Exposure question

[Suede]

I am not clear on this . . .

In a situation where a + one-stop exposure compensation needs to be dialled in, would the end result be the same if I increased exposure by one stop by either opening the lens or slowing the speed by like amount? I am using a manual-only camera (M6) which has no exposure compensation facility.

Thanks in advance for any inputs.

Happy New Year 2015 to all.

~ Pritam

[Dave Humphries]

Hi Pritam,

In a word: Yes.

Or raise the iso one stop, assuming it is a digital camera.

Happy New Year to you and yours, Dave

[Suede]

Thank you, Dave, for the quick reply.

In that case, would I be going too far if I said that the exposure compensation button/dial/switch on cameras, digital or film, is a feature which can be effectually be done away with?

[Dave Humphries]

In that case, would I be going too far if I said that the exposure compensation button/dial/switch on cameras, digital or film, is a feature which can be effectually be done away with?

If you're shooting Manual only (and understand what you're doing), absolutely.

However, the EC button/feature is of far greater convenience when shooting full or semi Auto exposure modes. I use it a lot when not shooting Manual exposure.

It could also be stated that when relating EXIF data here at CiC, it allows less experienced shooters to see what offset needed to be applied to the metered exposure to achieve the result, whereas pure EXIF data in manual doesn't provide that information, unless the photographer goes out of their way to provide it as part of a 'mini-tutorial'. Although having thought about it, that applies to the photographer themselves, when reviewing their own shots - unless they have been manually 'bracketing', in which case they can glean it from the other exposures.

[Mike Buckley]

would I be going too far if I said that the exposure compensation button/dial/switch on cameras, digital or film, is a feature which can be effectually be done away with?

To add to Dave's comment, no, you wouldn't be going too far insofar as being accurate is concerned. However, you could say the same thing about such basic capabilities as auto focus and the light meter built into the camera. Those tools, just like the exposure compensation button, are very handy. They make photography more enjoyable in many situations than if they weren't included.

[xpatUSA]

In that case, would I be going too far if I said that the exposure compensation button/dial/switch on cameras, digital or film, is a feature which can be effectually be done away with?

Going one step further but only with reference to my early Sigma cameras, the ISO selection can be ignored too, because they are true ISO-less cameras. A true ISO-less camera does no analog amplification in-camera. In other words, the captured raw data is not changed by the ISO selection. So, there is absolutely no difference between the the two following actions (in terms of raw data), all other things being equal:

Shoot at ISO 400 and expose for 0 EV.
Shoot at ISO 100 and expose for -2 EV.

The ISO setting is passed to the raw converter as meta-data and an appropriate multiplication is done during conversion.

Nice and simple, just the way I like it . .

[Suede]

Thanks again for putting my mind at rest. Yes, I feel reassured now about "understanding what I am doing" with the manual controls.

I am thoroughly enjoying using the M6, it forces me to slow down and think before activating the shutter release. My joy with digital (D700) is once again on the wane. I don't develop or print at home but I can see it as a distinctly possible 2015 project going on the anvil. . . and seeking out a reasonable scanner. A whole new can of worms!

[DanK]

I am not clear on this . . .

In a situation where a + one-stop exposure compensation needs to be dialled in, would the end result be the same if I increased exposure by one stop by either opening the lens or slowing the speed by like amount? I am using a manual-only camera (M6) which has no exposure compensation facility.

Thanks in advance for any inputs.

Happy New Year 2015 to all.

~ Pritam

Just to add to Dave's good answer: dialing in an EV correction is simply telling the camera to open the lens or slow the speed for you, rather than letting you do it yourself. Which it will do depends on how you have the camera set.

[Mike Buckley]

dialing in an EV correction is simply telling the camera to open the lens or slow the speed for you

Or, depending on the situation, possibly changing the ISO value.

[Suede]

Thanks Dave, Mike, Dan and xpatUSA for your insightful contributions to dispel the fog clouding that fundamental question in my mind. Much appreciated.

I am consciously trying to simplify my picture-taking experience. Using a single film body, the ISO cannot be changed for as long as the roll of film is in the camera. It's restraining but strangely liberating as well.

[TonyW]

Going one step further but only with reference to my early Sigma cameras, the ISO selection can be ignored too, because they are true ISO-less cameras. A true ISO-less camera does no analog amplification in-camera. In other words, the captured raw data is not changed by the ISO selection. So, there is absolutely no difference between the the two following actions (in terms of raw data), all other things being equal:

Shoot at ISO 400 and expose for 0 EV.
Shoot at ISO 100 and expose for -2 EV.

