ISO - Waste of time when shooting RAW?

This is all very silly. If you reduce a digital signal by 5 stops, you also reduce the sensitivity by 5 stops. e.g. a signal that starts as 256 will be able to distinguish 256 levels. A signal that start at 8 will only have 8 possible levels. What's the signal to noise ratio got to do with that? Further, if you reduce the signal by 5 stops you will throw away everything that would have been below 32, but the signal to noise ratio stays more or less the same.
 
This is all very silly. If you reduce a digital signal by 5 stops, you also reduce the sensitivity by 5 stops. e.g. a signal that starts as 256 will be able to distinguish 256 levels. A signal that start at 8 will only have 8 possible levels. What's the signal to noise ratio got to do with that? Further, if you reduce the signal by 5 stops you will throw away everything that would have been below 32, but the signal to noise ratio stays more or less the same.


Nope - the different types of noise are introduced at different intervals and some are subject to analogue amplification where as some are introduced at the digital conversion stage; The noise introduced prior to the analogue amplification stage can not be changed in post - it is embedded it to the file and is dependant on ISO set in camera. More noise is then added with the digital conversion and with an ISO invariant sensor this can 'almost' be changed at will in Post without any image degradation; this is why manufacturers/others are so excited about it :)

The CCD/CMOS does not care about 'sensitivity' - its pixels collect Photons from the whole scene, from dark areas there are a lot less and are also a lot more random - this creates a charge on the CCD/CMOS. The analogue converter changes this charge into a voltage. Increasing ISO does not increase the sensitivity of the CCD - it is just an amplifier that amplifies the voltage from the analogue converter before this signal enters the digital converter.
 
Last edited:
This is utter rubbish. Signal to noise ratio is only one of the concerns with photography (or many other things for that matter). How do you get the idea that you will get a better dynamic range by doing this. It will clearly be worse. Much, much worse.
 
To add to this.
You seem to be imagining an analogue system, but this isn't. It is a mostly digital system. The sensor signal is recorded digitally and from there on it is digital. That means that the detail finer than 1 is treated as zero. If you amplify zero you get ....... (drum roll) ....... zero!
 
To add to this.
You seem to be imagining an analogue system, but this isn't. It is a mostly digital system. The sensor signal is recorded digitally and from there on it is digital. That means that the detail finer than 1 is treated as zero. If you amplify zero you get ....... (drum roll) ....... zero!

Steve - you have a very clever advanced camera that can take the charge produced by photons 'landing' in photosites straight into a binary (digital) signal?

Mr Canon doesn't agree as he seems to use some 'imaginary analogue system'as well:

A/D Conversion Explained:

Simply put, an A/D converter samples the analogue electric signals from the camera’s CMOS or CCD image sensor (N.B. He doesn't state it is recorded digitally like you have?), which vary in intensity based on the number of photons captured in each pixel, and converts them into digital data consisting of 0’s and 1’s. In the case of the EOS 5D Mark II and 50D cameras, the raw digital data produced by the A/D converter is fed directly to the DIGIC 4 image processing circuit, which does the mathematical “heavy lifting” of converting the data to a usable image.

Here is a very simple analogy: The original analog signals are like an old-fashioned clockface -- the moving hands progress through the second markers, indicating every instant of passing time: In other words, the signal is stepless. For a digital camera to function, these analog ('stepless') signals must be converted into digital signals with definite, distinct steps. Using the clock analogy again: This digital signal acts more like a modern digital clockface, with a numeric display changing to indicate the passage of time, but without continuous movement between seconds.

Several factors may affect the strength or purity of the analogue signals reaching the A/D converter, such as electronic noise generated by the image sensor or its readout circuitry, or electronic noise generated by the signal amplification which occurs at high ISO speed settings (No........are you sure Mr Canon, that means you agree with me!). These factors primarily affect the dynamic range of the image. (What 'utter rubbish' according to Mr Bollygum on the TP Forum)

I've added the bits in red to the quote :)
 
Last edited:
This type of response I find quite 'tyresome' and is something I have seen over and over again in the 40+ years I have been interested in Photography!

Photographers; professional and keen amateurs are 'snobs' that do not like their 'interest' being made more simple! They take great pride in 'understanding' the exposure triangle, DoF, Shutter speeds etc............whoopy do!

