What's up with the ISO values on phones?

[theworstphotographerontheinternet]

Why do phones have such idiosyncratic ISO values? For example, my phone has settings like ISO 472, or 1024, which you don't find on "normal" cameras. I'm guessing it's not just about the sensor size, as I've used plenty of small sensor cameras which don't use these kind of ISO values. So what gives?

 

[Bob A L]

When that standard was written it might well have been as precise as they could measure at that time. Technology happens - lots of things once thought impossible are shown to be possible every day in today's world. Those of us that are old enough see impossible things (based on what we were taught when young) happen every day.

 

[Bob A L]

Way back when there was no iso 101, but like many standards from the past that has changed - now we have the capability to have iso 101 for today's standards. In the past automobiles only had one gear transmissions, then 2, then 3, now, limitless.

 

[Brian Kimball]

When that standard was written it might well have been as precise as they could measure at that time. Technology happens - lots of things once thought impossible are shown to be possible every day in today's world. Those of us that are old enough see impossible things (based on what we were taught when young) happen every day.

Well the ISO standard for digital still cameras that I keep referencing was last updated just a couple years ago in 2019. It's a modern standard.

And as others have pointed out, the difference between ISO 100 and 101 is meaningless and very likely not noticeable to human eyes.

So I wouldn't wrap this up as a story of technology marching on past an old standard and making the impossible possible just yet. My bet is still on an engineering team that just wasn't paying close enough attention to the standard.

 

[Brian Kimball]

Way back when there was no iso 101, but like many standards from the past that has changed - now we have the capability to have iso 101 for today's standards. In the past automobiles only had one gear transmissions, then 2, then 3, now, limitless.

What's this "way back" and "standards from the past" business? The modern ISO standard for digital cameras was last updated in 2019.

Just in case you're thinking of old school ISO film speed... ignore that. That's an entirely different standard. The separate and distinct standard for digital cameras was written intentionally to use the same terminology even though it describes properties of entirely different mediums.

 

[Bob A L]

Updated? or reprinted from original issue? What did it say before 2019?

 

[Brian Kimball]

Updated? or reprinted from original issue? What did it say before 2019?

Sounds like you might benefit from buying a copy or finding a local library that has access to the ISO standards. This is the one you're looking for:

https://www.iso.org/standard/73758.html

In brief, the 2019 version lists at least 7 changes from the previous 2006 version in its Foreword.

Beyond that, sorry, I can't sit here and paraphrase from the document all day. If you missed my original explanation for why ISO 101 and other weird ISO values are undefined, see my other post in this discussion:

https://www.dpreview.com/forums/post/65408781

 

[Phil A Martin]

Why do phones have such idiosyncratic ISO values? For example, my phone has settings like ISO 472, or 1024, which you don't find on "normal" cameras. I'm guessing it's not just about the sensor size, as I've used plenty of small sensor cameras which don't use these kind of ISO values. So what gives?

Ok maybe not ISO but years ago the humble Pentax ME Super had infinitely variable shutter speed when set to auto exposure.

--
http://instagram.com/thoughtful_joe

 

[J A C S]

dedicated camera photographers are obsessed with defined "stops"

phones dont have to worry about that, they just use what works since they are much more limited

The OP question made very good sense, and dismissing it as an obsession with "stops" is not an answer.

Typical ISOs in digital cameras are "hardwired". Full stops or 1/3 stops, or whatever - they are preset by the manufacturer and "baked in". Most of the time, they are not just rescaling of the numbers - the whole circuit works in a specific way for each ISO. When they are just rescaling - you can have every ISO you want. This happens usually for some intermediate ISOs or very high ones only.

So it would be interesting to know which are the "hardwired" ISOs in every particular phone, and which are not.

 

[Brian Kimball]

Why do phones have such idiosyncratic ISO values? For example, my phone has settings like ISO 472, or 1024, which you don't find on "normal" cameras. I'm guessing it's not just about the sensor size, as I've used plenty of small sensor cameras which don't use these kind of ISO values. So what gives?

Ok maybe not ISO but years ago the humble Pentax ME Super had infinitely variable shutter speed when set to auto exposure.

Indeed, I believe most cell phones have such nearly infinitely variable shutter speeds to compensate for their fixed wide-open apertures. Of course they also don't have physical shutters, so that helps.

 

[robert1955]

dedicated camera photographers are obsessed with defined "stops"

phones dont have to worry about that, they just use what works since they are much more limited

The OP question made very good sense, and dismissing it as an obsession with "stops" is not an answer.

Typical ISOs in digital cameras are "hardwired". Full stops or 1/3 stops, or whatever - they are preset by the manufacturer and "baked in". Most of the time, they are not just rescaling of the numbers - the whole circuit works in a specific way for each ISO. When they are just rescaling - you can have every ISO you want. This happens usually for some intermediate ISOs or very high ones only.

