Why do digital cameras need iso?

[cliveg]

Hello
I'm still a beginner in all this photography business, and am getting my head round the three components of exposure - shutter speed, aperture and iso. The first two are straight-forward, but I can't work out why a digital sensor needs iso. As it is a digital device, why can't the sensitivity always remain at the highest level for the particular sensor? Why should you need to adjust a digital sensor's sensitivity?
Anyone know a simple answer to this?
Thanks

[Donald]

Hello Clive and welcome to CiC.

This is very much a learning resource, so questions like yours are always welcome.

The best place to start understanding this is this CiC tutorial. Scroll down to the ISO Speed section. That should start you with the basics. The you can look at the list of other tutorials and start to expand your knowledge and understanding.

What you'll find, I think, is that if we just had shutter and aperture to control our exposures, we'd be very limited. Having ISO speed options as well significantly increases the number of options we have and the degree of control we can exercise.

By the way, you don't say where you are. If you wish, you can go to Edit Profile and enter your location so that it shows up alongside all your posts, just as in my details alongside this message. Then we all know where everyone is in the world.

[bambleweeney]

The lower the ISO the better the quality basically. The higher the ISO the more noise (due to signal amplification and a couple of other factors) and hence less quality. I tend to shoot always at the lowest ISO possible but I shoot primarily landscapes and often have longish exposures tripod mounted. Try a couple of shots one with the lowest and one with the highest ISO and look carefully at the results.

[tclune]

As it happens, current generation digital sensors do always shoot at the base ISO of the system, and insert a shift identifier in the raw data file to mark the gain to be applied for the ISO the user requested. But that is a very recent development. It used to be the case that the dynamic range of the sensor chips wasn't high enough to allow that sort of operation, so there was an amplifier stage that would increase the ISO of the acquired data. But now, with as much as 14 EV dynamic range in the sensor, the amplifier is no longer needed, and leaving it out not only lowers costs but also removes a major source of electronic noise in the camera circuitry.

[cliveg]

Good evening chaps, thanks very much for this. I did have a look at the tutorial, but I still scratch my head about the idea of iso being the quantity of light, and each pixel being a mini bucket that can be filled up with light.
I could understand iso in film, as higher iso and therefore faster film had a different chemical composition.
I was thinking that Digital sensors were on - off binary devices with an existing dynamic range, which seems to fit with tclune's reply.
So the idea of iso seems unnecessary in digital photography, if the sensor is a good one it should have sufficient sensitivity and the image would be adjusted by aperture, focus and shutter speed.
Hopefully this might make the basic question make more sense.
Thanks again.

PS have added my location - Worcester in England.

[Dave Humphries]

As it happens, current generation digital sensors do always shoot at the base ISO of the system, and insert a shift identifier in the raw data file to mark the gain to be applied for the ISO the user requested. But that is a very recent development. It used to be the case that the dynamic range of the sensor chips wasn't high enough to allow that sort of operation, so there was an amplifier stage that would increase the ISO of the acquired data. But now, with as much as 14 EV dynamic range in the sensor, the amplifier is no longer needed, and leaving it out not only lowers costs but also removes a major source of electronic noise in the camera circuitry.

While I can see this might well be possible on a P&S, which typically only produce 8 bit jpgs, leaving 6 bits available for iso shift (100 - 3200 or 200 - 6400 in 6 steps) as you describe Tom, I don't think it to be so for DSLRs, which still need to capture 14 bit RAW data for later PP.

That said, DSLRs do bit shift iso at the extremes of range; e.g. for the steps called "Lo" and "Hi1", "Hi2", etc. This does reduce their dynamic range though.

[MilT0s]

As it happens, current generation digital sensors do always shoot at the base ISO of the system, and insert a shift identifier in the raw data file to mark the gain to be applied for the ISO the user requested. But that is a very recent development. ...

That's very interesting Paul, didn't know about that development. Do you have any links?

In the usually configuration I think the amplifier is integrated (or close to?) to the actually photosensitive elements to provide amplification as soon the signal is captured. We can compare it to microphones that need a pre-amlifying unit before the signal is actually amplified at the main amplifier unit before reaching the speakers.