They actually don't want a camera that you can give to any Tom,Dick or Harry and achieve outstanding results from - they want newbies and wannabies to 'learn' their chosen craft just like they had to!

Let me give you a time line since I have been taking photographs:

(1) I was given my first camera by my father - a Canon Rangefinder in 1979 when he purchased a Nikon F2 with a non-metering prism; the photo press at the time 'shunned' the photomic prism stating 'real' photographers didn't need it - they used proper light meters to evaluate the scene and exposure or could guess it from experience!

(2) Nikon introduced the professional Nikon F3; the press at the time and professional photog's 'slagged' it off; bah real photographers don't use an electronically controlled camera with, wait for it...........Aperture priority mode; FFS this camera is for amateurs who don't understand manual exposure! (It went on to become Nikon's longest running Pro Camera but initially had to be sold for less than the older F2)

(3) Autofocus - the press said it was good for the visually challenged photographer but couldn't see any 'mainstream' use for it - surely in photography you had to see well enough to focus a lens; what were camera manufacturers thinking!

(4) Digital: naaaah; film produces much better quality images and digital would never get near the resolution of modern films! To be a PROPER photographer you had to have your own darkroom and process your own images.

So, stop burying your heads in the sand, wake up to advances in camera design and learn about them - otherwise it is you that is Lazy! I n the past we looked on great darkroom printers as a very skilled 'artist' why do we not accept Post Production artists with the same enthusiasm?

Phew............rant over!

I worked on photo magazines throughout this entire period. I don't recall it like being like that at all, not from the substantial majority for sure.
 
This is utter rubbish. Signal to noise ratio is only one of the concerns with photography (or many other things for that matter). How do you get the idea that you will get a better dynamic range by doing this. It will clearly be worse. Much, much worse.

Not rubbish. If you don't want to understand how it works, and there's no need to, just try it using a camera with an ISO-invariant sensor and see for yourself. You won't have to look hard.
 
Steve - you have a very clever advanced camera that can take the charge produced by photons 'landing' in photosites straight into a binary (digital) signal?

Mr Canon doesn't agree as he seems to use some 'imaginary analogue system'as well:



I've added the bits in red to the quote :)
Do you have an analogue camera or a digital camera? All post processing is done with a digital signal and definitely all ISO invariant cameras are digital. So why argue analogue?
 
Not rubbish. If you don't want to understand how it works, and there's no need to, just try it using a camera with an ISO-invariant sensor and see for yourself. You won't have to look hard.
I have 3 ISO invariant cameras. I think that they are great and I often underexpose by 1-2 stops, but not 4-5 stops. If I underexpose by 4-5 stops then those areas that would have been -5 are now -10. Do you think this is ok?
 
Going back to the microphone analogy - if the gain is turned way down and someone speaks quietly into the mic we can hardly hear them, if you turn the gain up (Higher ISO) you can then hear them but everything else (background sounds) is amplified as well. If you could keep the gain on the Mic low ( Low ISO) so that the background was quite but separated out his voice and digitally amplify just that (ISO Invariance) it would be the ideal solution.
This is close...
But the truth is the whole thing is defined/controlled by the amount of light collected (or lack of, photon shot noise). If you underexpose to save highlights you will have noisier shadow areas when recovered (in comparison). Take a high resolution smaller sensor (i.e. 24mp APS) where noise starts to be problematic at ISO's above 800. It doesn't matter how/why...when an area gets to an equivalent of ISO 800 light exposure it will become problematic/noisy.

As I have explained several times... *Invariance* simply means the results are no worse, not *better*.
 
This type of response I find quite 'tyresome' and is something I have seen over and over again in the 40+ years I have been interested in Photography!

Photographers; professional and keen amateurs are 'snobs' that do not like their 'interest' being made more simple! They take great pride in 'understanding' the exposure triangle, DoF, Shutter speeds etc............whoopy do!

They actually don't want a camera that you can give to any Tom,Dick or Harry and achieve outstanding results from - they want newbies and wannabies to 'learn' their chosen craft just like they had to!

Let me give you a time line since I have been taking photographs:

(1) I was given my first camera by my father - a Canon Rangefinder in 1979 when he purchased a Nikon F2 with a non-metering prism; the photo press at the time 'shunned' the photomic prism stating 'real' photographers didn't need it - they used proper light meters to evaluate the scene and exposure or could guess it from experience!