The ISO setting is passed to the raw converter as meta-data and an appropriate multiplication is done during conversion.

Nice and simple, just the way I like it . .

That does sound nice and simple but it raises a few of questions in my mind.

Firstly why don't all cameras use that? I suppose the answer to that question might be that some people would want a lot of processing done in the camera if they want a jpeg output from the camera. Also, we would generally want to be able to review the shot on the screen in the camera.

Secondly, in that case why do you need the ISO value in the camera at all? This could be set automatically during the raw conversion process provided the user can override the choice (as we often do now in tweaking the exposure during the conversion).

Thirdly, are there any implications about the bit-depth that would be needed?

[xpatUSA]

That does sound nice and simple but it raises a few of questions in my mind.

Firstly why don't all cameras use [ISO-less]?

There are pros and cons - too many to discuss here. I'll guess it was easier for Sigma at the time because, with the Foveon sensor, each camera has 3 ADC's - not just one like ordinary cameras. I've read that some Bayer cameras have amplification up to a certain ISO but don't increase the gain after that. All rather complex, I'm afraid.

I suppose the answer to that question might be that some people would want a lot of processing done in the camera if they want a jpeg output from the camera. Also, we would generally want to be able to review the shot on the screen in the camera.

Yes, early Sigmas had that facility: a thumbnail and a quarter size JPEG was embedded in the RAW file for the very purpose of LCD review images. JPEG outputs started with the SD14 which was the first to also embed a full-size JPEG in the raw file.

Secondly, in that case why do you need the ISO value in the camera at all? This could be set automatically during the raw conversion process provided the user can override the choice (as we often do now in tweaking the exposure during the conversion).

The ISO value in the camera is for those folks who are used to the conventional way of doing things, IMHO. And of course, the review images in the LCD are suitably bright thereby satisfying their expectations.

Thirdly, are there any implications about the bit-depth that would be needed?

None that I know of. What did you have in mind?

Again, Sigma was a bit peculiar in that regard. The ADC's are 12-bit - but quite a bit of digital processing goes on during the writing to the card with the result that the raw output file is 16bit and the raw values can be up to 14,000 (depending on the camera model).

[george013]

Secondly, in that case why do you need the ISO value in the camera at all? This could be set automatically during the raw conversion process provided the user can override the choice (as we often do now in tweaking the exposure during the conversion).

The ISO value in the camera is for those folks who are used to the conventional way of doing things, IMHO. And of course, the review images in the LCD are suitably bright thereby satisfying their expectations.

I don't understand what is stated here. The main goal for a high isso is to gain a suitebal, shorter, shutter speed. So it's changing the RAW data.

George

[ajohnw]

I don't understand what is stated here. The main goal for a high isso is to gain a suitebal, shorter, shutter speed. So it's changing the RAW data.

George

Ted's being a bit naughty here George. ISO could be replaced with something else but it has been around for a long time and attempts to replace this sort of thing have failed in the past.

ISO on a normal camera rather than Ted's foveon is obtained via "amplification" with a strong possibility that some values are interpolated in software. Amplification in quotes as data sheets aren't to clear about how it's obtained. As far as I am aware data sheets on larger sensors such as used in dslr's aren't available. The ones used in cheap and many technical cameras are available and there is no reason to suppose that the big ones are any different - especially these days when most cameras offer mechanical shutterless operation one way or the other or part use an electronic shutter all of the time.

These chips generally have separate R G and B gain controls and what is usually called a chip gain. The RGB ones are used for colour balance and the chip one for sensitivity. Even at this level sensors often have facilities for both electronic and mechanical shutter operation. At this level they are also virtually a camera on a chip and don't need much else to function. Larger sensors might use a separate ASIC for some functionality along with a general purpose microcontroller or maybe the latter looks after everything that isn't built into the sensor. I'm not about to take one of mine apart to find out. A definition of ASIC can be found on the web. There is a pretty strong possibility that things like buffering are handled by something like that.

John
-

[xpatUSA]

I don't understand what is stated here.

The point of my post was to explain that there are some cameras that do not obey the conventional rules and do not have a variable gain analog amplifier (VGA) in front of the analog to digital converter (ADC).

The main goal for a high isso is to gain a suitebal, shorter, shutter speed. So it's changing the RAW data.