The main problem with amplifiers (all of them) is that by amplifing a signal you also amplify the noise. This way you make noise more prominent (the noise to singal-to-ratio -or SNR- drops). This "pre-amp" is a high quality unit which provides a low degree amplification that doesn't introduce more noise to the signal (at least much, unfortunately this is inevitable). That's why it is better to do this amlification stage on camera by increasing the ISO and not during post production, e.g. by using the exposure slider in lighroom.

OK what I said is not totally correct since I tried to simply things (probably with no success..)

Hope it helps,
Miltos

[tclune]

While I can see this might well be possible on a P&S, which typically only produce 8 bit jpgs, leaving 6 bits available for iso shift (100 - 3200 or 200 - 6400 in 6 steps) as you describe Tom, I don't think it to be so for DSLRs, which still need to capture 14 bit RAW data for later PP.

Actually, it is the Sony DSLR sensors (also used on Nikons) that I am aware of this being done for. The tip-off can be seen if you look at the DXO sensor measurements (choose camera sensor database; choose a brand; choose a camera; choose measurements; choose dynamic range) and you will see starting with the D7000-generation sensors, the dynamic range graph is flat with ISO. But, if you go back a ways, you will see that the graph curves down as ISO increases. The added curve is a reflection of the amplifier noise. There was a very cool demonstration when the D7000 first came out: the guy took a series of photos where he adjusted ISO, and then used the base ISO image and simply shifted the raw data to match the higher ISO. The apparently black base ISO image was indistinguishable from the properly-exposed 3200 ISO image after the shifting had been done.

[ETA: the linear drop in ISO should be 1 bit per each doubling of ISO to reflect fixed ISO, in case that was not obvious from the description.]

[orlcam88]

Good evening chaps, thanks very much for this. I did have a look at the tutorial, but I still scratch my head about the idea of iso being the quantity of light, and each pixel being a mini bucket that can be filled up with light.
I could understand iso in film, as higher iso and therefore faster film had a different chemical composition.
I was thinking that Digital sensors were on - off binary devices with an existing dynamic range, which seems to fit with tclune's reply.
So the idea of iso seems unnecessary in digital photography, if the sensor is a good one it should have sufficient sensitivity and the image would be adjusted by aperture, focus and shutter speed.
Hopefully this might make the basic question make more sense.
Thanks again.

PS have added my location - Worcester in England.

In simplicity sake, it was done to resemble film cameras. I can't fined the original link that I got this information from but basically businesses wanted to make a smooth transition and not get any resentment to new technology. As there still a requirement to allow more light into the camera, they still needed a way to lighten a photo. So they probably could have called it "light sensitivity" and work on the similar bases but they also needed a way for photographers to transition to digital with it resembling film cameras they kept ISO. I remember when the first DSLR came out how people said it wouldn't last, it looks too digital etc. So I can see the business marketing point. But not all have converted as my cousin still gives me that argument! What's funny is, that she still uses film but post her photo's on the internet!!!! duh!

Here's a link from wikipedia that gets all technical about it. http://en.wikipedia.org/wiki/ISO_speed It doesn't explain why but if you want to know what's different in ISO in film vs dlsr, this should help. mind you, it got too deep for my tiny little brain so I stopped after the smoke started coming out of my ears.

I'll keep looking for that article. It may have been a book.

[dje]

Good evening chaps, thanks very much for this. I did have a look at the tutorial, but I still scratch my head about the idea of iso being the quantity of light, and each pixel being a mini bucket that can be filled up with light.
I could understand iso in film, as higher iso and therefore faster film had a different chemical composition.
I was thinking that Digital sensors were on - off binary devices with an existing dynamic range, which seems to fit with tclune's reply.
So the idea of iso seems unnecessary in digital photography, if the sensor is a good one it should have sufficient sensitivity and the image would be adjusted by aperture, focus and shutter speed.
Hopefully this might make the basic question make more sense.
Thanks again.

PS have added my location - Worcester in England.

Hi Clive

I'll preface my comments by saying that I'm not an expert on camera sensor technology. My remarks are based on the reading I have done on a variety of technical articles (low level).

Camera sensors aren't really just "on-off" digital devices. They have a "front end" which reads the analogue light signals coming in and this produces an analogue voltage for each pixel corresponding to the light level falling on each pixel. This analogue signal is then converted to digital by an A/D converter built into the sensor chip.

