ISO discussion, digital SLR compared to film SLR

[Derrel]

Re-read the above discussion, and strive to place my comments into context. The discussion was -not- about "making pictures".

 

[Designer]

The dynamic range of modern DSLRs is astounding! I once tried to duplicate what I was seeing with my eye in my dining room. The room was lit by very muted and shady daylight entering through the glass patio doors. I wanted to capture what I saw because I liked the range of light (fairly bright fading to quite dark).

I tried several settings but was not able to get a photograph to look just like what I was seeing.

 

[Overread]

So what? Can you make great images with it or not? Isn't that what matters?

In technical discussions - no no its not what matters.
Creative ability is limited by the creator and also by the tool they use. However when you are comparing tools against other tools you have to boil things down to technical differences. To actual boring dull maths and stats and to things that are really boring photos that look at things purely technically.

It's not art its not creative its a technical discussion.

If you try to bring "Oh but I made this great photo with it" then its not really a valid argument. Because you can make great photos with mobile camera phones; with point and shoots; with film; with digital; with glass plate. So it doesn't really give people much to measure and compare with other than the fact that someone made a great picture of something with it.

 

[fmw]

Re-read the above discussion, and strive to place my comments into context. The discussion was -not- about "making pictures".

My apologies.

 

[Gary A.]

So what? Can you make great images with it or not? Isn't that what matters?

In technical discussions - no no its not what matters.
Creative ability is limited by the creator and also by the tool they use. However when you are comparing tools against other tools you have to boil things down to technical differences. To actual boring dull maths and stats and to things that are really boring photos that look at things purely technically.

It's not art its not creative its a technical discussion.

If you try to bring "Oh but I made this great photo with it" then its not really a valid argument. Because you can make great photos with mobile camera phones; with point and shoots; with film; with digital; with glass plate. So it doesn't really give people much to measure and compare with other than the fact that someone made a great picture of something with it.

Significance is the odd element of all this tech stuff. If I was purchasing a camera ... sure I'd want a camera with the greatest DR ... but for what I shoot ... I've never had a photo fail because of lack of DR ... no one ever told me "... that shot could have been successful if it had two more stops of dynamic range...". Often I will increase the contrast and darken the shadows to eliminate distracting details and to add more drama to the image.

I've never shot astro and maybe DR is vitally important in shooting the heavens ... I mostly shoot people ... with people, the impact of the moment is vital not the DR.

Tech and specs are important and should be discussed, but we all need to realize that the bottom line is the final image. The photographer needs to know and understand a minimal amount of tech and specs in order to consistently capture the exceptional image. But also remember that tech and specs are merely the means towards the end.

 

[petrochemist]

the trouble with pedantry is that it turns into an infinite regress of "well no, actually"

for example, "well no, Ysarex, decreased exposure doesn't cause more noise, it causes a lower s/n ratio by reducing the numerator"

so what we actually do in practical terms is work with a simplified mental model that's good enough. rather than pedantically insisting that the sensor is the array of sensels, the bayer array, and the support circuits but not the amplifiers then we get to do the dramatic bit about how changing the ISO doesn't change the sensor's sensitivity, and then the noobs look on in confused wonder and then we can wander off into the weeds mixing up noise and signal-to-noise ratio, but nobody will notice because so so so very many words

if instead you include the amplifiers that implement ISO into what you mean by the word sensor, which is a perfectly reasonable thing to do (I am an ex-systems guy) then by god changing the ISO setting does indeed make your sensor (or "sensor system" if you prefer) into amore sensitive, albeit noisier, sensor (-system)

No. If sensitivity were increased then additional data would be recorded. That doesn't happen. Amplifying existing data doesn't record more data. Raising ISO on a digital camera does not increase light sensitivity -- it doesn't allow the "system" to record data deeper into the shadows. The trouble with oversimplification is that it can encourage incorrect assumptions about cause and effect and false assumptions about what's possible. The mental model that the sun revolves around the earth worked for millions of people for a long time. I think you can still join in fact: The Flat Earth Society

Joe

But it DOES increase the information.
Considering a low light exposure, such as stars in the night sky, with the ISO low the result of the A/D conversion for the brightest pixel could easily be much lower than 1/10th of full scale.
When the signal is amplified 8 fold (3 stops higher ISO) before the A/D the digital output of a hypothetical brightest pixel which was 10% maximum would now be ~80% of maximum.
Meanwhile the luminosity recorded by many dark pixels that may have a digital value of only 0-5 (out of 256 for 8-bit jpegs or more for RAW) would now be spread over the range 0-15 the extra data here could be very significant, or could only be down to shot noise from the random nature of individual photons.
Each pixel of the intermediate light levels will also be digitalized over a 3x larger number of values, shot noise will not be significant for these so meaningful extra data is generated.

If the brightness of the shot is adjusted in post processing (after A/D conversion) then no extra data is produced the existing values are just spread further apart, but this is not how cameras usually work!

