Manual vs Av and Other Modes

[Arthur Morris]

Below is a feature from an old Bulletin that may help the newer folks.

I still plan on doing a blog post on when and why I use various modes not just manual :)

WORKING IN MANUAL MODE

Here is an excerpt from ABP II for those who need help with working in Manual mode.

Many professional and serious amateur photographers work in Manual mode most of the time. I do not because working in Av mode and entering exposure compensation is faster whenever the background is of a relatively constant tonality. When the background tonality is changing from moment to moment but the light is constant, it is best, however, to work in Manual mode. In either case, I rely on Evaluative Metering. Here are some examples of rapidly changing backgrounds: a shorebird on a rock along the edge of the ocean with waves breaking behind it. Birds flying against a blue sky with occasional white clouds. Cranes flying by in front of a variety of backgrounds that might include sky, mountains, yellowed grasses, or water.
As many folks are confused as to how to work in or set exposure compensation when working in Manual mode, I offer the following basic tutorial.

#1: When you work in Manual mode you select and set the shutter speed and you select and set the aperture. With my Canon cameras the default has you changing the shutter speed with your index finger dial and the aperture with the thumb wheel.

#2: After selecting Manual mode, point your camera at a scene or stationary subject and lock your tripod head so the framing remains constant. Next select and set the desired aperture. Then adjust the shutter speed until the analog scale in the viewfinder nulls out to zero. With Canon pro bodies this scale is laid out vertically along the right side of the viewfinder display (when you are working in horizontal format). With many of the pro-sumer bodies the analog scale is laid out horizontally at the bottom of the viewfinder display. The zero or null indicator is at the center of the analog scale. The three full stops above the null symbol (marked in 1/3-stop increments) indicate overexposure. The three full stops below the null symbol (also marked in 1/3-stop increments) indicate underexposure. If you change the aperture and you do not see the small square moving, check either the top or the bottom of the analog scale. You will note a small triangle at the top if you are way overexposed or a small triangle at the bottom if you are way underexposed. If the former, rotate the dial and choose faster shutter speeds, if the latter, choose slower shutter speeds. In either case, you will soon see the small square moving up or down the analog scale. At first, you will simply want to practice nulling the meter, that is, getting the small square to rest on the null symbol. This indicates that you have now set the metered exposure (as determined by the camera's Evaluative Metering system.

#3: When you work in Manual mode it is not possible to set exposure compensation. To come up with the exposure that you wish, simply change the aperture or shutter speed as above until the small square indicates the amount of over- or under-exposure that you desire. If you wish to work at +2 stops, you need adjust either the shutter speed or the aperture until the small square rests on the symbol that is two full stops above the null symbol. If you wish to underexpose by 1/3 stop, you need adjust either the shutter speed or the aperture until the small square rests on the symbol that lies just below the null symbol.

With a bit of practice you should quickly become comfortable whenever the need to work in Manual mode arises.

[Charles Glatzer]

Steve; can you explain to me why you need a meter at all in manual exposure? In manual exposure you are not compensating for anything.

regards~Bill

Bill,

The meter pattern provides a suggestion in any/all Priority Modes, thereafter you more than likely wiill need to alter the initial recommendation by adding or subtracting exposure via f/stop, shutter-speed, ISO or a combination therein to obtain the correct exposure for the subject at hand.

Meter patterns, reference values, etc are used to determine Exposure, regardless of the Priority mode being used.

Chas

[Charles Glatzer]

Below is a feature from an old Bulletin that may help the newer folks.

I still plan on doing a blog post on when and why I use various modes not just manual :)

WORKING IN MANUAL MODE

Here is an excerpt from ABP II for those who need help with working in Manual mode.

Many professional and serious amateur photographers work in Manual mode most of the time. I do not because working in Av mode and entering exposure compensation is faster whenever the background is of a relatively constant tonality. When the background tonality is changing from moment to moment but the light is constant, it is best, however, to work in Manual mode. In either case, I rely on Evaluative Metering. Here are some examples of rapidly changing backgrounds: a shorebird on a rock along the edge of the ocean with waves breaking behind it. Birds flying against a blue sky with occasional white clouds. Cranes flying by in front of a variety of backgrounds that might include sky, mountains, yellowed grasses, or water.
As many folks are confused as to how to work in or set exposure compensation when working in Manual mode, I offer the following basic tutorial.

