ISO

[fjrabon]

Increasing the ISO makes the sensor more sensitive to light. ...........

Technically, incorrect.

Raising the ISO simply changes the algorithms used by the camera's processor on the data the sensor produces. The sensor will 'see' the exact same thing, whether the ISO is set to 50 or 512,000.

Also incorrect. ISO changes the analog gain of the amplifier. This is done on the analog side of the camera, not the digital side. There are no algorithms involved. Changing ISO is not the same in the RAW processor as just boosting up the "exposure" slider in your raw image processor.

Thanks for that. I always felt like that had to be the case, because of the issue I outlined above. But I had been told like 5-6 times what Sparky said above.

 

[unpopular]

It is important to remember that a CCD or CMOS chip is an analog device, in fact, a good portion of a digital camera is a computer controlled analog instrument. The only "digital" thing about a digital camera deals with converting that analog information into computer files which we can record and manipulate.

In fact, the Sony a700 and the a900 even incorporated analog noise reduction.

 

[480sparky41 40]

Increasing the ISO makes the sensor more sensitive to light. ...........

Technically, incorrect.

Raising the ISO simply changes the algorithms used by the camera's processor on the data the sensor produces. The sensor will 'see' the exact same thing, whether the ISO is set to 50 or 512,000.

Also incorrect. ISO changes the analog gain of the amplifier, rather than applying digital algorithms to recorded data. This is done on the analog side of the camera, not the digital side. There are no algorithms involved. Changing ISO is not the same in the RAW processor as just boosting up the "exposure" slider in your raw image processor.

You say toe-may-toe, I say tuh-mah-toe.

 

[480sparky41 40]

Given this, I've never understood why you can't change ISO in post if you shoot RAW? I mean I guess I understand that you can somewhat, but it almost seems to me like ISO should be 100% adjustable in post.

Because its about the electronics of the camera. It is what is was. You can't take a picture of a fish and change it into a dog in ACR. You can't take a picture at ISO 4000 and change it to ISO 100 in ACR.

I understand that. My point is that ISO doesn't actually change what the sensor records (or at least that's the way I interpreted what Sparky said). ISO doesnt make the sensor more sensitive, it just changes how the image was processed, which, as I've understood RAW, it doesn't make use of the processing, it simply records what the settings were. This is just like how white balance is more or less completely adjustable in post if you shoot RAW, because while the camera does note what it thought the settings were white balance wise, it doesn't change what the sensor recorded.

Shutter speed and aperture are physical characteristics of the lens, while ISO is about how the image was processed. Given that, it seems like to me you should be able to completely adjust 100% in post. ie it seems like if you shot an image at ISO 100 that was 4 stops underexposed, you should be able to bump it up to ISO 400 in post, and it would look exactly the same. I've never 100% understood why that isn't the case. Sure, we can basically do it for 2 stops, and everything mostly looks the same, but much more than that, and you see way more noise than you would if you had shot it at a higher ISO in camera.

If you shot the image at 4-stop under, you won't have any (or very very little) data recorded at many of the photoreceptors. Hence, you have little or no data to work with no matter what. You can't take a photoreceptor that recorded zero photons and magically make the data appear. Same thing for over-exposure. If the photoreceptor what overwhelmed and returns the value for pure white, you have no way of reducing that to what it truly should be.

 

[unpopular]

Technically, incorrect.

Raising the ISO simply changes the algorithms used by the camera's processor on the data the sensor produces. The sensor will 'see' the exact same thing, whether the ISO is set to 50 or 512,000.

Also incorrect. ISO changes the analog gain of the amplifier, rather than applying digital algorithms to recorded data. This is done on the analog side of the camera, not the digital side. There are no algorithms involved. Changing ISO is not the same in the RAW processor as just boosting up the "exposure" slider in your raw image processor.

You say toe-may-toe, I say tuh-mah-toe.

LOL. I suppose. But it's hard to imagine an operational amplifier as the analog computers which they are...

 

[fjrabon]

Technically, incorrect.

Raising the ISO simply changes the algorithms used by the camera's processor on the data the sensor produces. The sensor will 'see' the exact same thing, whether the ISO is set to 50 or 512,000.

Also incorrect. ISO changes the analog gain of the amplifier, rather than applying digital algorithms to recorded data. This is done on the analog side of the camera, not the digital side. There are no algorithms involved. Changing ISO is not the same in the RAW processor as just boosting up the "exposure" slider in your raw image processor.