This analogue front end also produces a "fixed amount" of noise. The higher the signal level (up to the saturation point of the sensor), the higher the signal to noise ratio (S/N). So if the exposure setting on the camera produces an image illumination of the sensor where the highlights are somewhat lower than the saturation level, you can amplify the output signal (before the A/D converter) to bring these highlights up to the maximum level. This is what happens with the higher ISO settings. However you also amplify the noise when doing this. The amplifier itself will also add some noise but not much.

Hi Tom
That is an interesting concept that I haven't heard of before and frankly it has me a bit puzzled. I had a look at the DXO site and the D7000 Dynmic Range curve and it seemed to have the same trend as other cameras. ie linear slope down of Dynaic Range vs ISO. Am I missing something ?

Dave

[orlcam88]

I can't find it on the internet. I'll have to check my books. btw, here's an article related to ISO and sharpness. I had noticed this also (sharper image at higher iso with low light) but thought it was just a coincidence.
http://cameradojo.com/2007/12/01/com...he-lowest-iso/

[tclune]

Hi Tom
That is an interesting concept that I haven't heard of before and frankly it has me a bit puzzled. I had a look at the DXO site and the D7000 Dynmic Range curve and it seemed to have the same trend as other cameras. ie linear slope down of Dynaic Range vs ISO. Am I missing something ?

Hi, Dave. The line will of course go down, but not curve. It is the curve that shows more information loss than would be accounted for by simple increase in ISO alone.

[AB26]

Hello
As it is a digital device, why can't the sensitivity always remain at the highest level for the particular sensor? Why should you need to adjust a digital sensor's sensitivity?
Anyone know a simple answer to this?
Thanks

Hi Clive,
Welcome to the wonderful world of Photography. It gets more demanding than a game of Gholf.

A very good and valid question. I have asked myself the same question and the answer I had for myself is as follows:

In the days of film we used “fixed” ISO film. A roll of film did not have frames with different ISO. You had to decide what you were going to shoot and put the correct ISO film in your camera. No matter what the ISO was, you were stuck with that ISO for as long as that roll of film was in your camera.
Now, if you were going to shoot low light scenes you would probably use ISO 800-1600. Those ISO ranges were also very “grainy” – in digital terms “noisy”. Some photographers would use high ISO for specific effects, especially in B&W.
If you were going to shoot a moonlit scene, ISO 800 would be just fine but comes sunrise you will run into various problems. For instance if your cameras shutter speed was limited to 1/1000 sec and the aperture on the lens was limited to F8 you would not be able to get a correct exposure in a sunlit scene. The solution, change film to ISO 100.

With the coming of the digital age it simply makes sense to have the camera’s sensitivity to light adjustable. If ISO is fixed at say 3600 the shutter speed would have to increase to say 1/16000 or even 1/32000. I think the best of our DSLR’s have a shutter speed of 1/8000sec. Imagine the cost of manufacturing a camera with a shutter speed of 1/32000? How about a lens with an aperture of say F.50 –F88 imagine the cost (and size) of that.
If the ISO was fixed at 100 you would not be able to shoot handheld low light scenes. So it makes sense to have the ISO value adjustable. Camera exposure still works in a triangle – Aperture/Shutter/ISO.

I still treat ISO like in the days of film, shoot with as low an ISO setting as possible for prevailing lighting conditions. If the light gets too low it is possible to increase ISO to be able to shoot handheld way beyond the limitations of the days of film. It is like changing a roll of film from ISO 100 to ISO 400 without the loss in unexposed frames.

Digital and in camera ISO adjustment have opened a vast expanse of increased possibilities for photographers.
Hope all this makes a little sense to you.

Challenge yourself by setting the ISO on your camera to 3200. Do not adjust it and shoot like that for a day or two, see the results.

[Colin Southern]

Put simply, without an ISO variable, an exposure of a given time for a given aperture may appear horrifically under-exposed. Granted, that even with ISO increases the sensor is still working with the same amount of light, but without increasing the ISO you'd have no way of reviewing the image until it was adjusted in post-processing.

There are also additional noise advantages to be gained by increasing the ISO due to the types of noise and the point at which they're amplified, but I won't go into that here.

[NikonFL]

Hello
I'm still a beginner in all this photography business, and am getting my head round the three components of exposure - shutter speed, aperture and iso. The first two are straight-forward, but I can't work out why a digital sensor needs iso. As it is a digital device, why can't the sensitivity always remain at the highest level for the particular sensor? Why should you need to adjust a digital sensor's sensitivity?
Anyone know a simple answer to this?
Thanks

WHAT A GREAT QUESTION!!! Did I miss something or has there not been a SIMPLE answer?

