from the ISO setting indicated by the cameras, I assumed that if the measured ISO on the Fuji was lower than the ISO setting, the actual shutter speed should be slower to compensate for the lower actual ISO. But since I don't have a Fuji camera I couldn't check my assumption is right.

[reh321]

Hi all,

I learned something new recently... evidently the ISO settings in most cameras are implemented in two different ways. For lower ISO numbers, this is an analog amplification stage applied to voltage read from the sensor. (This makes sense, and is useful.) At higher ISO settings though, the ISO increase is implemented digitally, as a simple multiplication of values after the analog-to-digital conversion. This is useless and can be done just as effectively in Photoshop or the application of your choice.

What do I gain by doing the multiplying in PP that the camera is willing to do for me?

What do I gain by doing the multiplying in PP that the camera is willing to do for me?

You might retain a little more detail using a lower ISO and compensating for it in PP if your camera has aggressive noise reduction (sometimes even with raw) kicking in above a certain ISO.

[photoptimist]

What do I gain by doing the multiplying in PP that the camera is willing to do for me?

You "gain" potentially ugly banding in flat areas such as the sky.

Imagine you under-expose an image by using an ISO that is 5-stops low and then multiply by 32 (= 5 stops) in PP to boost it. Now, instead of adjacent pixels in smooth areas differing by an unnoticeable 1 gray value, they differ by a jump of least 32 gray values.

This is why I'm skeptical that cameras use "post ADC multiplication" to any great extent -- it would create both artifacts in the image and a very strange spiky histogram.

[interested_observer]

I was looking for some information and stumbled on to this video the other day. It's Tony Northrup's take on ISO. What was interesting was that they had the K1 and K3 as the few cameras that actually had ISO 100 "valued" reasonably accurately.

You Tube

You Tube

Then Fstoppers had this follow up video on the Northrup video

You Tube

You Tube

[stevebrot]

I was looking for some information and stumbled on to this video the other day. It's Tony Northrup's take on ISO. What was interesting was that they had the K1 and K3 as the few cameras that actually had ISO 100 "valued" reasonably accurately.

ISO is totally FAKE. Seriously. - YouTube

Then Fstoppers had this follow up video on the Northrup video

Mythbusting Tony's ISO Claims - YouTube

Eeek! Tony and I are singing the same tune!


Steve

[stevebrot]

Then Fstoppers had this follow up video on the Northrup video

This is where things get interesting. The assumption is that Photoshop (or Lightroom for that matter) leave the base data alone when one "pushes" exposure in PP. The truth is quite different. Our PP tools do quite a bit of creative interpretation and "filling in the blanks". This is immediately obvious when one observes the distribution of values in the histogram. Inexplicably values at the low end tend to be retained even in the areas of "near black" rather than being boosted the same number of stops as the highlights.


Steve