(2) Nikon introduced the professional Nikon F3; the press at the time and professional photog's 'slagged' it off; bah real photographers don't use an electronically controlled camera with, wait for it...........Aperture priority mode; FFS this camera is for amateurs who don't understand manual exposure! (It went on to become Nikon's longest running Pro Camera but initially had to be sold for less than the older F2)

(3) Autofocus - the press said it was good for the visually challenged photographer but couldn't see any 'mainstream' use for it - surely in photography you had to see well enough to focus a lens; what were camera manufacturers thinking!

(4) Digital: naaaah; film produces much better quality images and digital would never get near the resolution of modern films! To be a PROPER photographer you had to have your own darkroom and process your own images.

So, stop burying your heads in the sand, wake up to advances in camera design and learn about them - otherwise it is you that is Lazy! I n the past we looked on great darkroom printers as a very skilled 'artist' why do we not accept Post Production artists with the same enthusiasm?

Phew............rant over!

Your points are not invalid - I just don't agree with them.

If you want to be all about the post-production, go and be a graphics designer. Of course, like every industry things evolve over a long period of time and as that happens we all gradually shift our mindsets. However, I genuinely feel that appreciating depth of field, aperture etc etc etc is not about snobbery, it is about the love of this wonderful craft. To take those things into consideration and deliberately apply them to a composition that perhaps only you can see, and to get the result you aimed for - it is an exhilarating feeling. So yes, as I am entitled to my opinion, it does make me feel a little sick in my mouth when I see people writing things such as "I can't be bothered to actually think about taking good photographs, look at this black image of a housing estate, now look how I have horrendously altered the exposure and contrast in post-production to produce an image I could have taken in seconds with some basic knowledge anyway" (paraphrasing obviously).

To say it is about being stuck-up is wrong. I am an amateur photographer, I have a full-time career and only make a small amount on the side through portraits etc. So to say I am looking down on others (or imply it) is so far off the mark. I am one of the people at the bottom, doing it right - for the love of photography. I maintain my original stance, that relying on RAW too heavily as a crutch is a bad thing, and the market is saturated with lazy people that think this way now.
 
I can see why you think that, but it isn't - not if you use it sensibly. It'd be very wrong to dismiss ISO-invariance as anything less than a major advance, allowing you to do things that you simply cannot achieve with a conventional sensor, no matter good your craftsmanship (without resorting to HDR technique).

There's no need to change anything about the way you work, no downsides, just shoot as normal. But there will always be the major upside of much more shadow detail to exploit, plus the opportunity to retain very bright highlights just by reducing exposure a little.

I will give it a go when I next get out, but I am one of those that prefers to get everything right in-camera as far as humanly possible, for me that is all part of the photography experience. I would find it soul-destroying always heading home with a series of black images and hoping I could do something with them.
 
I have 3 ISO invariant cameras. I think that they are great and I often underexpose by 1-2 stops, but not 4-5 stops. If I underexpose by 4-5 stops then those areas that would have been -5 are now -10. Do you think this is ok?

Okay, so you have three ISO-invariant cameras, and you think they're great. I'm not sure what the problem is then, unless you just want to argue analogue/digital semantics.
 
To Hoppy and Fraser.
Just to make sure I am not misinterpreting the original scenario, let me describe it.
A) You take a tripod photo at ISO 100 and 1/1000 sec of a static subject. This photo registers on the camera as being 5 stops underexposed.
B) You take another photo, of the same subject, this time ISO 3200 at 1/1000 sec. This photo registers on the camera as being correctly exposed.
In this example nothing is overexposed in either shot.
I interpret what you are saying to mean that there is no discernible difference between these photos after optimal post processing (increasing the exposure in shot A) by +5). Is this correct?
 
Last edited:
To Hoppy and Fraser.
Just to make sure I am not misinterpreting the original scenario, let me describe it.
A) You take a tripod photo at ISO 100 and 1/1000 sec of a static subject. This photo registers on the camera as being 5 stops underexposed.
B) You take another photo, of the same subject, this time ISO 3200 at 1/1000 sec. This photo registers on the camera as being correctly exposed.
In this example nothing is overexposed in either shot.
I interpret what you are saying to mean that there is no discernible difference between these photos after optimal post processing (increasing the exposure in shot A) by +5). Is this correct?