So it would be interesting to know which are the "hardwired" ISOs in every particular phone, and which are not.

Possibly none is 'hardwired' and all are digitally scaled

 

[Brian Kimball]

dedicated camera photographers are obsessed with defined "stops"

phones dont have to worry about that, they just use what works since they are much more limited

The OP question made very good sense, and dismissing it as an obsession with "stops" is not an answer.

Typical ISOs in digital cameras are "hardwired". Full stops or 1/3 stops, or whatever - they are preset by the manufacturer and "baked in". Most of the time, they are not just rescaling of the numbers - the whole circuit works in a specific way for each ISO. When they are just rescaling - you can have every ISO you want. This happens usually for some intermediate ISOs or very high ones only.

So it would be interesting to know which are the "hardwired" ISOs in every particular phone, and which are not.

Possibly none is 'hardwired' and all are digitally scaled

The few phones on photonstophotos.net don't show any scaling found:

Photographic Dynamic Range versus ISO Setting

 

[john Clinch]

Why do phones have such idiosyncratic ISO values? For example, my phone has settings like ISO 472, or 1024, which you don't find on "normal" cameras. I'm guessing it's not just about the sensor size, as I've used plenty of small sensor cameras which don't use these kind of ISO values. So what gives?

Those particular values of 472 and 1024, if actually used by a phone's camera, are made up numbers that are undefined by the ISO standard.

To measure a digital camera's ISO one must first measure or otherwise determine a certain number called an exposure index that meets certain photographic and technical criteria.

To convert that exposure index to an ISO value the standard merely supplies a long table of ranges like this (grossly simplified):


(the actual standard seems to provide ISO values in 1/3 stop increments. the tables are quite long)

Note this table leaves no room for intermediate ISO values not shown in the standard.

So yeah, if we want to be pedantic about it, there's no such thing as "ISO 472" or "ISO 1024".

DXOmark is full of intermediate iso values. Are they using there own different systems. From memory they are. Hence their graph of their iso against the manufacturers





--

John Clinch

Explore John Clinch’s 1,859 photos on Flickr!

www.flickr.com

Flickriver: John Clinch's most interesting photos

Flickriver - view images as a 'river of photos' and more...

www.flickriver.com

 

[Brian Kimball]

Why do phones have such idiosyncratic ISO values? For example, my phone has settings like ISO 472, or 1024, which you don't find on "normal" cameras. I'm guessing it's not just about the sensor size, as I've used plenty of small sensor cameras which don't use these kind of ISO values. So what gives?

Those particular values of 472 and 1024, if actually used by a phone's camera, are made up numbers that are undefined by the ISO standard.

To measure a digital camera's ISO one must first measure or otherwise determine a certain number called an exposure index that meets certain photographic and technical criteria.

To convert that exposure index to an ISO value the standard merely supplies a long table of ranges like this (grossly simplified):


(the actual standard seems to provide ISO values in 1/3 stop increments. the tables are quite long)

Note this table leaves no room for intermediate ISO values not shown in the standard.

So yeah, if we want to be pedantic about it, there's no such thing as "ISO 472" or "ISO 1024".

DXOmark is full of intermediate iso values. Are they using there own different systems. From memory they are. Hence their graph of their iso against the manufacturers

Yup, and Bill Claff also uses calculated intermediate ISO values on photonstophotos.net.

I'm not saying they're conceptually impossible or useless.

I'm just saying it's interesting that the standard doesn't leave room for them as official ISO values. If you look at ISO 101 and say "Huh, that's weird. Hey Brian, what's the Irei or Isos or Isat range of values in the standard that would lead to a camera setting having a reported ISO of 101?" my answer will be "none, that ISO value doesn't exist, because the entire range of EI values are already covered by the official 3rd-stop ISO increments listed in the standard."

It's just a quirk of the standard that I find interesting. It's also interesting to see so many people push back on it!

 

[mamallama]

Sometimes the ISO from my phone is weird.

For example the EXIF file shows an ISO of 101, why 101 rather than 100?

That's what the algorithm determined. The algorithm is not confined to discrete steps.

The way they implemented their algorithm is not confined to discrete steps, but the ISO standard confines ISO Speed/SOS/REI to discrete and well-defined values, so the cell phone company(ies) are certainly getting a little sloppy here.

Sloppy or refined? Where does it say ISO speeds are confined?

In the actual standard the values are very clearly confined by their definitions. See my post here:

https://www.dpreview.com/forums/post/65408781

Sorry, I had to paraphrase the standard because I'm not comfortable directly quoting from it or posting screenshots of it.

Ideally they could have kept their continuous function but rounded it in EXIF to the nearest actually defined ISO value.

Why?

Because there's no such thing as ISO 101 (or the other ISOs mentioned by OP) according to the standard.