[xpatUSA]

WHAT A GREAT QUESTION!!! Did I miss something or has there not been a SIMPLE answer?

HAS THERE NOT ??? . How about these:

As it is a digital device, why can't the sensitivity always remain at the highest level for the particular sensor?

The OP talks about "it" as being the sensor. The sensitivity of the sensor actually does remain at the same "level", irrespective of any camera settings.

Why should you need to adjust a digital sensor's sensitivity?

The sensitivity of a sensor is fixed and can not be "adjusted" by the User.

. . . two SIMPLE answers, neither of which is of any use to Clive, however.

How about a SIMPLE answer for the question in the title:

Why do digital cameras need iso?

For the same reason that film cameras need different "speed" films.

[jcuknz]

It is really much simpler to use digital ISO as one used film ISO and let the geeks worry about how it all works.
The end results are quite similar with the advantage that the ISO is tied to the shot/file and not the roll of film.
I cannot see how it is any more difficult to organize a high shutter speed as AB26 writes than a slow one since once one has 'sync speed' it is simply a matter of blind width. Sync speeds have been getting faster in my lifetime and I do not doubt this will continue ....imagine a 1/20 sync speed of a camera built before electronic flash became generally available.

[Andrew1]

I'm by no means an electrical engineer so it's just my 2 cents worth.

I think each generation of any particular sensor would have a point where there is an optimum performance level. The output from those components would be engineered to work at an incident so as to maximize the reading of the input (light) while minimize the negative effects of the circuitry (attenuation and noise). The amplification or gain applied is developed with downstream requirements in mind.

In the case of photography, visible light in the THz range would take very little time to expose the sensor to enough light to produce an image. Shutter speed and ISO considerations to adequately record the slowest purple light would really be non-issues for most of us. A GAIN adjustment might be useful but full auto already manages that quite well. Uses such as Hubble taking 20 minute duration shots to see galaxies in what is black to the human eye would be an exception. If you have the cash you too can have exotic custom made cameras not made for the masses.

In the digital world shutter and ISO don't work like they did with film. A long shutter speed on film changed the reaction of silver halide producing darker particles in the film. Exposure to light and movement of the subject made physical changes in the chemical makeup of the film. Long exposures on a digital sensor result in many samplings of the output of the sensor which are stored then manipulated in the electronics and software to produce a single visible output representative of what we were used to in film. ISO is much the same thing. A single electronic input produced by the sensor and fed into circuitry and software that produces results mirroring those of film.

No shutter or ISO considerations for us would make photography in the digital world a pretty boring endeavour. Those two adjustable factors account for a majority of the creativity available to us and have been carried over from the film world for exactly those reasons.

Camera manufacturers could easily have changed the paradigms of photography as we know it. Instead of the settings we are familiar with like the shutter speeds and f-stops they could have gone with a totally new sequence of numbers. It’s the same as you editing your photos in raw. Without standards you can make your photo look however you like without considering any other photographer.

It’s just data. It only pretends to work like film for our benefit.

[xpatUSA]

Shutter speed and ISO considerations to adequately record the slowest purple light would really be non-issues for most of us.

Not sure I understood that. I thought that all light travels at the same speed ~ 300 million meters a second. I also thought that purple light has no frequency per se because it is a mixture of blue and red light.

[cliveg]

Good afternoon everyone, and thanks for having a go at explaining it!
I'm one of those who really likes to know how something works, and with that understanding can hopefully have a stab at using it.....
Having gone through the great replies here it would seem that that an analogy would be a microphone and amplifier to capture sound. If the sound is loud, sustained and clear the microphone will pick up and faithfully record it without the need for much amplification, and with only limited loss of quality through distortion. So for light if its bright and steady, the iso 'amplification' needs to be at its lowest setting to allow good picture quality with minimal 'noise'.
However if the sound is really quiet or is very short it will need amplification to catch it, but there will inevitably be a loss if quality involved (unless using very expensive equipment). Similarly, if the light is very low or is gone in a flash, the camera needs to amplify the light by means of a higher iso - but the result is 'noisier'.
Does that sound about right?
By the way - I knew it would be a good idea to join this forum!
Cheers.