Is the term ISO “totally fake”?

[tmroper]

On #2, film was (and still is) the same way. "Box speed" is only a rough suggestion, and each photographer must do their own tests, and calibrate things accordingly. So your light meter only "works as you expect" after you've determined what those expectations are. Also, not all light meters are the same, so you have to check that, too. Oh, and if you print in the darkroom, different batches of paper will be a little different, so you can't just use the same time and aperture from one to make identical prints later (that's what my darkroom instructor/guru says anyway--I'm not at that level yet). So yeah, life's not as simple as the marketing people claim. But you should know that already.

 

[Kit.]

Yes, there is: ISO (organization) standard 12232:2006 and 2019

So, basically (for shooting in RAW): "It's when you have a half stop extra to overexpose a 100% diffuse white object without highlight clipping".

 

[knight427]

I found the following link to be the most informative guide to understanding ISO as gain, with one layer drilling down into the details to be aware of the difference between upstream and downstream read noise (at least as it related to astrophotogrpahy use cases).

http://dslr-astrophotography.com/iso-dslr-astrophotography/

But Tony's video is far more entertaining with high conflict generation potential, and therefore will reach far more people.

 

[magarity]

Horseless carriages don't have horses pulling them but the engine power is described as 'horsepower'. So why not 'ISO' for 'sensitivity'? Is the problem that it isn't standardized?

 

[AlanF]

Just came across this by jrista, who used to make great contributions here (https://photo.stackexchange.com/questions/25917/why-is-sensor-sensitivity-called-iso)
"Just a note. When it comes to digital sensor "sensitivity", the term sensitivity in that context is actually a bit of a misnomer. A digital sensor is a fixed, linear, analog device. It always has the same real sensitivity. When you adjust the ISO setting to a higher level, all that really does is reduce the maximum saturation point. The sensor does not detect more light...it detects the same, so its still just as "sensitive". Its just that instead of pure white occurring at say 40,000 electrons in a pixel (ISO 100), it occurrs at 20,000 electrons (ISO 200), or 10,000 electrons (ISO 400), etc " jrista

 

[peterzuehlke]

Even without iso invariant sensors, it is still valuable to know about it specifically for astrophotography. There comes a point where lower ISOs may introduce more noise with longer exposures on some sensors which is why you tend to see most shots at higher ISOs like 1600-3200.

For performance photography too. When you are at the bleeding edge of exposure and are trying to make decisions between f stop, shutter speed and noise, ISO helps in that judgement. I often end up shooting at f/1.4 to f/2.8 when I would like to be at f/4 for DOF but I would also like to be above 1/125s because of subject movement but don't want to go too high ISO because of noise (I am usually at 6400 or higher)

 

[gwooding]

The better Canon sensors are isoinvariant above about iso160-200, which covers much of their range: see
http://www.photonstophotos.net/Charts/PDR.htm#Canon EOS 1D X Mark II,Canon EOS 5D Mark IV,Canon EOS 80D

for 5DIV (=EOS R), 80D, 1dXII
View attachment 183314

I do not believe that it is very easy to determine a sensors ISO-in-variance by looking at the sensors dynamic range.

The same site however does provide charts for what I believe to be an indication of ISO-in-variance. They call it "Photographic Dynamic Range Shadow Improvement versus ISO".

http://www.photonstophotos.net/Gene...ographic_Dynamic_Range_Shadow_Improvement.htm

If I understand correctly, what this chart shows is how much extra shadow detail is gained by increasing the ISO instead of raising an underexposed image in post. In other words whether it is worth using higher ISO's instead of just underexposing and raising in post.
http://www.photonstophotos.net/Charts/PDR_Shadow.htm#Canon EOS 1D X Mark II,Canon EOS 6D Mark II,Nikon D850,Pentax K-1


From this chart we can see that basically the closest of the 4 cameras to ISO-in-variance is the Pentax K-1 which only sees about a 0.2EV improvement increasing the ISO over base,

The Nikon D850 is ISO-in-variant above 500, with 500 yielding about a 1EV improvement.

The 1DX MK2 however only appears to see a no significant improvement once we get to about ISO 3200 (1.7 EV improvement). I suppose you could argue that the 0.3EV improvement from 800 to 3200 isn't much.

With the 6D MK2 you can clearly see that using higher ISO's are very beneficial (interestingly you can also easily see where the analogue gain is used and where the digital gain is used).