You say toe-may-toe, I say tuh-mah-toe.

There's actually a fairly large difference though. If the method was digital, it would be completely, 100% manipulable in post. The fact that it's analog amplification means it isn't.

 

[unpopular]

There is, but Sparky was referring to the analog signal processor, not the digital signal processor which we're more familiar with.

 

[HughGuessWho99 14]

I understand that. My point is that ISO doesn't actually change what the sensor records (or at least that's the way I interpreted what Sparky said). ISO doesnt make the sensor more sensitive, it just changes how the image was processed, which, as I've understood RAW, it doesn't make use of the processing, it simply records what the settings were. This is just like how white balance is more or less completely adjustable in post if you shoot RAW, because while the camera does note what it thought the settings were white balance wise, it doesn't change what the sensor recorded.

Shutter speed and aperture are physical characteristics of the lens, while ISO is about how the image was processed. Given that, it seems like to me you should be able to completely adjust 100% in post. ie it seems like if you shot an image at ISO 100 that was 4 stops underexposed, you should be able to bump it up to ISO 400 in post, and it would look exactly the same. I've never 100% understood why that isn't the case. Sure, we can basically do it for 2 stops, and everything mostly looks the same, but much more than that, and you see way more noise than you would if you had shot it at a higher ISO in camera.

You can't change the f stop or aperture in post either. It is what it was at the time of capture. You can change the appearance of exposure in post but if you shot at 1/250 f8 ISO 400 that's what it is and will ever be. Changing a value and changing the appearance is two different things.

 

[fjrabon]

If you shot the image at 4-stop under, you won't have any (or very very little) data recorded at many of the photoreceptors. Hence, you have little or no data to work with no matter what. You can't take a photoreceptor that recorded zero photons and magically make the data appear. Same thing for over-exposure. If the photoreceptor what overwhelmed and returns the value for pure white, you have no way of reducing that to what it truly should be.

I get that. My point was, operating under your original statement that ISO is a 'digital algorithm from the camera's processor', all that would matter is what hit the sensor. so, as far as the image that hit the sensor, it wouldn't matter if it was ISO 800 or ISO 100, as long as the aperture and shutter speed were the same, the same amount of data would be recorded by the photoreceptors. ISO doesn't make more or less photons hit the sensor. As unpopular pointed out, the reason this doesn't work is that ISO comes from an amplifier, that doesn't work completely linearly and is the result of physical circuitry instead of digital processing.

For example, if you shot a scene that would be properly exposed with ISO 400, f/4 and 1/30, it would be 4 stops underexposed at ISO100, f/4 and 1/30, correct? Yet the same light would be hitting the sensor. If ISO were a result of processing, this could be done and undone, such that you could always fix ANY misses in exposure and they would look EXACTLY the same as if they had been shot at the 'correct' ISO. However, since it's the result of an imperfect physical amplifier, we can't do that without introducing a lot of noise.

 

[fjrabon]

I understand that. My point is that ISO doesn't actually change what the sensor records (or at least that's the way I interpreted what Sparky said). ISO doesnt make the sensor more sensitive, it just changes how the image was processed, which, as I've understood RAW, it doesn't make use of the processing, it simply records what the settings were. This is just like how white balance is more or less completely adjustable in post if you shoot RAW, because while the camera does note what it thought the settings were white balance wise, it doesn't change what the sensor recorded.

Shutter speed and aperture are physical characteristics of the lens, while ISO is about how the image was processed. Given that, it seems like to me you should be able to completely adjust 100% in post. ie it seems like if you shot an image at ISO 100 that was 4 stops underexposed, you should be able to bump it up to ISO 400 in post, and it would look exactly the same. I've never 100% understood why that isn't the case. Sure, we can basically do it for 2 stops, and everything mostly looks the same, but much more than that, and you see way more noise than you would if you had shot it at a higher ISO in camera.

You can't change the f stop or aperture in post either. It is what it was at the time of capture. You can change the appearance of exposure in post but if you shot at 1/250 f8 ISO 400 that's what it is and will ever be. Changing a value and changing the appearance is two different things.

Yes, I get that. My original question had to do with WHY this was the case with regards to ISO. If ISO were a result of in camera digital processing (as I've seen stated here by different people several times) then while that is the case, then that shouldn't be the case. My question with regards to that has been answered, the answer being that ISO isn't the result of digital processing by the camera, it's the result of a solid state amplifier in the camera.