That is exactly my understanding of what ISO in-variance means. Interested to see if more knowledgeable folks confirm this.
 
Ok, so next step.
Imagine 3 pixels in case B
The first is 256,128,64 (RGB)
The second is 240,100,20
The third is 100,20,10
What are the same pixels in case A?
 
Ok, so next step.
Imagine 3 pixels in case B
The first is 256,128,64 (RGB)
The second is 240,100,20
The third is 100,20,10
What are the same pixels in case A?


There is a massive amount of 'processing' in the camera before you get the pixel numbers you have quoted; these processes add 'noise' to the image and make the pixel number 'inaccurate' in relation to the scene you are capturing. using ISO invariance tries to remove these 'errors' from the pixel data.
 
That is exactly my understanding of what ISO in-variance means. Interested to see if more knowledgeable folks confirm this.

If you have a scene infront of you and that contains a large tonal range and you take a picture of it the dark areas of that scene contain more noise due to the way photons work - you have no control over this.

If you keep this noise and the noise generated by the actual CCD in the camera at it's minimum (low ISO = less amplification of this noise) then the dark areas of the picture in the end result have less noise but you can then increase the highlights in an ISO invariant camera to produce a cleaner image; less noise in the dark areas & only a very small increase in the bright areas.

This is close...
But the truth is the whole thing is defined/controlled by the amount of light collected (or lack of, photon shot noise). If you underexpose to save highlights you will have noisier shadow areas when recovered (in comparison). Take a high resolution smaller sensor (i.e. 24mp APS) where noise starts to be problematic at ISO's above 800. It doesn't matter how/why...when an area gets to an equivalent of ISO 800 light exposure it will become problematic/noisy.

As I have explained several times... *Invariance* simply means the results are no worse, not *better*.

That's not what I am reading but I have to go to work so will go into it in more depth later.
 
Bear with me. Can you tell me what the new figures will be?

Please fully read the quote I made from Canon earlier - you need to fully understand how the camera gets to it's digital output.

Canon said:
A/D Conversion Explained:

Simply put, an A/D converter samples the analogue electric signals from the camera’s CMOS or CCD image sensor, which vary in intensity based on the number of photons captured in each pixel, and converts them into digital data consisting of 0’s and 1’s. In the case of the EOS 5D Mark II and 50D cameras, the raw digital data produced by the A/D converter is fed directly to the DIGIC 4 image processing circuit, which does the mathematical “heavy lifting” of converting the data to a usable image.


Several factors may affect the strength or purity of the analogue signals reaching the A/D converter, such as electronic noise generated by the image sensor or its readout circuitry, or electronic noise generated by the signal amplification which occurs at high ISO speed settings. These factors primarily affect the dynamic range of the image.
 
Last edited:
To summarise this thread for those that aren't particularly bothered with the semantics...

A) There is no real difference in the resulting noise on ISO invariant sensors, so there is no advantage in underexposing. You can still adjust many images that are correctly exposed in post without creating unpleasant noise.

B) There is a disadvantage in not seeing a correctly exposed photo on the rear screen. You can't really argue this point. The argument has been made that the advancement of technology is wonderful and people should embrace it, but some seem to be unwilling to accept the advance in technology of being able to look at an image on the rear screen.

C) The only advantage seems to be that you can selectively lighten areas of an image to preserve highlights, which really limits this usage to static subjects or forms of photography where you would be using a tripod. But there are better ways of doing this such as bracketing which would not have the disadvantage of a poorly exposed shot on the rear screen.

Fraser, I honestly think that you are overstating the significance of this technology, whether you now just want to win this debate or not is unclear to me. I really mean no offence in saying that.
 
Last edited:
A
To summarise this thread for those that aren't particularly bothered with the semantics...

A) There is no real difference in the resulting noise on ISO invariant sensors, so there is no advantage in underexposing. You can still adjust many images that are correctly exposed in post without creating unpleasant noise.

B) There is a disadvantage in not seeing a correctly exposed photo on the rear screen. You can't really argue this point. The argument has been made that the advancement of technology is wonderful and people should embrace it, but some seem to be unwilling to accept the advance in technology of being able to look at an image on the rear screen.