I also believe the preceding chart illustrates that the only camera of the ones I picked only the Pentax conforms to Tony Northrup's ideas (I must confess I didn't watch the video I just read the summary provided). Clearly if we look at something like the 6D2 the ISO setting (at least the full stops) has a big difference and works as you would expect although increasing ISO by one stop only gives you about 0.8EV extra sensitivity.

The quote by Jrista provided by AlanF is 100% correct. I just thought I could expand a bit on what increasing the ISO in camera actually does. As mentioned the signal coming off the sensor is an analogue signal so it needs to be read by an analogue to digital converter (ADC) which quantizes the signal (converts an analogue voltage into a digital number between 0 and 2^14 for a 14-bit ADC). This process introduces noise. If we then amplify the signal after this (raise in post or digital ISO) we also increase the magnitude of the noise thus we don't get as much improvement as we expected. One way around this is to amplify the signal before the ADC reads it, this reduces the headroom but normally improves the signal to noise ratio so as long as you are not clipping highlights it should be an improvement.

If the noise introduced by the ADC is negligible and the ADC has sufficient bit depth we can theoretically do all amplification digitally without any adverse effects, this would be a perfect ISO-in-variant sensor. The introduction of the low noise on sensor ADC's has definitely reduced the need for pre-amplification or perhaps eliminated it completely with certain cameras. I suspect the Pentax K-1 either has such good ADC's that it doesn't benefit from them or it doesn't actually have them (I think the latter is more likely since that would explain why the 1DX MK2 has better dynamic range at ISO's over 800).
http://www.photonstophotos.net/Charts/PDR.htm#Canon EOS 1D X Mark II,Pentax K-1

 

[AlanF]

I do not believe that it is very easy to determine a sensors ISO-in-variance by looking at the sensors dynamic range.

The same site however does provide charts for what I believe to be an indication of ISO-in-variance. They call it "Photographic Dynamic Range Shadow Improvement versus ISO".

http://www.photonstophotos.net/Gene...ographic_Dynamic_Range_Shadow_Improvement.htm

If I understand correctly, what this chart shows is how much extra shadow detail is gained by increasing the ISO instead of raising an underexposed image in post. In other words whether it is worth using higher ISO's instead of just underexposing and raising in post.
http://www.photonstophotos.net/Charts/PDR_Shadow.htm#Canon EOS 1D X Mark II,Canon EOS 6D Mark II,Nikon D850,Pentax K-1
View attachment 183321

From this chart we can see that basically the closest of the 4 cameras to ISO-in-variance is the Pentax K-1 which only sees about a 0.2EV improvement increasing the ISO over base,

The Nikon D850 is ISO-in-variant above 500, with 500 yielding about a 1EV improvement.

The 1DX MK2 however only appears to see a no significant improvement once we get to about ISO 3200 (1.7 EV improvement). I suppose you could argue that the 0.3EV improvement from 800 to 3200 isn't much.

With the 6D MK2 you can clearly see that using higher ISO's are very beneficial (interestingly you can also easily see where the analogue gain is used and where the digital gain is used).

I also believe the preceding chart illustrates that the only camera of the ones I picked only the Pentax conforms to Tony Northrup's ideas (I must confess I didn't watch the video I just read the summary provided). Clearly if we look at something like the 6D2 the ISO setting (at least the full stops) has a big difference and works as you would expect although increasing ISO by one stop only gives you about 0.8EV extra sensitivity.

The quote by Jrista provided by AlanF is 100% correct. I just thought I could expand a bit on what increasing the ISO in camera actually does. As mentioned the signal coming off the sensor is an analogue signal so it needs to be read by an analogue to digital converter (ADC) which quantizes the signal (converts an analogue voltage into a digital number between 0 and 2^14 for a 14-bit ADC). This process introduces noise. If we then amplify the signal after this (raise in post or digital ISO) we also increase the magnitude of the noise thus we don't get as much improvement as we expected. One way around this is to amplify the signal before the ADC reads it, this reduces the headroom but normally improves the signal to noise ratio so as long as you are not clipping highlights it should be an improvement.