 

[o hey tyler]

Increasing the ISO makes the sensor more sensitive to light. ...........

Technically, incorrect.

Raising the ISO simply changes the algorithms used by the camera's processor on the data the sensor produces. The sensor will 'see' the exact same thing, whether the ISO is set to 50 or 512,000.

Glad you decided to "correct" the post that was the most basic rundown I could type out on my phone. Too bad your explanation doesn't actually help the OP understand the function of ISO as a standalone statement.

 

[480sparky41 40]

If you shot the image at 4-stop under, you won't have any (or very very little) data recorded at many of the photoreceptors. Hence, you have little or no data to work with no matter what. You can't take a photoreceptor that recorded zero photons and magically make the data appear. Same thing for over-exposure. If the photoreceptor what overwhelmed and returns the value for pure white, you have no way of reducing that to what it truly should be.

I get that. My point was, operating under your original statement that ISO is a 'digital algorithm from the camera's processor', all that would matter is what hit the sensor. so, as far as the image that hit the sensor, it wouldn't matter if it was ISO 800 or ISO 100, as long as the aperture and shutter speed were the same, the same amount of data would be recorded by the photoreceptors. ISO doesn't make more or less photons hit the sensor. As unpopular pointed out, the reason this doesn't work is that ISO comes from an amplifier, that doesn't work completely linearly and is the result of physical circuitry instead of digital processing.

For example, if you shot a scene that would be properly exposed with ISO 400, f/4 and 1/30, it would be 4 stops underexposed at ISO100, f/4 and 1/30, correct? Yet the same light would be hitting the sensor. If ISO were a result of processing, this could be done and undone, such that you could always fix ANY misses in exposure and they would look EXACTLY the same as if they had been shot at the 'correct' ISO. However, since it's the result of an imperfect physical amplifier, we can't do that without introducing a lot of noise.

When you set the ISO in the camera, you're determining what shutter speed/aperture combination will be needed to create a 'correct' exposure. This is determined by how much light is coming from the subject. Let's say there's not much light, so we'll use your ISO 400-f/4-1/30sec. You need the large aperture and 'slower' shutter to record enough photons in the dark areas to create the data needed. If you decrease the ISO in the camera without changing the aperture/shutter speed, you'll end up with no data in some of the pixels.

How many photons strike the sensor is determined by the amount of light present in the scene, as well as how long the shutter is open and what the aperture is set for. For a given light level on a scene, only a certain number of shutter/aperture combinations will produce a 'correct' exposure. Anything outside those combination will result in either underexposure in parts of the image (no data returned, hence the pixel is pure black, or the photoreceptor is overwhelmed and returns a pure white ['blown-out'] pixel).

Changing the ISO in post will not recover that data that could have been recorded in a 'correct' exposure.


FWIW, I didn't say the algorithm used by the camera on the sensor data for ISO is digital. I merely stated that ISO is a function of the camera's internal processing, not some physical change induced on the 'sensitivity of light' of the sensor.

 

[fjrabon]

If you decrease the ISO in the camera without changing the aperture/shutter speed, you'll end up with no data in some of the pixels.

No, changing the ISO doesn't cause more or less photons to hit the sensor. Changing the ISO only causes the photons that do hit the sensor to be more or less amplified. If the photo was properly exposed at ISO 1600, you could go down th ISO100 and the same exact photons would be hitting the sensor.

The difference isn't that those photons 'aren't there', it's that they are less (or more) amplified.

 

[KmH]

Technically, incorrect.

Raising the ISO simply changes the algorithms used by the camera's processor on the data the sensor produces. The sensor will 'see' the exact same thing, whether the ISO is set to 50 or 512,000.

Also incorrect. ISO changes the analog gain of the amplifier, rather than applying digital algorithms to recorded data. This is done on the analog side of the camera, not the digital side. There are no algorithms involved. Changing ISO is not the same in the RAW processor as just boosting up the "exposure" slider in your raw image processor.

You say toe-may-toe, I say tuh-mah-toe.

Not really. You were describing software (apples), while unpopular was describing hardware (oranges).

For the OP: Visit -What is a pixel, and how does it work

 

[brandolsx]

Whoa, Getting way too heavy technical in here. I am good about figuring how a device works once I know what it does. O Hey Tyler gave me a simple easy answer regardless if it was super technically right or wrong. I am just worried about what effect it has on the photos, not which part senses, captures, or manipulates the image.