ISO is Fake - Seriously

[pdk42]

Re: ISO is Fake - Seriously

What does he say that is nonsense? Please share, I genuinely don't know where you get that from? Did you actually watch it?

The basic point is that ISO on a digital camera is not a standard despite what manufacturers try and convince us it is. On Astro cameras they are honest and simply call it Gain.

Yes, Tony does create clickbait but at least if you are going to be critical have some facts that prove he is wrong.

The problem with the video is that he's just doing a subjective viewing of the image brightness and of course the final brightness will be subject to all sorts of processing changes. I don't doubt that different cameras (and processing) can produce different final brightness for a specific ISO setting, but Tony's conclusion is simply that the ISO is "wrong". That's just a huge oversimplification and he states it simply to gain clicks and feed the ignorant.

For film, it's easy to measure ISO since all you do is determine the exposure required to build a particular density on the emulsion after development. Of course, you need to calibrate the development process, the measuring eqpt etc, but that's all quite doable. For digital though, it's not at all so clear since the final image brightness can be manipulated within very wide margins after exposure. There are at least three approaches you could take:

- The exposure needed to saturate the sensor (clipping) can be used as the baseline. Then, the ISO is (somewhat arbitrarily) calculated as the light needed to produce a particular pixel saturation based on the reflected light from an 18% grey card at a given illumination level (i.e. what your Weston Master would tell you!). The standard is for that 18% grey to produce a 12.7% pixel saturation.

- The level of acceptable noise in the final image. In this test, the S/N ratio of the image is used to determine what an acceptable exposure is. There are two thresholds used for the S/N - 40:1 and 10:1, determined by subjective analysis of a particular print resolution and viewing distance.

- A so-called Standard Output Sensitivity test which measures the exposure needed to produce a specific bit value in an RGB file with a specific gamma. This will include all the processing that happens on a file and can be thought of as a "JPEG" measure.

Most manufacturers will use some combination of the above to arrive at a number, but there are other complications such as spectral characteristics and white balance settings which should also be taken into account.

The bottom line is that measuring ISO is complex and trying to claim that manufacturers are in some way misleading us is, well, misleading!

[wornish]

Re: ISO is Fake - Seriously

For film then yes ISO is a calculated standard as Tony says if you did watch it.

For digital its meaningless. It is just the gain applied to the sensor output again as Tony says. It's not a standard by any measure merely a misused marketing ploy. Oly is one of the worst offenders, but they all do it. That was the point Tony was making. But yes I know he loves generating click bait, sometimes what he says is however true.

[Petrochemist]

Re: ISO is Fake - Seriously

There are two different industry standard interpretations of ISO calibration (at least).

Ian

I know of two for film:
ISO 5800 for colour negative
ISO2240 for Black & white


For digital we have ISO 12232 which apparently has 5 different techniques that can give different results. There were no official speed designations above ISO10,000 in the 2006 release, a new 2019 version is now available I don't know how significant the changes are.



Oh and it seems several high sensitivity films were marketed as having higher ISO than the standards gave... https://en.wikipedia.org/wiki/Film_speed

[Naughty Nigel]

Re: ISO is Fake - Seriously

I think the critical issue for us as users is rarely what the absolute ISO number is for a given camera/image, but rather whether the camera is capable of being used to produce images that we find properly exposed.

The true ISO sensitivity of digital cameras only becomes relevant when you need to work with external devices such as exposure meters, studio flash or other cameras.

I have often used my OM4Ti to meter a scene for my Mamiya RZ and have fount it extremely accurate. (The metering head for the RZ is a bit big and heavy to carry very far along disused railway lines.)

My current OM-D E-M1 is also pretty close, but my old Olympus E1 was miles out, especially at higher ISO settings. I know I am cynical about these things but I suspect the noise from the E1 at a true 400 ASA might have had something to do with this.

Likewise I had a Canon PowerShot that was even further out, with an indicated ISO 400 actually being closer to ISO 240; probably for the same reasons.

Oddly enough, I use an exposure meter app on my Sony Z3 and Z3 Compact mobile phones, (which of course have very small sensors), and that is remarkably accurate.

[birdboy]

Re: ISO is Fake - Seriously

What does he say that is nonsense? Please share, I genuinely don't know where you get that from? Did you actually watch it?

The basic point is that ISO on a digital camera is not a standard despite what manufacturers try and convince us it is. On Astro cameras they are honest and simply call it Gain.

Yes, Tony does create clickbait but at least if you are going to be critical have some facts that prove he is wrong.

