Richard Ditch and I just completed a marathon exchange on what influences noise in our images.
http://www.naturephotographers.net/imagecritique/ic.cgi?a=vp&pr=193675&CGISESSID=376ca7fee59731c8e5ea459ca71721b8&u=19485
It originated from my stating that iso, aperture and shutter speed were all important to observed noise. Richard maintained that only iso was important. This has practical consequences for how we pursue low noise imaging in fhe field. Also, when we compare noise levels in our images at different iso levels we should know whether to think about exposure. When we compare images from different cameras at the same iso I can see that other variables associated with sensor design can enter. However, even there, I think when we talk about noise comparisons we will need to take exposure settings into account.
It's easy to get all wrapped up in words for this topic. So, I think the easiest way to summarize it would be to refer you to an article with images:
"As you can see, having a low ISO setting means almost nothing if we end up adjusting the exposure heavily. If a photo is made at ISO 100 and requires two stops of exposure adjustment in post, we will have introduced a lot more noise than if we had exposed the initial image at ISO 400 (a two stop sensitivity increase)."
I realize that noise is a potentially very complex topic. Also, like Richard, we are all so accustomed to just thinking of only iso with respect to noise. I'd like to know if the NPN community agrees with this article or not. If we do, then how does this influence our efforts to minimize noise in our photography and to compare noise in different images?
In our long exchanges in the avian gallery you never brought up the topic of exposure; you only referred to shutter speed and aperture parameters in the same manner as ISO.
I agree completely with the short article you cited here, and the second one in your post on your tern image. Under exposing leads to increased noise in the image (I wrote that more than once in my comments); increasing exposure in post processing to compensate for under exposure increases noise (I wrote that more than once as well).
Neither point has anything to do with the specific aperture or shutter speed used - you can achieve proper exposure with a set of coupled shutter speed and aperture settings that maintain the same Exposure Value. Make your exposure with any combination of shutter sped and aperture you like to use as long as that combination doesn't underexpose the image for the amount of light you are working in and ISO you have set.
You can use the histogram on the back of the camera to assist in this. It is why so many of us "expose to the right;" i.e., keep the histogram as far to the right (bright end) without completely bunching it up against the right edge (which blocks up all the bright areas in blown highlights).
1. ISO (higher values produce more noise than lower values for any camera)
2. increasing exposure in raw conversion (bringing up the shadows makes the noise more prominent) [there's that under exposure]
3. cropping (effectively enlarging the noise)
4. sharpening (increasing the edge contrast of each noisy pixel)
It looks like this is just a misunderstanding, Richard. I'm glad that you agree with the articles. As you recall, this discussion started because I did not think it was correct to compare two images considering only iso when aperture and ss (i.e. exposure) were so different. I mentioned in my e-mail to you about this forum post, I initiated this post to know what is the correct answer. I would enjoy seeing comments from others [Richard, by "from others" I just mean that you and I are both pretty weary of trying to communicate with each other about this topic. I hope someone else can help.], especially with regards to the questions of a field strategy for minimizing noise in our photography and for comparing noise in different images.
I'm not sure there is anything more to "help" with. Using a low ISO and not underexposing will pretty much minimize noise. You can't really bring into it "aperture and shutter speed" as you can get the same exposure using a variety of apertures and shutter speeds. You should really only refer to under, over, or correct exposure.
Steve, I suspect the reason that there have been few (no?) others entering into this thread is because Richard is doing an exemplary job of objectively clarifying the issues surrounding the topic of electronic noise. I, for one, am grateful to you for having introduced the topic of noise and a number of misconceptions related thereto because it has provided us all the opportunity to benefit from your questions and perspectives and from Richard's patient, objective, and closely reasoned responses to some unfortunately commonly repeated misunderstandings on the matter. These topics are broached frequently by my students and by participants in workshops and classes and I thank you both for this discussion. Differing perspectives are always helpful. They have been well stated and It would seem, now, time to move on.