#1: When you work in Manual mode you select and set the shutter speed and you select and set the aperture. With my Canon cameras the default has you changing the shutter speed with your index finger dial and the aperture with the thumb wheel.

#2: After selecting Manual mode, point your camera at a scene or stationary subject and lock your tripod head so the framing remains constant. Next select and set the desired aperture. Then adjust the shutter speed until the analog scale in the viewfinder nulls out to zero. With Canon pro bodies this scale is laid out vertically along the right side of the viewfinder display (when you are working in horizontal format). With many of the pro-sumer bodies the analog scale is laid out horizontally at the bottom of the viewfinder display. The zero or null indicator is at the center of the analog scale. The three full stops above the null symbol (marked in 1/3-stop increments) indicate overexposure. The three full stops below the null symbol (also marked in 1/3-stop increments) indicate underexposure. If you change the aperture and you do not see the small square moving, check either the top or the bottom of the analog scale. You will note a small triangle at the top if you are way overexposed or a small triangle at the bottom if you are way underexposed. If the former, rotate the dial and choose faster shutter speeds, if the latter, choose slower shutter speeds. In either case, you will soon see the small square moving up or down the analog scale. At first, you will simply want to practice nulling the meter, that is, getting the small square to rest on the null symbol. This indicates that you have now set the metered exposure (as determined by the camera's Evaluative Metering system.

#3: When you work in Manual mode it is not possible to set exposure compensation. To come up with the exposure that you wish, simply change the aperture or shutter speed as above until the small square indicates the amount of over- or under-exposure that you desire. If you wish to work at +2 stops, you need adjust either the shutter speed or the aperture until the small square rests on the symbol that is two full stops above the null symbol. If you wish to underexpose by 1/3 stop, you need adjust either the shutter speed or the aperture until the small square rests on the symbol that lies just below the null symbol.

With a bit of practice you should quickly become comfortable whenever the need to work in Manual mode arises.

Badda Bing!!!

Chas

[Don Thompson]

Below is a feature from an old Bulletin that may help the newer folks.

I still plan on doing a blog post on when and why I use various modes not just manual :)

WORKING IN MANUAL MODE

Here is an excerpt from ABP II for those who need help with working in Manual mode.

Thank you. Very simple and easy to understand. This is as I understood it. All the other discussions had me confused and thinking that me doubting myself.

Now, understanding and doing, especially quickly, is another matter. Still looking forward to your post as to more information on when to use all the different modes.

[Arthur Morris]

Many thanks Don. I can teach, and I can write, and I can make things simple and understandable. All of those every bit as much responsible for my success as my images. Though those don't suck either!

[Steven Kersting]

Steve, I am starting to get frustrated trying to get you to understand the important points below one last time.

When the light is relatively consistent and the background is changing or may change rapidly from moment to moment nobody is smart/fast enough to work in an automatic mode like Av. So we determine the correct exposure (usually via a histogram/blinkies check) for a given subject and then set that exposure manually. If a subject with a different tonality comes along it is easy to make the required changes quickly and easily say by adjusting the shutter speed (as most of us have learned to do).

As has been said above in so many words, trying to spot meter a bird in flight is totally insane.

Respectfully.

You're probably getting frustrated because you think I must be dense, you must be right, and there is no validity to what I am saying.

Let me give you an example assuming I am stationary with a scene I can pre expose and push the exposure to the right (i generally don't do that though) based upon the histogram/blinkies.

In this case I will probably be in manual mode but I will have the iso set to an auto range. For various subjects I will use EC or change metering pattern and the camera will trade ISO (I paid a lot of money for that ISO performance). If I choose to change aperture or SS for the scene (not subject. i.e. I suddenly care about shallow DOF more than optimal aperture) then the auto ISO will correct back to my original exposure. If I choose to both change metering mode and aperture for subject and scene I hold down a button (change metering mode) and rotate a dial and click, the ISO adjusts to give me a good exposure....all in fractions of a second.

If you are in "full manual" your first/ perhaps only option is to give on Aperture or SS (changing ISO manually on my Nikon is the hardest thing to do). Those are both things I'd rather not "give" on first.