 

[Ysarex]

the trouble with pedantry is that it turns into an infinite regress of "well no, actually"

for example, "well no, Ysarex, decreased exposure doesn't cause more noise, it causes a lower s/n ratio by reducing the numerator"

so what we actually do in practical terms is work with a simplified mental model that's good enough. rather than pedantically insisting that the sensor is the array of sensels, the bayer array, and the support circuits but not the amplifiers then we get to do the dramatic bit about how changing the ISO doesn't change the sensor's sensitivity, and then the noobs look on in confused wonder and then we can wander off into the weeds mixing up noise and signal-to-noise ratio, but nobody will notice because so so so very many words

if instead you include the amplifiers that implement ISO into what you mean by the word sensor, which is a perfectly reasonable thing to do (I am an ex-systems guy) then by god changing the ISO setting does indeed make your sensor (or "sensor system" if you prefer) into amore sensitive, albeit noisier, sensor (-system)

No. If sensitivity were increased then additional data would be recorded. That doesn't happen. Amplifying existing data doesn't record more data. Raising ISO on a digital camera does not increase light sensitivity -- it doesn't allow the "system" to record data deeper into the shadows. The trouble with oversimplification is that it can encourage incorrect assumptions about cause and effect and false assumptions about what's possible. The mental model that the sun revolves around the earth worked for millions of people for a long time. I think you can still join in fact: The Flat Earth Society

Joe

But it DOES increase the information.
Considering a low light exposure, such as stars in the night sky, with the ISO low the result of the A/D conversion for the brightest pixel could easily be much lower than 1/10th of full scale.
When the signal is amplified 8 fold (3 stops higher ISO) before the A/D the digital output of a hypothetical brightest pixel which was 10% maximum would now be ~80% of maximum.
Meanwhile the luminosity recorded by many dark pixels that may have a digital value of only 0-5 (out of 256 for 8-bit jpegs or more for RAW) would now be spread over the range 0-15 the extra data here could be very significant, or could only be down to shot noise from the random nature of individual photons.
Each pixel of the intermediate light levels will also be digitalized over a 3x larger number of values, shot noise will not be significant for these so meaningful extra data is generated.

If the brightness of the shot is adjusted in post processing (after A/D conversion) then no extra data is produced the existing values are just spread further apart, but this is not how cameras usually work!

What you are describing is a post processing procedure that occurs after the exposure of the sensor. You can't increase the data the sensor has recorded once the exposure has ended. What you can do, which you describe, is improve the efficiency with which you retain the data recorded -- in other words do a better job avoiding data loss through the system.

The issue in question here is: does raising the ISO increase the sensitivity of the sensor. If it did that then you would be able to raise the ISO and expect the camera to reach deeper into the shadows recording more data. That doesn't happen. What does happen in most cameras is what you described, the ISO gain pre-ADC prevents loss that would otherwise occur in an inefficient system. That's very different from increasing the data recorded.

Work with a functionally ISO invariant system and you can easily see the above. Here's an example:

Exact same exposure in both with the only difference that the ISO was 200 in one exposure and 3200 in the other. Above are the JPEGs created by the camera.

The Fuji X-E2 is functionally ISO invariant. If raising the ISO made the sensor more light sensitive the shadow detail in the above images would bear that out. It bears out the opposite. Raising the ISO to 3200 provided no benefit in increased shadow detail. The light sensitivity of the sensor is unaffected. With an ISO invariant camera the ISO analog gain adds no value. That's not the case with most cameras including Fuji's new X-Trans III sensor cameras. With those cameras the process you describe is valuable but it's not increasing the light sensitivity of the sensor and allowing the sensor to record more data. It's an engineering difference that does a better job retaining the data recorded.

Joe

 

[Overread]

You know I think we are at the point were it might be wise to call it quits. not only are we trying to cobble together how sensors and ISO works but we are trying to do it for different generations and types of sensor; with resulting arguments and differences of opinions.

 

[petrochemist]

With those cameras the process you describe is valuable but it's not increasing the light sensitivity of the sensor and allowing the sensor to record more data. It's an engineering difference that does a better job retaining the data recorded.

Joe

I don't know about ISO invariant cameras, I don't have one & don't expect to in the next few years.

For other cameras (still the majority unless I'm much mistaken), the data recorded is only the digital output from the A/D, so it's changing the data recorded to show a different range. It has no effect on retaining the data recorded at all that's down to systems after the A/D.

I'll try to take on @Overread's wisdom & leave it there.

 

[Derrel]

THE BIGGEST difference between ISO in film and in modern d-slrs is the ability to change the ISO value now, from shot to shot, as-needed. No more "36 frames stuck at ISO 64" for super-quality color slides...no more being stuck at ISO 400 pushed two stops for detail-free shadows with pushed Tri-X B&W film...nope! With digital, ISO speeds can be changed, and exposures accordingly changed, as-needed or as-desired. THAT is the single, biuggest difference.

As far as equivalency: a photographer friend and I used his Canon EOS 10D (ten-Dee!) back in the day, with its 6 megapixel sensor: we felt that it had finer grain,and more detail than ISO 100 color negative film had.

 

[Ysarex]

With those cameras the process you describe is valuable but it's not increasing the light sensitivity of the sensor and allowing the sensor to record more data. It's an engineering difference that does a better job retaining the data recorded.

Joe

I don't know about ISO invariant cameras, I don't have one & don't expect to in the next few years.

For other cameras (still the majority unless I'm much mistaken), the data recorded is only the digital output from the A/D, so it's changing the data recorded to show a different range. It has no effect on retaining the data recorded at all that's down to systems after the A/D.

I'll try to take on @Overread's wisdom & leave it there.

After exposure the data in the sensor wells is read -- it is fundamentally an electron count. That is the data recorded. That's what the sensor does -- it records data as an electron count. Reading that data is the beginning of processing that data. Amplified or not the read data is then converted into numerical form in the ADC. Both the analog gain if applied and the ADC conversion are applied to the recorded data and constitute post processing which cannot alter the original electron count. Post processing can do a more or less efficient job with the data. Changing ISO on a digital camera does not/can not change the data the sensor is able to record. It can change how the data is post processed.

Joe