Dave in East River, SD
I haven't done any scientific comparisons, but in my experience for a given camera, proper exposure is the main issue. I have demonstrated this to myself time and again until if finally sank in and is now my normal practice. Particularly with the newer CMOS sensor cameras, it is much better to properly expose at higher ISO than to underexpose and correct by even one stop in post. Again without doing a lab type test, I haven't seen that noise is better or worse at different aperture settings as long as exposure is correct. Plus the physics (as I understand them) don't support that aperture would affect noise. Shutter speed, is the other hand can be a factor at long exposures. This is supported by the fact that the camera manufacturers build long exposure NR into the camera. However, again with my limited knowledge, that only becomes an issue at extremely long ss (i.e. multiple sedonds).
Thank you Alan and David. Before we move on: how would you answer both questions I posed.
i. How to minimize noise in the field: O.K., I understand that we don't want to under-expose at a particular iso. Of course exposure and iso are the fundamental parameters. However, our control variables are ss, aperture and iso. I could have a stab at this but it would be interesting to know what you would advise. In particular, how do we judge when to shift to the next iso based on what we see in the histogram? There is a trade-off between noise generated from low exposure and that generated by increased iso. What do we know about how noise varies with low exposure as opposed to increased iso?
ii. How to compare images? Can we agree that when we compare the noise in two images, say from the same camera to simplify matters, if we want to say that increased iso has caused the increase in noise we must be examining two images taken at the same camera exposure. I understand that different combinations of aperture and ss can result in the same exposure. A third question might be whether, in practice, it matters to the noise level whether one combination or another of aperture and ss has produced that exposure. I would expect not.....but, maybe there is more to this.
When I look closely at the luminance noise in my photos (after reducing color noise and sharpening to accentuate it in ACR) and in the crops of this photo I see luminance noise does not vary pixel to pixel but rather there are small groups of dark pixels adjacent to small groups of light pixels. This makes the noise more noticeable when viewed at more normal magnifications. Why isn't it a pixel to pixel variation if it is random noise at each sensor photosite?
i. How to minimize noise in the field: O.K., I understand that we don't want to under-expose at a particular iso. Of course exposure and iso are the fundamental parameters. However, our control variables are ss, aperture and iso. I could have a stab at this but it would be interesting to know what you would advise. In particular, how do we judge when to shift to the next iso based on what we see in the histogram? There is a trade-off between noise generated from low exposure and that generated by increased iso. What do we know about how noise varies with low exposure as opposed to increased iso?
A. Expose properly. Push the histogram as far to the right as possible without stacking everything up against the right edge.
B. Use the lowest ISO possible for the light and the subject. That means only increase the ISO if you can't get a high enough shutter speed to freeze subject motion, or stop down the aperture for enough depth of field. Using a tripod to prevent camera motion will vastly increase the range of slow shutter speeds you can get away with, and thus allow for using much lower ISO settings. Use whatever pairing of shutter speed and aperture gives the proper exposure that you like - it doesn't matter in terms of noise as long as you stay away from very long exposure times where the electrons moving in the circuits warm them up enough to create more noise.
ii. How to compare images? Can we agree that when we compare the noise in two images, say from the same camera to simplify matters, if we want to say that increased iso has caused the increase in noise we must be examining two images taken at the same camera exposure. I understand that different combinations of aperture and ss can result in the same exposure.
In any scientific test it is imperative to only change a single variable at a time. When magazines publish noise tests in their camera reviews they use a fixed target and controlled light. Each time they increase the ISO by a factor of two (e.g., going from 400 to 800) they must decrease the exposure by the same factor of two (by increasing the shutter speed by a factor of two, or stopping down one more stop, or decreasing the flash output by half). For every exposure they get the same histogram so that the results are not influenced by under exposure. "The same exposure" is not the same as "the same combination of shutter speed and aperture."
You do not want to compare two images where you have changed the ISO but not the exposure to compensate because you would need to increase exposure in post processing to make them have the same brightness.
A third question might be whether, in practice, it matters to the noise level whether one combination or another of aperture and ss has produced that exposure. I would expect not.....but, maybe there is more to this.
We've covered this countless times already. The answer is still the same: it does not matter as long as the exposure is the same.
Steve: I've been following this thread tying to understand your reasoning. I'm not following it. The words you have been using are that shutter speed and/or aperture have an effect on noise. They do not. ISO affects noise. I think what I'm reading is that you are using shutter speed and aperture as a synonym for exposure. They are not synonyms. They are just 2 of the knobs to modify exposure. Then you added the issue of underexposing an image in camera, then later increasing the exposure in post processing and how that will affect noise. An underexposed image, post processed to brighten it up will generally have a negative affect on noise. Much depends on the camera, how much underexposed and what the tonal values in the image are whether the increased noise will be visible or not. I'll add a couple of concrete examples (I'm assuming these will only demonstrate points you fully understand, but maybe will help clarify the language used in this thread that I think is causing the confusion and disagreement).