George

If you meant that statement to apply to all digital cameras, the statement is not always correct.

This is a conventional camera signal block diagram for which your statement is correct:



Note that the block "ISO Gain Adj" is before the ADC block. Selecting a higher ISO amplifies the sensor signal thus presenting a bigger voltage to the ADC. Therefore the value of the signal *.RAW is changed, just as you said. BUT, in ISO-less cameras, the ISO Gain Adj block is not there, or is not used. Therefore, in that case, the value of the signal *.RAW can not be changed by the ISO setting.

There are also cameras where the ISO gain block value is changed only for part of the ISO selection range.

Try Googling "ISO-less camera" for more information to help you understand the concept of ISO-less.

Nikon D7000 owners might find this of interest:

http://1000wordpics.blogspot.com/201...-say-that.html
.

[george013]

The point of my post was to explain that there are some cameras that do not obey the conventional rules and do not have a variable gain analog amplifier (VGA) in front of the analog to digital converter (ADC).



If you meant that statement to apply to all digital cameras, the statement is not always correct.

This is a conventional camera signal block diagram for which your statement is correct:



Note that the block "ISO Gain Adj" is before the ADC block. Selecting a higher ISO amplifies the sensor signal thus presenting a bigger voltage to the ADC. Therefore the value of the signal *.RAW is changed, just as you said. BUT, in ISO-less cameras, the ISO Gain Adj block is not there, or is not used. Therefore, in that case, the value of the signal *.RAW can not be changed by the ISO setting.

There are also cameras where the ISO gain block value is changed only for part of the ISO selection range.

Try Googling "ISO-less camera" for more information to help you understand the concept of ISO-less.

Nikon D7000 owners might find this of interest:

http://1000wordpics.blogspot.com/201...-say-that.html
.

I did do a search for iso-less and found a lot of discussion about it but not what it is. So I'm still wondering.
I also read that article and printed it to study it more carefully.

I don't believe that ISO adjustments is a matter of enlarging the signal from the sensor only, If so, an image shot at 1600 ISO with clipping area's could be transformed to an 100 iso image without clipping. Everybody who is starting editing will say that is impossible.

Going one step further but only with reference to my early Sigma cameras, the ISO selection can be ignored too, because they are true ISO-less cameras. A true ISO-less camera does no analog amplification in-camera. In other words, the captured raw data is not changed by the ISO selection. So, there is absolutely no difference between the the two following actions (in terms of raw data), all other things being equal:

Shoot at ISO 400 and expose for 0 EV.
Shoot at ISO 100 and expose for -2 EV.

The ISO setting is passed to the raw converter as meta-data and an appropriate multiplication is done during conversion.

Nice and simple, just the way I like it . .

It's not only that simple, I think it's wrong. First, for the same reason as I explained before, second, for you forget the 2 major tools a photographer has: shutterspeed to play with movements and aperture to play with DOF. If you do that, you're reducing your camera to a flatbed scanner.

I can't change the exposurevalue in the RAW-converter or in PP.

George

[ajohnw]

The amplification is probably needed to help remove noise problems introduced by the A/D so I don't accept the general circuit diagram as there is very likely to always be some amplification what ever the ISO is set at. That gain may even be in the charge to voltage converter - if that is what it is. My understanding is that sensor sites behave like capacitors so charge is represented by voltage. The only things that store charge as current are inductors. What I suspect who ever did did the diagram really means is a typical method of sampling low capacitance values accurately and quickly - a differential amplifier that provides the power to the output drawing very very little from the input. A sample and hold type circuit can offer similar functionality and also scale the voltages being measured as required. Just how they do this in practice is open to debate. CCD's for instance can use strange avalanche techniques to boost the signal.CMOS sensors probably also have some sort of buffer to isolate the actual light sensor built into each cell. It might even be possible to read from these directly.

What I think Ted is saying in terms of ISO less is that raw files are typically 12 or 14 bit data in a 16bit format for convenience. That means that the signal could me amplified in software by a factor of 4^2 in one case and 2^2 in another without overflowing. Unfortunately currently life isn't as simple as that as as 16 for instance would only provide 4 EV variation. The other fact is that it's generally accepted that increasing ISO decreases dynamic range. What's not clear is if that is just down to noise amplification or other physical factors as well. In practice both probably have an effect. These days lower iso's are sometimes gained by simply reducing the exposure as far as the sensor but not the user is concerned. Software then fixes that. ISO changes could be handled in a similar fashion at least in part via messing with the exposure time as far as the sensor is concerned.