So my yardstick is marked in 1/8 inch increments means there is no such thing as 1.34678 inch. i got it.

 

[Brian Kimball]

Sometimes the ISO from my phone is weird.

For example the EXIF file shows an ISO of 101, why 101 rather than 100?

That's what the algorithm determined. The algorithm is not confined to discrete steps.

The way they implemented their algorithm is not confined to discrete steps, but the ISO standard confines ISO Speed/SOS/REI to discrete and well-defined values, so the cell phone company(ies) are certainly getting a little sloppy here.

Sloppy or refined? Where does it say ISO speeds are confined?

In the actual standard the values are very clearly confined by their definitions. See my post here:

https://www.dpreview.com/forums/post/65408781

Sorry, I had to paraphrase the standard because I'm not comfortable directly quoting from it or posting screenshots of it.

Ideally they could have kept their continuous function but rounded it in EXIF to the nearest actually defined ISO value.

Why?

Because there's no such thing as ISO 101 (or the other ISOs mentioned by OP) according to the standard.

So my yardstick is marked in 1/8 inch increments means there is no such thing as 1.34678 inch. i got it.

That's a pretty poor analogy that doesn't at all apply to this situation.

Please read (or re-read) this: https://www.dpreview.com/forums/post/65408781

The standard is pretty explicit that, for example, all Isos measurements between 89.09 and 112.2 give a reported ISO value of 100, while all Isos measurements between 112.2 and 141.4 give a reported ISO value of 125.

One could imagine a fantasy world where instead of a stepwise lookup-table the ISO standard provided a continuous function that took the requisite measured exposure index as input and then output a positive real number for the reported ISO.

But that's just a fantasy. Such a thing does not exist in the standard. All we have is a stepwise lookup table that explicitly precludes the possibility of "ISO 101" existing.

That's not to say we all don't know what "ISO 101" or "ISO 472" or "ISO 1024" are implying. I'm just pointing out that it's interesting that those aren't valid ISOs as defined by the standard.

BTW, have you read the standard?

 

[Kaspah]

...... My bet is still on an engineering team that just wasn't paying close enough attention to the standard.

Yep, that is exactly what it is.

Reading through various threads here on DPR I often see comments saying that even modern "real" camera manufacturers implement ISO differently to some extent in their cameras.

 

[Brian Kimball]

...... My bet is still on an engineering team that just wasn't paying close enough attention to the standard.

Yep, that is exactly what it is.

Reading through various threads here on DPR I often see comments saying that even modern "real" camera manufacturers implement ISO differently to some extent in their cameras.

I know what you're saying and I don't mean to correct you, but just to be clear to other readers: the ISO standard for digital cameras does not specify any implementation details. It truly doesn't care how a camera implements its ISO settings. (It does seem to care how a camera's ISO values are reported, the subject of most of this discussion).

Interestingly, note I said camera, not sensor. ISO is a measured value of a camera's image output, not just its sensor's output. So for example it would be totally legit for all ISO lightening to happen in the JPG pipeline as opposed to much of it happening on the sensor and other hardware (like in most real-world digital cameras). The ISO standard truly doesn't care.

It's just a black box that takes a measured amount of light at the focal plane, matches that amount of light up with certain characteristics of a final image out of the camera (JPG/TIFF/etc, not raw), and then assigns that relationship an ISO value pulled from a long table of 1/3rd stop values (that doesn't include 101! ).

 

[mamallama]

Sometimes the ISO from my phone is weird.

For example the EXIF file shows an ISO of 101, why 101 rather than 100?

That's what the algorithm determined. The algorithm is not confined to discrete steps.

The way they implemented their algorithm is not confined to discrete steps, but the ISO standard confines ISO Speed/SOS/REI to discrete and well-defined values, so the cell phone company(ies) are certainly getting a little sloppy here.

Sloppy or refined? Where does it say ISO speeds are confined?

In the actual standard the values are very clearly confined by their definitions. See my post here:

https://www.dpreview.com/forums/post/65408781

Sorry, I had to paraphrase the standard because I'm not comfortable directly quoting from it or posting screenshots of it.

Ideally they could have kept their continuous function but rounded it in EXIF to the nearest actually defined ISO value.

Why?

Because there's no such thing as ISO 101 (or the other ISOs mentioned by OP) according to the standard.

So my yardstick is marked in 1/8 inch increments means there is no such thing as 1.34678 inch. i got it.

That's a pretty poor analogy that doesn't at all apply to this situation.

Please read (or re-read) this: https://www.dpreview.com/forums/post/65408781

The standard is pretty explicit that, for example, all Isos measurements between 89.09 and 112.2 give a reported ISO value of 100, while all Isos measurements between 112.2 and 141.4 give a reported ISO value of 125.

One could imagine a fantasy world where instead of a stepwise lookup-table the ISO standard provided a continuous function that took the requisite measured exposure index as input and then output a positive real number for the reported ISO.