Scenario #1
At at given ISO:
1/250 @ f/8
1/1000 @ f/4
Those two combinations of shutter/aperture will generate the same exposure (i.e. light to the sensor).
Both of those settings will generate an image with the same noise characteristics.
Scenario #2
Now assume that from the previous example the ISO is doubled because the ambient light is exactly half of what it used to be.
1/125 @ f/8
1/500 @ f/4
Again those two combinations will generate the same exposure
Both of those settings will generate an image with the same noise characteristics. In comparison to Scenario #1, the exposures will be the same (the ambient light was half but the amount of light to the sensor was doubled), but the noise in scenario #2 will be increased because the ISO was doubled.
You asked when you should think about underexposing in the field, rather than increasing ISO. Again, the answer is it depends. If the camera you are using can handle the increased ISO, and still generate a file that is useable for your purposes for noise, then increase the ISO. If you are maxed out on ISO, and your aperture is already wide open then you don't have much choice but to underexpose and deal with the image in post processing, or drop your shutter speed down to provide the "proper" exposure. If that shutter speed is too slow to satisfy the needs of the situation, then you'll have to deal with an underexposed image to keep the shutter speed up.
I think Richard has explained it well. The best thing to do is fully understand the capabilities of your camera, how the sensor in your camera deals with noise and do the best you can to "properly" expose images in the field. Note that "proper" is in quotes as that often does not mean the metered value.
As I read through this thread I think you are unnecessarily complicating the issue. Given any lighting condition we can arrive at a correct exposure by adjusting any of three variables. Of those three, with correct exposure, the only one that will affect noise is ISO.
We would never have noise issues if we always had enough light, or could use long exposure times. Unfortunately we often have to trade some noise in exchange for faster shutter speeds even at wide open apertures. Hopefully as technology advances that will become ever less of an issue.
Thank you, everyone. I'm amazed at the communication problem here. I'm much more accustomed to people understanding what I write and me correctly interpreting their comments. Anyway, I'd like to add a couple of more thoughts to this discussion:
i. Considering Dan's comment that underexposure is more of a threat that high iso. That is: "It is much better to properly expose at higher ISO than to underexpose and correct by even one stop in post." maybe we can come up with modification to Richard's strategy of using " the lowest ISO possible for the light and the subject". To guard against the possibility of underexposure perhaps we should error on the side of using a higher than necessary iso. It depends how concerned we are about light conditions changing and/or our ability to define the correct match of exposure and iso via the histogram.
ii. There is a difference between in-camera exposure and raising exposure in post processing using software. Perhaps that distinction should be made more clear (I hesitate to try to do it ).
"To guard against the possibility of underexposure perhaps we should error on the side of using a higher than necessary iso"
That doesn't really make any sense to me. We should try to expose correctly, so if we use a "higher than necessary ISO" we would have to compensate by either closing the aperture or using a faster shutter speed than needed. Otherwise you would over expose the image which has its own set of problems.
To guard against the possibility of underexposure perhaps we should error on the side of using a higher than necessary iso. It depends how concerned we are about light conditions changing and/or our ability to define the correct match of exposure and iso via the histogram.
You are still confusing ISO with exposure control.
To guard against the possibility of under exposure learn to use the metering system in your camera and pay attention to the histogram. If you use the same Program or auto exposure mode then you will always get the same histogram no matter what ISO you have set. Raising the ISO and following the same exposure method will only give you a faster shutter speed and/or a smaller aperture - it will not move the histogram closer to the right edge of the LCD.
If your histogram is consistently pushed to the left (or even the middle of the plot) then it might help to dial in some positive exposure compensation - that will make the histogram shift to the right.
ISO does not control the exposure; it only controls how sensitive the sensor is to the light. It does that by turning up the gain and amplifying the lower number of photons striking the sensor while at the same time amplifying the random noise inherent in the sensor. It reduces the signal to noise ration and makes the noise more prominent.