If the noise introduced by the ADC is negligible and the ADC has sufficient bit depth we can theoretically do all amplification digitally without any adverse effects, this would be a perfect ISO-in-variant sensor. The introduction of the low noise on sensor ADC's has definitely reduced the need for pre-amplification or perhaps eliminated it completely with certain cameras. I suspect the Pentax K-1 either has such good ADC's that it doesn't benefit from them or it doesn't actually have them (I think the latter is more likely since that would explain why the 1DX MK2 has better dynamic range at ISO's over 800).
http://www.photonstophotos.net/Charts/PDR.htm#Canon EOS 1D X Mark II,Pentax K-1

Very good points, which fit in with nice discussions such as this: https://photographylife.com/iso-invariance-explained

However, a simple way of looking for isoinvariance is to look at DR vs iso charts and homing in on the regions where DR decreases linearly with iso- see:
http://www.dslrbodies.com/cameras/camera-blogs/the-d5d500-blog/iso-variance.html

"Simplest explanation: An ISO invariant sensor is going to have a dynamic range chart that has a constant slope downwards as you increase ISO values. An ISO variant camera is going to have ups and downs, sometimes only a couple of clear jumps at particular ISO values where the digital gain is changed, sometimes changes at all values, as with the D5 low ISO values." Thom Hogan

But, I would add a further mathematical criterion about the slope of the plot:
Isoinvariance is characterised by a linear plot of DR vs ln(2)iso of slope 1, (where ln(2)iso = ln to base 2 of the iso)

This criterion is obeyed by the 5DIV, 80D, 1DXII in the linear regions pretty well. The Pentax K-1 fits it close to perfect.

 

[RGF]

Difference between 2 cameras and LENS. Could the difference be the T factor on the lens. a full stop is a lot but 1/3 to 2/3 is possible!

 

[Michael Clark]

Tony Northrup is the YouTube version of Ken Rockwell. He says just enough that is actually correct to make the rest of what he says dangerous.

He's in it for the clicks (money), and he'll say whatever he thinks will get the most clicks. Everyone in any kind of media today (and these days, YouTube is about as financially viable for many folks as what we call the "traditional media") follows the model of Howard Stern: It doesn't matter if they like or hate the way you are constantly stirring the pot, as long as they are talking about you and listening/watching you.

 

[Graphic.Artifacts]

I'm no expert, but I'd be very surprised if Canon wasn't using some sort of analog gain prior to the A/D circuits at higher ISO's. I like Tony Northrup and most of what he say's I'd agree with regarding ISO. However, if analog signal amplification is being used by Canon at higher ISO's, those images will have different properties than those that are "under-exposed" and lifted in post. That being said, within my normal range for handheld shooting, I rarely worry about ISO and typically shoot M/auto ISO and sort exposure out in post. I'd certainly like to see complete ISO invariance but I think Canon has a way to go on that.

 

[AlanF]

Yet another reason how and why digital photography ruined the business... ( except for them )
A baboon can shoot now ! And why shoot it in one frame ? Let's fix it in post ! LMAO !
I miss the days when photographers actually knew their trade...
I wish digital photography would die and go away already.

It's now Planet of the Baboons, and I am happy to be one of them. And you know what, I know the tools of the digital trade, and a lifetime ago a I developed and enlarged my own negs.

 

[neuroanatomist]

...the premise is immediately and obviously faulty.

On one hand, this statement is very appropriate for pretty much any ‘technical’ video from Northrup.

On the other hand, from his perspective I suspect the real premise for all his videos is simply to generate clicks and thus income, and thus any misinformed piece of clickbait can be a success in those terms.

 

[dtaylor]

I like the videos that the Northrup's put out, but Tony does get caught up in memes at times. And ISO invariance is a meme.

1 - Even ISO invariant sensors are not ISO invariant because they typically have at least two analog gain stages. That's why fstoppers saw a difference testing arguably the highest DR / most ISO invariant sensor made. Older sensors had more stages and complicating factors (read noise). Future sensors might also have more stages and complicating factors. Incorporating the belief that you can ignore ISO will lead to mistakes.

2 - Outside of astrophotography knowing the ranges of invariance for a sensor doesn't help you at all. If you're in an AE mode are you going to constantly fight your meter with your EC dial because your sensor is ISO invariant? If you're in full manual are you going to constantly recalculate your meter readings? On mirrorless are you going to give up exposure preview? And...what happens to your back-of-camera shot review when you underexpose by 5 stops?

Saying it doesn't matter if you shoot at X and push to Y in post is the same thing as saying it doesn't matter if you go ahead and shoot at Y so that your meter, AE, EVF preview, and shot review are all in sync and working as designed. Which is what everyone does any way.

3 - ISO invariance is terribly confusing to new photographers and will lead to mistakes even if they are operating within an invariant range.