John
-

[xpatUSA]

I don't believe that ISO adjustments is a matter of enlarging the signal from the sensor only, If so, an image shot at 1600 ISO with clipping area's could be transformed to an 100 iso image without clipping. Everybody who is starting editing will say that is impossible.

Sorry, George, that is false logic and is not relevant to ISO-less cameras.

A true ISO-less camera does no analog amplification in-camera. In other words, the captured raw data is not changed by the ISO selection. So, there is absolutely no difference between the the two following actions (in terms of raw data), all other things being equal:

Shoot at ISO 400 and expose for 0 EV.
Shoot at ISO 100 and expose for -2 EV.

It's not only that simple, I think it's wrong.

Please consider the following points, step-by-step, one at a time.

1) For any camera sensor, the output is dependent only on the exposure (illuminance X time, lux-sec).

2) Changing the ISO setting does not change the sensor output.

(Please note that, at this point, we are only talking about the sensor, therefore the so-called "exposure triangle" does not apply)

3) In the signal flow, there is an analog to digital converter (ADC) to convert the sensor output from analog to digital.

4) If the camera is not ISO-less, there is a variable gain amplifier (VGA), also called a programmable gain amplifier (PGA) before the ADC. It's gain is set to 1 at the camera's base ISO, normally 100 except for some Nikons at 200. Each ISO step upward increases the gain by a fixed amount. For whole steps of ISO, e.g. 100, 200, 400, etc., the gain is increased by X2.

5) If the camera is ISO-less, there is no variable gain amplifier. If necessary, I can prove this by showing the histogram for two Sigma Raw files shot as per my claim above. The histograms will be virtually the same, proving conclusively that the raw data is unchanged by the ISO on my cameras (not yours).

If you understand and agree with all of the above points, we may be able to continue with our discussion. If not, I am waiting for your proof, point-by-point, that any are incorrect. It would really help if your statements were less general in nature.

I can't change the exposure value in the RAW-converter or in PP.

George

I can. I move a slider called "Exposure Compensation".

[Manfred M]

Ted, any of the literature that I've read seems to indicate that there is a fairly simple amplifier circuit built into each individual photo site on the sensor chip, so gain is applied at the source (photodiode output). In standard CMOS sensors these are in the top backplane layer, just below the optical elements and in the "back lit" devices they sit between the microlenses and the bottom of the optical path.

Intuitively this makes sense, as applying gain as close to the source as possible minimizes electrical noise. The downside would be that the circuit elements would have to be quite simple (non-linear?) to fit into the relatively small area. I have no idea as to how the Foveon sensor in the Sigma cameras would be built, but at a high level the electronics would have to sit in each respective colour layer?

I would assume that there would be a secondary level elsewhere on the chip to voltage match the A/D input requirements

[xpatUSA]

CMOS sensors probably also have some sort of buffer to isolate the actual light sensor built into each cell.

Exactly. The electrons induced by the captured incident photons flow into the photo-diode's intrinsic capacitance, creating a charge which, for a capacitor is represented by a voltage. This voltage is buffered by a transistor (source follower) which has the property of an extremely high input resistance. The voltage "gain" at this stage is of the order of 5uV/electron; so if the sensor saturates at say 45,000 electrons then the sensor's saturation voltage is 225mV.

The Foveon sensor in my cameras has three analog outputs, one for each photodiode layer; it saturates at 550mV and an exposure of 0.8 lx-sec. These outputs are fed direct to three 12-bit ADCs. Processing in-camera before writing to the card gives rise to values greater than 2^12 (scaling, linearization, and more).

What I think Ted is saying in terms of ISO less is that raw files are typically 12 or 14 bit data in a 16bit format for convenience. That means that the signal could be amplified in software by a factor of 4^2 in one case and 2^2 in another without overflowing.
John
-

So, for example, my SD14 raw file data saturates (shows sensor clipping) at around 8,000 (approx 13bits which is 0-8191). The saturation in raw data is not affected by the ISO setting. If I took a shot that is exposed -2EV with respect to saturated (ignoring metering), the raw data would be 2,000. If I actually had set ISO to 400, the converter would apply X4 while creating the linear ROMM working file. If I had left the ISO at 100, both the LCD and the converted file would look dark but moving the converter's EC slider to +2 makes the perceived exposure look correct. Either way, the raw data is the same.