The only reason to use a higher ISO than necessary is to make noise more obvious in the image. A higher ISO will let you use a faster shutter speed so that is good insurance if you are shooting action, but once the ISO is high enough for the situation increasing it further will only degrade the image by pumping up the noise.
There is a difference between in-camera exposure and raising exposure in post processing using software. Perhaps that distinction should be made more clear
Of course there is.
The sensor in the camera records the photons of light that strike the photosites. There is a threshold level (the noise) below which no image data can be detected. The more photons that are collected the better the signal to noise ratio and the more image detail you see and the less noise you see. Dark areas in an image are the result of few photons being collected, and noise is more prominent in the dark areas as a result. So, it is best to record as many "dark photons" as possible to get some image detail above the noise level.
If you under expose then the image data/detail is barely different from the inherent noise in the dark areas. When you "increase exposure" in raw conversion or with the Levels slider in post processing you are increasing the noise along with the data that is barely different from it: by bringing up the detail in the shadows you are also bringing up the noise. There's no magic way for your software to invent data that wasn't collected by the camera sensor.
Thank you, everyone, for taking the time to participate in this discussion.
Posted by rob ashcroft on July 10, 2012 at 6:27 AM
Digital noise is caused by a weak signal at the photo-sensor site (pixel). A weak signal is caused by poor (low level) light. Let's assume you’re shooting in bright light, with an exposure of 1/100 second at f8 at ISO 100. You then change to a much dimmer light and to use the same settings of 1/100 second and f8, you need to raise the ISO to 800. The amount of light reaching the sensor is 1/8th of the first shot at ISO 100. There is less ‘signal’, and the result is more noise in the final image. That’s the basic reason why there is more noise at high ISO settings than low ones.
The best ways to avoid noise are...
1. Use a camera that has high-quality pixel sensors. My Canon 5DM2 is better than my Canon 350D
2. Shoot as far to the right as you can without clipping the whites. The way the bits are arranged across the exposure scale gives more data to the right (over-exposed) side. If you have a scene that is fairly even with no real high-spots, then over-expose the shot and correct it down in RAW edit, which is a non-destructive edit. Don't do it in JPEG/TIFF edit.
3. Shoot as low an ISO that you can. If the light is low, use a tripod! Even holding the camera on a fence post will make an amazing difference, and allow you to use ISO100.
4. If you can, use flash - either flash-guns or studio flash. All of my indoor still-life shots are done with flash at ISO100. I never get a noise problem. It just doesn't happen. I'd say this is the most important point. You need good quality light so the camera doesn't have to struggle to get a good signal.
The way the bits are arranged across the exposure scale gives more data to the right (over-exposed) side.
That's true, but has almost nothing to do with noise. That has more to do with banding and posterization, and ability to manipulate the file. The reason why exposing to the right works is purely because it increases the signal-to-noise ratio by letting more light hit the sensor.
(And yes, this is contrary to a couple of articles out there online, notably one on Luminous-Landscape. And yes, those articles are wrong and a couple have since been corrected/clarified.)
Posted by rob ashcroft on July 10, 2012 at 1:07 PM
David
"That's true, but has almost nothing to do with noise. That has more to do with banding and posterization, and ability to manipulate the file. The reason why exposing to the right works is purely because it increases the signal-to-noise ratio by letting more light hit the sensor."
Are you saying that if part of my image is in a shadow area (let us say its about 80%black) but has a lot of detail and I take a normal exposure I will get one result, and if I then increase the exposure by one stop to the right (more data) I'll not notice any difference in that dark area with regard to noise?
I understood that noise is cause when a photo-sensor gets a weak signal because not enough light is hitting it for long enough, and it therefore misreads the signal.
Are you saying that if part of my image is in a shadow area (let us say its about 80%black) but has a lot of detail and I take a normal exposure I will get one result, and if I then increase the exposure by one stop to the right (more data) I'll not notice any difference in that dark area with regard to noise?
No, not at all. You'll definitely see a difference. But the reason is because you increased the signal in that area, and the noise stayed constant; So you have a higher signal-to-noise ratio.
I think the problem is that "more data" can mean two things...Hopefully I can explain...
There are two reasons I've seen given for why exposing to the right (or Expose To The Right, or just ETTR) decreases noise :
1) ETTR uses the brighter part of the histogram which has more brightness levels, and hence finer gradations, which makes for less noise. (This is wrong.)