C) The only advantage seems to be that you can selectively lighten areas of an image to preserve highlights, which really limits this usage to static subjects or forms of photography where you would be using a tripod. But there are better ways of doing this such as bracketing which would not have the disadvantage of a poorly exposed shot on the rear screen.

Fraser, I honestly think that you are overstating the significance of this technology, whether you now just want to win this debate or not is unclear to me. I really mean no offence in saying that.
A is correct
B is incorrect. Sony A7R2&3 will allow you to see a correctly exposed image
C Not sure what this is saying
 
I think you may need to re-read my post in the context of this whole thread.
 
Do you mean that that is what the consensus of the thread is, not what is correct?
 
I don't think a Sony A7R will show you a corrected version of a photo you've just taken if it has been taken 5 stops under. I know that it can in live view, but that isn't what is being discussed.
 
What is the difference between live view and what you are talking about?
 
Steve, if you take a photo 5 stops under exposed the preview of that photo on your rear screen will be underexposed. You don't seem to have read some of the earlier comments in this thread discussing how this would be a disadvantage as you couldn't check that you've nailed the focus etc.
 
Not being a technical person I won't get caught up in all the scientific discussion. Seems to be two schools of thought, those who are prepared to embrace 'invariance' and those that dismiss it, both points of view are fine and nothing wrong with either.

I use Canon gear so am ISO invariant challenged, however my view is that invariance is simply another tool that can be used to reach the goal, I don't think those that are prepared to embrace it will be underexposing everything by 5 stops (I could be wrong) and getting everything back in post. I can see situations where it will be useful whilst still striving to achieve nirvana in camera.
 
To Craigus On my cameras it isnt. It gives you a choice to show it as underexposed or not .
 
Last edited:
Do you mean that that is what the consensus of the thread is, not what is correct?
I think you're beating your head against a brick wall Steve.

Basic electronic theory clearly doesn't apply and post-processing to recover lost bits is the future.

Try underexposing an 8-bit image by 5EV and recovering that in post.
 
I think you're beating your head against a brick wall Steve.

Basic electronic theory clearly doesn't apply and post-processing to recover lost bits is the future.

Try underexposing an 8-bit image by 5EV and recovering that in post.
It is a strange conversation. :) .
 
To summarise this thread for those that aren't particularly bothered with the semantics...

A) There is no real difference in the resulting noise on ISO invariant sensors, so there is no advantage in underexposing. You can still adjust many images that are correctly exposed in post without creating unpleasant noise.

B) There is a disadvantage in not seeing a correctly exposed photo on the rear screen. You can't really argue this point. The argument has been made that the advancement of technology is wonderful and people should embrace it, but some seem to be unwilling to accept the advance in technology of being able to look at an image on the rear screen.

C) The only advantage seems to be that you can selectively lighten areas of an image to preserve highlights, which really limits this usage to static subjects or forms of photography where you would be using a tripod. But there are better ways of doing this such as bracketing which would not have the disadvantage of a poorly exposed shot on the rear screen.

Fraser, I honestly think that you are overstating the significance of this technology, whether you now just want to win this debate or not is unclear to me. I really mean no offence in saying that.

Great reply Craigus :) some of the points are true (especially the last paragraph!)

Let me present why I think this development is fantastic and you can judge for yourself- again I will try and answer the other points in the small hours of the morning when I'm not replying on my phone! (Hope you don't mind)

Ever since I can remember photography has relied on the exposure triangle; everything in engineering is a compromise and we have to work within this triangle. The closer we move to the edge of this triangle the more we are 'c'ompromising' one or two settings.

Now, we are heading towards an ISO-less sensor, where you no longer have an ISO setting. This then breaks the exposure triangle and we are no longer constrained by it!

A simple example - taking a dusk landscape - set at an aperture for optimum DoF then you can still use a fast shutter speed and achieve the ''correct' exposure - no need for tripod.

Add to this improved IQ (still under debate here) and it's awesome!

Lens design (fast aperatua lenses not as necessary unless DoF control wanted) is made more simple etc.

The digital image on the rear screen is just ''digitally generated' by software so, in the future this would be displayed 'correctly' but can't at the moment.

Thoughts?
 