But that's just a fantasy. Such a thing does not exist in the standard. All we have is a stepwise lookup table that explicitly precludes the possibility of "ISO 101" existing.

That's not to say we all don't know what "ISO 101" or "ISO 472" or "ISO 1024" are implying. I'm just pointing out that it's interesting that those aren't valid ISOs as defined by the standard.

That doesn’t mean they don’t exist.

BTW, have you read the standard?

How they are reported does not mean nothing exists between the reported values. Sort of like rounding.

The fact that smartphones have so much more processing power they can work with more precision and don’t need to round off so roughly.

 

[MrBrightSide]

This has been a very educational back and forth, strong work by all.

But I was shocked to learn how much rounding is allowed under the standard because some Sekonic meters will measure in tenths of a stop, now I wonder if that is useless precision given the vagaries of ISO implementation.

Sometimes the ISO from my phone is weird.

For example the EXIF file shows an ISO of 101, why 101 rather than 100?

That's what the algorithm determined. The algorithm is not confined to discrete steps.

The way they implemented their algorithm is not confined to discrete steps, but the ISO standard confines ISO Speed/SOS/REI to discrete and well-defined values, so the cell phone company(ies) are certainly getting a little sloppy here.

Sloppy or refined? Where does it say ISO speeds are confined?

In the actual standard the values are very clearly confined by their definitions. See my post here:

https://www.dpreview.com/forums/post/65408781

Sorry, I had to paraphrase the standard because I'm not comfortable directly quoting from it or posting screenshots of it.

Ideally they could have kept their continuous function but rounded it in EXIF to the nearest actually defined ISO value.

Why?

Because there's no such thing as ISO 101 (or the other ISOs mentioned by OP) according to the standard.

So my yardstick is marked in 1/8 inch increments means there is no such thing as 1.34678 inch. i got it.

That's a pretty poor analogy that doesn't at all apply to this situation.

Please read (or re-read) this: https://www.dpreview.com/forums/post/65408781

The standard is pretty explicit that, for example, all Isos measurements between 89.09 and 112.2 give a reported ISO value of 100, while all Isos measurements between 112.2 and 141.4 give a reported ISO value of 125.

One could imagine a fantasy world where instead of a stepwise lookup-table the ISO standard provided a continuous function that took the requisite measured exposure index as input and then output a positive real number for the reported ISO.

But that's just a fantasy. Such a thing does not exist in the standard. All we have is a stepwise lookup table that explicitly precludes the possibility of "ISO 101" existing.

That's not to say we all don't know what "ISO 101" or "ISO 472" or "ISO 1024" are implying. I'm just pointing out that it's interesting that those aren't valid ISOs as defined by the standard.

That doesn’t mean they don’t exist.

BTW, have you read the standard?

How they are reported does not mean nothing exists between the reported values. Sort of like rounding.

The fact that smartphones have so much more processing power they can work with more precision and don’t need to round off so roughly.

 

[mcsk]

Why do phones have such idiosyncratic ISO values? For example, my phone has settings like ISO 472, or 1024, which you don't find on "normal" cameras. I'm guessing it's not just about the sensor size, as I've used plenty of small sensor cameras which don't use these kind of ISO values. So what gives?

Those particular values of 472 and 1024, if actually used by a phone's camera, are made up numbers that are undefined by the ISO standard.

To measure a digital camera's ISO one must first measure or otherwise determine a certain number called an exposure index that meets certain photographic and technical criteria.

To convert that exposure index to an ISO value the standard merely supplies a long table of ranges like this (grossly simplified):


(the actual standard seems to provide ISO values in 1/3 stop increments. the tables are quite long)

Note this table leaves no room for intermediate ISO values not shown in the standard.

So yeah, if we want to be pedantic about it, there's no such thing as "ISO 472" or "ISO 1024".

DXOmark is full of intermediate iso values. Are they using there own different systems. From memory they are. Hence their graph of their iso against the manufacturers

Yup, and Bill Claff also uses calculated intermediate ISO values on photonstophotos.net.

I'm not saying they're conceptually impossible or useless.

I'm just saying it's interesting that the standard doesn't leave room for them as official ISO values. If you look at ISO 101 and say "Huh, that's weird. Hey Brian, what's the Irei or Isos or Isat range of values in the standard that would lead to a camera setting having a reported ISO of 101?" my answer will be "none, that ISO value doesn't exist, because the entire range of EI values are already covered by the official 3rd-stop ISO increments listed in the standard."

It's just a quirk of the standard that I find interesting. It's also interesting to see so many people push back on it!

Because your posts read less as simply pointing out that imprecision is built into the standard and more that these are the specific lawful values handed down by the gods and anything else is an unfathomable mystery. Being obtuse and deliberately misunderstanding things to show off that you read a technical document doesn't really add anything.