2) ETTR increases the signal (the amount of light) hitting the sensor, which increases the signal-to-noise ratio. (This is correct.)
Item #1 does not really affect the amount of noise in your image... if you had finer gradations in the shadows the noise may look different, but it would be the same in amount. Item #2 is the real reason ETTR works, because the variation caused by noise gets swamped by the actual signal, so you don't really see the noise (or it has less effect).
So when you say "more data", you could be referring to an increase in signal (and increase in light) or an increase in the number of brightness levels available. If you're referring to an increase in signal, you're correct; If you're referring to an increase in brightness levels, you're incorrect.
Posted by Steve Balke on July 11, 2012 at 10:46 AM
Rob and David, Thank you very much for your additional comments. They are all very useful and informative.
Discussion
22 replieshttp://www.naturephotographers.net/imagecritique/ic.cgi?a=vp&pr=193675&CGISESSID=376ca7fee59731c8e5ea459ca71721b8&u=19485
It originated from my stating that iso, aperture and shutter speed were all important to observed noise. Richard maintained that only iso was important. This has practical consequences for how we pursue low noise imaging in fhe field. Also, when we compare noise levels in our images at different iso levels we should know whether to think about exposure. When we compare images from different cameras at the same iso I can see that other variables associated with sensor design can enter. However, even there, I think when we talk about noise comparisons we will need to take exposure settings into account.
It's easy to get all wrapped up in words for this topic. So, I think the easiest way to summarize it would be to refer you to an article with images:
http://photo.tutsplus.com/articles/post-processing-articles/quick-tip-the-real-secret-to-avoiding-noise/
A key statement in this article is as follows:
"As you can see, having a low ISO setting means almost nothing if we end up adjusting the exposure heavily. If a photo is made at ISO 100 and requires two stops of exposure adjustment in post, we will have introduced a lot more noise than if we had exposed the initial image at ISO 400 (a two stop sensitivity increase)."
I realize that noise is a potentially very complex topic. Also, like Richard, we are all so accustomed to just thinking of only iso with respect to noise. I'd like to know if the NPN community agrees with this article or not. If we do, then how does this influence our efforts to minimize noise in our photography and to compare noise in different images?
I agree completely with the short article you cited here, and the second one in your post on your tern image. Under exposing leads to increased noise in the image (I wrote that more than once in my comments); increasing exposure in post processing to compensate for under exposure increases noise (I wrote that more than once as well).
Neither point has anything to do with the specific aperture or shutter speed used - you can achieve proper exposure with a set of coupled shutter speed and aperture settings that maintain the same Exposure Value. Make your exposure with any combination of shutter sped and aperture you like to use as long as that combination doesn't underexpose the image for the amount of light you are working in and ISO you have set.
You can use the histogram on the back of the camera to assist in this. It is why so many of us "expose to the right;" i.e., keep the histogram as far to the right (bright end) without completely bunching it up against the right edge (which blocks up all the bright areas in blown highlights).
As I wrote in the discussion in Avian:
Noise comes from:
1. ISO (higher values produce more noise than lower values for any camera)
2. increasing exposure in raw conversion (bringing up the shadows makes the noise more prominent) [there's that under exposure]
3. cropping (effectively enlarging the noise)
4. sharpening (increasing the edge contrast of each noisy pixel)
Dave in East River, SD
i. How to minimize noise in the field: O.K., I understand that we don't want to under-expose at a particular iso. Of course exposure and iso are the fundamental parameters. However, our control variables are ss, aperture and iso. I could have a stab at this but it would be interesting to know what you would advise. In particular, how do we judge when to shift to the next iso based on what we see in the histogram? There is a trade-off between noise generated from low exposure and that generated by increased iso. What do we know about how noise varies with low exposure as opposed to increased iso?
ii. How to compare images? Can we agree that when we compare the noise in two images, say from the same camera to simplify matters, if we want to say that increased iso has caused the increase in noise we must be examining two images taken at the same camera exposure. I understand that different combinations of aperture and ss can result in the same exposure. A third question might be whether, in practice, it matters to the noise level whether one combination or another of aperture and ss has produced that exposure. I would expect not.....but, maybe there is more to this.