I would say that the video is misleading. As someone who is into astrophotography I have really tried to understand the ISO effect. From what I have read from trusted sources and books for Olympus cameras ISO is a pre ADC gain i.e it boosts the analogue voltage signal of the photon hitting the photo sensor. You cannot change the ISO once the image has been taken, you can change the brightness but not the ISO. If it was a gain post ADC you could change it. Now in astophotography (or very low light) only a few photons will be hitting the photo sensor. Read errors and ADC quantisation add noise after the ADC. If you kept the ISO at native ISO say 200 (Olympus) you will get more noise in your image than if you had upped the ISO to say 1600 (3x gain pre ADC). That is because the signal is more pre ADC therefore giving a better s/n ratio.

I am not sure if I have explained this that well But ISO in cameras is such a confusing and often misquoted issue.

[Naughty Nigel]

Re: ISO is Fake - Seriously

You cannot change the ISO once the image has been taken, you can change the brightness but not the ISO.

How does ISO bracketing work then, when only one exposure is made?

Does the camera apply differing levels of gain to each image, with ISO numbers given accordingly?

[birdboy]

Re: ISO is Fake - Seriously

How does ISO bracketing work then, when only one exposure is made?

Does the camera apply differing levels of gain to each image, with ISO numbers given accordingly?

I thought it took several images at different ISO's and merged them in camera, so yes I think it does apply different levels of gain (pre ADC).

[Naughty Nigel]

Re: ISO is Fake - Seriously

I thought it took several images at different ISO's and merged them in camera, so yes I think it does apply different levels of gain (pre ADC).

[birdboy]

Re: ISO is Fake - Seriously

My understanding is that bracketing does not merge the images. With ISOBKT it takes the 3 frames with whatever ISO value you have set +/- the step. It only takes one shutter button press to create the 3 images. You could get the same effect by varying the ISO yourself but you would need 3 press's to do it. You then need to merge the images in software such as LR Enfuse plugin. I do not have Photoshop.

EM1 MKII manual p93
"The camera varies the sensitivity over three shots while keeping the shutter speed and aperture fixed. You can select the bracketing increment from 0.3 EV, 0.7 EV, and 1.0 EV. Each time the shutter button is pressed, the camera shoots three frames with the set sensitivity (or if auto sensitivity is selected, the optimal sensitivity setting) on the first shot, negative modification on the second shot, and positive modification on the third shot."

[Naughty Nigel]

Re: ISO is Fake - Seriously

My understanding is that bracketing does not merge the images. With ISOBKT it takes the 3 frames with whatever ISO value you have set +/- the step. It only takes one shutter button press to create the 3 images. You could get the same effect by varying the ISO yourself but you would need 3 press's to do it. You then need to merge the images in software such as LR Enfuse plugin. I do not have Photoshop.

EM1 MKII manual p93
"The camera varies the sensitivity over three shots while keeping the shutter speed and aperture fixed. You can select the bracketing increment from 0.3 EV, 0.7 EV, and 1.0 EV. Each time the shutter button is pressed, the camera shoots three frames with the set sensitivity (or if auto sensitivity is selected, the optimal sensitivity setting) on the first shot, negative modification on the second shot, and positive modification on the third shot."

From my observation using the OM-D E-M1 Mk1 it only takes one shot. That is evident when taking longish exposures. Only after the exposure do the previews pop up on the screen.

My guess is that a single image is taken, the data temporarily held in memory and than amplified or attenuated to the required ISO bracketed values.

If that is the case I would be better off making gain adjustments in Photoshop.

[birdboy]

Re: ISO is Fake - Seriously

From my observation using the OM-D E-M1 Mk1 it only takes one shot. That is evident when taking longish exposures. Only after the exposure do the previews pop up on the screen.

My guess is that a single image is taken, the data temporarily held in memory and than amplified or attenuated to the required ISO bracketed values.

If that is the case I would be better off making gain adjustments in Photoshop.

I have just tried ISOBKT with my EM1 MKI and it takes 3 images, althoug I only hear one shutter press and if you you a 1 sec shutter it clearly takes 3 secs to take the picture. However I do find this feature very difficult to get to work, and hardy ever use it. You have to keep pressing the ok button having selected ISOBKT and the words BKT should appear on the evf or back screen. The manual clearly states it takes 3 images. Manual EM1 MK1 p83. If you are not getting 3 images then you cannot be taking bracketing shots.

[Petrochemist]

Re: ISO is Fake - Seriously

Surely the ISO sensitivity of a film or digital camera is a logarithmic function, so it is simply calculated from a standard of (say) 100?

https://en.wikipedia.org/wiki/Film_speed) is quite interesting if you want more technical details.
https://www.iso.org/obp/ui/#iso:std:...232:ed-3:v1:en gives a preview of the new version of the standard. Possibly only FOUR procedures with the deletion of Annex E.


Less relevant perhaps we also have:
ISO 2720 for exposure meters
ISO 2721 for camera auto metering
ISO 5763 for flash
ISO 7187 for direct positive colour film
ISO 10157 for flash meters