To Hoppy and Fraser.
Just to make sure I am not misinterpreting the original scenario, let me describe it.
A) You take a tripod photo at ISO 100 and 1/1000 sec of a static subject. This photo registers on the camera as being 5 stops underexposed.
B) You take another photo, of the same subject, this time ISO 3200 at 1/1000 sec. This photo registers on the camera as being correctly exposed.
In this example nothing is overexposed in either shot.
I interpret what you are saying to mean that there is no discernible difference between these photos after optimal post processing (increasing the exposure in shot A) by +5). Is this correct?
That is exactly my understanding of what ISO in-variance means. Interested to see if more knowledgeable folks confirm this.
Yes...
 
Not being a technical person I won't get caught up in all the scientific discussion. Seems to be two schools of thought, those who are prepared to embrace 'invariance' and those that dismiss it, both points of view are fine and nothing wrong with either.

I use Canon gear so am ISO invariant challenged, however my view is that invariance is simply another tool that can be used to reach the goal, I don't think those that are prepared to embrace it will be underexposing everything by 5 stops (I could be wrong) and getting everything back in post. I can see situations where it will be useful whilst still striving to achieve nirvana in camera.

Good summary (y)
 
To summarise this thread for those that aren't particularly bothered with the semantics...

A) There is no real difference in the resulting noise on ISO invariant sensors, so there is no advantage in underexposing. You can still adjust many images that are correctly exposed in post without creating unpleasant noise.

B) There is a disadvantage in not seeing a correctly exposed photo on the rear screen. You can't really argue this point. The argument has been made that the advancement of technology is wonderful and people should embrace it, but some seem to be unwilling to accept the advance in technology of being able to look at an image on the rear screen.

C) The only advantage seems to be that you can selectively lighten areas of an image to preserve highlights, which really limits this usage to static subjects or forms of photography where you would be using a tripod. But there are better ways of doing this such as bracketing which would not have the disadvantage of a poorly exposed shot on the rear screen.

Fraser, I honestly think that you are overstating the significance of this technology, whether you now just want to win this debate or not is unclear to me. I really mean no offence in saying that.

A - yes
B - yes
C - no

On C, an ISO-invariant sensor in practise has greatly extended sensitivity at the shadow end. So just shoot as normal and pull up those shadows to taste in post. Or, you can exploit it further when highlights are so bright that they blow with 'correct' exposure, simply by reducing exposure by maybe a stop or two. You will still have plenty of shadow detail for a high quality result, and a usable LCD image. There is no limitation around static subjects or tripods.

Basically, there are no downsides to ISO-invariance, but significant upsides (in extended dynamic range) if you care to exploit them.
 
Ok, so next step.
Imagine 3 pixels in case B
The first is 256,128,64 (RGB)
The second is 240,100,20
The third is 100,20,10
What are the same pixels in case A?
This is not nearly as simple as you would like...
First, what we are actually concerned with is light sensitivity (DR) and light gathered by the sensor, not the output. And at the pixel level, at best I would have 1 R/G/B pixel each for a 3 pixel sample (Bayer type). Secondly, we don't know where the minimum sensitivity/usability is. From those numbers the only thing I can tell is than none of them are at the minimum.... so I can't tell you how far the exposure can be reduced before some register as zero (black).

And "black" (or white) in the output does *NOT* mean that no usable data was recorded there... it only means that we cannot discern/see it in the output... that's what "recoverability" means. Another way to think of it is; how many times can you divide a number in half before reaching zero? It's pretty much infinite depending on how small of a unit/result you can use...
 
Last edited:
1. Imagine a nice analogue sine wave.
2. Pass it through a 14-bit ADC. The output will be a digital representation with 16,384 (2^14) discrete values.
3. Reduce the amplitude of the analogue sine wave by a factor of 32 (2^5).
4. Pass it through the same 14-bit ADC. The output will be a digital representation with 512 (2^(14-5)) discrete values.
5. Amplify the second digitised sine wave by a factor of 32. The output will be a digital representation with 512 discrete values.
6. Make a nice cup of tea and accept that there is a difference albeit that it probably makes bugger-all difference to the price of custard creams.
7. Go out and take some pictures, exposing them however you like. But remember to take the lens cap off first.
 
With all the arguments going on, I'm wondering, as most cameras these days have Auto ISO, if your skill level is that low that you cannot expose properly, why you don't just use that option?
 
Back
Top