A. Expose properly. Push the histogram as far to the right as possible without stacking everything up against the right edge.
B. Use the lowest ISO possible for the light and the subject. That means only increase the ISO if you can't get a high enough shutter speed to freeze subject motion, or stop down the aperture for enough depth of field. Using a tripod to prevent camera motion will vastly increase the range of slow shutter speeds you can get away with, and thus allow for using much lower ISO settings. Use whatever pairing of shutter speed and aperture gives the proper exposure that you like - it doesn't matter in terms of noise as long as you stay away from very long exposure times where the electrons moving in the circuits warm them up enough to create more noise.
In any scientific test it is imperative to only change a single variable at a time. When magazines publish noise tests in their camera reviews they use a fixed target and controlled light. Each time they increase the ISO by a factor of two (e.g., going from 400 to 800) they must decrease the exposure by the same factor of two (by increasing the shutter speed by a factor of two, or stopping down one more stop, or decreasing the flash output by half). For every exposure they get the same histogram so that the results are not influenced by under exposure. "The same exposure" is not the same as "the same combination of shutter speed and aperture."
You do not want to compare two images where you have changed the ISO but not the exposure to compensate because you would need to increase exposure in post processing to make them have the same brightness.
We've covered this countless times already. The answer is still the same: it does not matter as long as the exposure is the same.
To fix the histogram (get proper exposure) we change the shutter speed and/or aperture, not the ISO.
We change the ISO only when we don't have enough shutter speed to stop motion, or when we need to use a smaller aperture for more depth of field.
Scenario #1
At at given ISO:
1/250 @ f/8
1/1000 @ f/4
Those two combinations of shutter/aperture will generate the same exposure (i.e. light to the sensor).
Both of those settings will generate an image with the same noise characteristics.
Scenario #2
Now assume that from the previous example the ISO is doubled because the ambient light is exactly half of what it used to be.
1/125 @ f/8
1/500 @ f/4
Again those two combinations will generate the same exposure
Both of those settings will generate an image with the same noise characteristics. In comparison to Scenario #1, the exposures will be the same (the ambient light was half but the amount of light to the sensor was doubled), but the noise in scenario #2 will be increased because the ISO was doubled.
You asked when you should think about underexposing in the field, rather than increasing ISO. Again, the answer is it depends. If the camera you are using can handle the increased ISO, and still generate a file that is useable for your purposes for noise, then increase the ISO. If you are maxed out on ISO, and your aperture is already wide open then you don't have much choice but to underexpose and deal with the image in post processing, or drop your shutter speed down to provide the "proper" exposure. If that shutter speed is too slow to satisfy the needs of the situation, then you'll have to deal with an underexposed image to keep the shutter speed up.
I think Richard has explained it well. The best thing to do is fully understand the capabilities of your camera, how the sensor in your camera deals with noise and do the best you can to "properly" expose images in the field. Note that "proper" is in quotes as that often does not mean the metered value.
As I read through this thread I think you are unnecessarily complicating the issue. Given any lighting condition we can arrive at a correct exposure by adjusting any of three variables. Of those three, with correct exposure, the only one that will affect noise is ISO.
We would never have noise issues if we always had enough light, or could use long exposure times. Unfortunately we often have to trade some noise in exchange for faster shutter speeds even at wide open apertures. Hopefully as technology advances that will become ever less of an issue.
Peter
i. Considering Dan's comment that underexposure is more of a threat that high iso. That is: "It is much better to properly expose at higher ISO than to underexpose and correct by even one stop in post." maybe we can come up with modification to Richard's strategy of using " the lowest ISO possible for the light and the subject". To guard against the possibility of underexposure perhaps we should error on the side of using a higher than necessary iso. It depends how concerned we are about light conditions changing and/or our ability to define the correct match of exposure and iso via the histogram.
ii. There is a difference between in-camera exposure and raising exposure in post processing using software. Perhaps that distinction should be made more clear (I hesitate to try to do it
That doesn't really make any sense to me. We should try to expose correctly, so if we use a "higher than necessary ISO" we would have to compensate by either closing the aperture or using a faster shutter speed than needed. Otherwise you would over expose the image which has its own set of problems.
Peter
You are still confusing ISO with exposure control.