I don't know why this meme is pushed. DPReview dedicates an entire page of each review to this meme, and I swear it's just so that they can bring up the DR topic again. Now the Northrup video. I can just set my camera to ISO 640 and push to ISO 25,600 in post. Or...or...you could just set the camera to ISO 25,600 and not complicate meter readings, previews, exposure settings, and post shot reviews.

Side note: ETTR did not become a technique due to DR or ISO invariance in early sensors. It became a technique to maximize shadow tonality. Due to the nature of how pixels and ADCs work higher zones (say, zone 8) have finer tonality than lower zones (zone 2). If you shift your exposure to the right on the histogram then adjust it back down in post your shadows will have finder gradations. This remains true today even on so-called ISO invariant sensors.

I just recently watched a video by a professional architecture photographer who uses a 100mp Phase One. He brought up the Phase One's DR but mentioned he never pushes shadows any way because he blends two exposures in those situations. Why? For better tonality. I had to laugh when he said it because it seems the photographic community has forgotten this in the rush to see who can underexpose their cat by the greatest amount and still get a usable print.

 

[photo212]

Dynamic range, color range/spectrum. Gee, ISO setting is something that might actually matter.

The upper ISO setting does matter. The manufacturer decides what is the useful ISO upper limit. Sure, they could allow you to go well beyond that number, but then people would post images of this thousands of dollars of camera that look quite crappy. If the upper limit did not matter, try comparing ISO 1600 a Canon 10D versus a 80D (or whatever more modern sensor).

https://www.mattgranger.com/tutorials/item/238-the-real-downsides-to-high-iso

 

[dtaylor]

I'm no expert, but I'd be very surprised if Canon wasn't using some sort of analog gain prior to the A/D circuits at higher ISO's. I like Tony Northrup and most of what he say's I'd agree with regarding ISO. However, if analog signal amplification is being used by Canon at higher ISO's, those images will have different properties than those that are "under-exposed" and lifted in post. That being said, within my normal range for handheld shooting, I rarely worry about ISO and typically shoot M/auto ISO and sort exposure out in post. I'd certainly like to see complete ISO invariance but I think Canon has a way to go on that.

Sony sensors have two gain stages.

 

[Quarkcharmed]

The Tony's statements are oversimplified to the point where the whole video becomes just a lie. In general he's giving a lot of misleading information when it comes to the technical details, I don't think he's a good source. His channel is well presented and has many subscribers but often broadcasts technically inaccurate and misleading information.

The signal from the sensor first gets amplified in analog amplifier before it gets through the ADC. Then it becomes a digital signal and can also be amplified digitally. Analog amplification also amplifies the noise but it's more accurate than digital. So setting ISO 800 in camera is *not* the same as adding +3ev to an ISO 100 image in Lightroom.

For Canon, as far as I remember, analog amplification goes for ISO 200, 400, 800, and interim values are obtained through analog amplification + digital up/down scaling. Say ISO 320 is analog amplification to ISO 400 then digital downscaling. (The numbers above are just examples as I'm too lazy to search for more accurate figures).
Base ISO 100 is called 'base' because it doesn't get amplified, just converted to digital, thus it's the ISO with the most accurate data possible.

 

[Quarkcharmed]

I found the following link to be the most informative guide to understanding ISO as gain, with one layer drilling down into the details to be aware of the difference between upstream and downstream read noise (at least as it related to astrophotogrpahy use cases).

http://dslr-astrophotography.com/iso-dslr-astrophotography/

But Tony's video is far more entertaining with high conflict generation potential, and therefore will reach far more people.

Good one, I didn't see your post before writing my comment above. Only that 'reaching more people' and 'providing accurate information' are totally different goals and I don't think Tony pursues the latter one.

 

[hamoser]

I need to relearn basics of photography.

Maybe you do. Just don't try to learn them from Tony.

 

[Deleted member 378875]

Yet another reason how and why digital photography ruined the business... ( except for them )
A baboon can shoot now ! And why shoot it in one frame ? Let's fix it in post ! LMAO !
I miss the days when photographers actually knew their trade...
I wish digital photography would die and go away already.

Yeah just like the days of professional typists which were completely ruined by digital word processing ...

Yet another reason how and why digital photography ruined the business... ( except for them )
A baboon can shoot now ! And why shoot it in one frame ? Let's fix it in post ! LMAO !
I miss the days when photographers actually knew their trade...
I wish digital photography would die and go away already.

You ain't seen nothing yet ... Only a matter of time until lightroom is built into the camera and will automatically crop, post process and even rate your pictures completely automatically ...