To guard against the possibility of under exposure learn to use the metering system in your camera and pay attention to the histogram. If you use the same Program or auto exposure mode then you will always get the same histogram no matter what ISO you have set. Raising the ISO and following the same exposure method will only give you a faster shutter speed and/or a smaller aperture - it will not move the histogram closer to the right edge of the LCD.
If your histogram is consistently pushed to the left (or even the middle of the plot) then it might help to dial in some positive exposure compensation - that will make the histogram shift to the right.
ISO does not control the exposure; it only controls how sensitive the sensor is to the light. It does that by turning up the gain and amplifying the lower number of photons striking the sensor while at the same time amplifying the random noise inherent in the sensor. It reduces the signal to noise ration and makes the noise more prominent.
The only reason to use a higher ISO than necessary is to make noise more obvious in the image. A higher ISO will let you use a faster shutter speed so that is good insurance if you are shooting action, but once the ISO is high enough for the situation increasing it further will only degrade the image by pumping up the noise.
Of course there is.
The sensor in the camera records the photons of light that strike the photosites. There is a threshold level (the noise) below which no image data can be detected. The more photons that are collected the better the signal to noise ratio and the more image detail you see and the less noise you see. Dark areas in an image are the result of few photons being collected, and noise is more prominent in the dark areas as a result. So, it is best to record as many "dark photons" as possible to get some image detail above the noise level.
If you under expose then the image data/detail is barely different from the inherent noise in the dark areas. When you "increase exposure" in raw conversion or with the Levels slider in post processing you are increasing the noise along with the data that is barely different from it: by bringing up the detail in the shadows you are also bringing up the noise. There's no magic way for your software to invent data that wasn't collected by the camera sensor.
The best ways to avoid noise are...
1. Use a camera that has high-quality pixel sensors. My Canon 5DM2 is better than my Canon 350D
2. Shoot as far to the right as you can without clipping the whites. The way the bits are arranged across the exposure scale gives more data to the right (over-exposed) side. If you have a scene that is fairly even with no real high-spots, then over-expose the shot and correct it down in RAW edit, which is a non-destructive edit. Don't do it in JPEG/TIFF edit.
3. Shoot as low an ISO that you can. If the light is low, use a tripod! Even holding the camera on a fence post will make an amazing difference, and allow you to use ISO100.
4. If you can, use flash - either flash-guns or studio flash. All of my indoor still-life shots are done with flash at ISO100. I never get a noise problem. It just doesn't happen. I'd say this is the most important point. You need good quality light so the camera doesn't have to struggle to get a good signal.
That's true, but has almost nothing to do with noise. That has more to do with banding and posterization, and ability to manipulate the file. The reason why exposing to the right works is purely because it increases the signal-to-noise ratio by letting more light hit the sensor.
(And yes, this is contrary to a couple of articles out there online, notably one on Luminous-Landscape. And yes, those articles are wrong and a couple have since been corrected/clarified.)
Are you saying that if part of my image is in a shadow area (let us say its about 80%black) but has a lot of detail and I take a normal exposure I will get one result, and if I then increase the exposure by one stop to the right (more data) I'll not notice any difference in that dark area with regard to noise?
I understood that noise is cause when a photo-sensor gets a weak signal because not enough light is hitting it for long enough, and it therefore misreads the signal.
No, not at all. You'll definitely see a difference. But the reason is because you increased the signal in that area, and the noise stayed constant; So you have a higher signal-to-noise ratio.
I think the problem is that "more data" can mean two things...Hopefully I can explain...
There are two reasons I've seen given for why exposing to the right (or Expose To The Right, or just ETTR) decreases noise :
1) ETTR uses the brighter part of the histogram which has more brightness levels, and hence finer gradations, which makes for less noise. (This is wrong.)
2) ETTR increases the signal (the amount of light) hitting the sensor, which increases the signal-to-noise ratio. (This is correct.)
Item #1 does not really affect the amount of noise in your image... if you had finer gradations in the shadows the noise may look different, but it would be the same in amount. Item #2 is the real reason ETTR works, because the variation caused by noise gets swamped by the actual signal, so you don't really see the noise (or it has less effect).
So when you say "more data", you could be referring to an increase in signal (and increase in light) or an increase in the number of brightness levels available. If you're referring to an increase in signal, you're correct; If you're referring to an increase in brightness levels, you're incorrect.