In most cases Aperture is the critical choice. In those cases I will be in aperture priority with the camera set to an auto ISO range. In that case the camera will make exactly the same choices I would make. It will trade ISO first, then it will trade SS. To adjust for variances in subject I change metering mode or EC. Usually the only thing I have to do is change metering mode. Again, I'm giving on ISO first because that's what I care the least about.

If you can give on ISO first in full manual as quickly as you can SS/aperture, then great. If not I would suggest that my method is probably "better". It is for me.

As for the dynamic spot metering...ok, but I have good results. Not perfect, but good. If the metering is off, so is the focus. I much more often have auto focus issues than I have exposure issues with BIF. (I don't find trap/zone focus a suitable alternative for BIF) This is an area I am still working on as I have not found a great solution.

I did admit to having many more images using center weighted metering than spot, but I also have many more images taken w/ a D200/D300 than my current D3.

We don't have to agree, but I'm not an idiot. I learned manual mode when that's all there was and I could even get decent results from my FM when the battery froze and the meter died.

[Steven Kersting]

Steve; can you explain to me why you need a meter at all in manual exposure? In manual exposure you are not compensating for anything.

regards~Bill

What do you use as your starting point for exposure? How do you know if you have enough compensation set?

I suppose you could start with "f/16, SS=ISO". I had to do that on more than a few occasions *a long time ago*....

[Geoff Warnock]

Steven, nicely explained. Perhaps it is your point, but I would not consider using AutoISO in manual mode 'working manually', in fact I think it's near the same level as using aperture priority - the camera is choosing your exposure by altering ISO.

I too use Nikon, and have used autoISO, but mainly for indoor people photography. It's a useful tool, though I rarely use it lately.

I too also use manual when the background is changing, or in a tough exposure situation (sports - rugby - where one team is in a light coloured strip with the other in black for example) but tend to use Av with some exp compensation whenever possible otherwise. As Artie says - I find it faster.

[Elliotte Rusty Harold]

Let me repeat; too fast a shutter-speed will not effect image quality. This gives you any shutter-speed above the the minimum to work with, and any exposure can be achieved with right or left clicks with one just one thumb wheel. Basically you start with a faster speed than you need, reduce toward the motion blur point to increase exposure, and go the other way to decrease exposure. Need more light and you've reached the minimum shutter-speed? Just up the ISO and start over; a higher SS than you need and right or left clicks with the SS thumbwheel as needed.
The starting point is determined with the histogram.
You will always have both the minimum SS and the DOF you need. No errors, no surprises. Being comfortable with higher ISOs and an understanding of digital noise and how it can be minimized becomes important and quite useful with this type of exposure.
So is there anything wrong with this method? Also I can't imagine that this is an original idea.

regards~Bill

That might work in Florida or southern California. In the Northeast in winter and shade, I doubt it. There's just not enough light. Even shooting at the maximum ISO I can plausibly use on my camera (800), I pretty much need to be at wide open aperture (f/4 or f/5.6 depending on lens) to hope to get *up* to the desired shutter speed. Coming down to that speed is not an option. Shutter speeds range from about 1/60 to 1/800. If the sun actually deigns to come out, and I'm not shooting under trees, then I'll drop the ISO to 400, and maybe then if I'm still getting shutter speeds over 1/1000 I'll consider manual exposure.

At more southerly latitudes, or in summer, it's not hard to get enough sun to shoot at 1/1250 and f/11. There you're right that you do have a lot more room to play with the shutter speed. I don't meter the same way in southern Texas I do in southern New York, especially not in winter. But 9-10 months out of the year up here in New York the only plausible aperture is wide open, especially for handheld cameras or moving subjects. That value is essentially fixed no matter what metering mode you use. And at that maximum aperture, I'm taking the best shutter speed I can get, not picking the one I want. :-(

[Greg Basco]

Interesting discussion all around on this thread. I would simply add that choice of modes may differ according to what and where one is photographing.

I shoot full-time in tropical rainforests and, as Eliotte implies above, that is another place that is quite different from the kinds of things and the kind of environment in which many posters on this thread apparently are shooting.

I shoot about half the time in manual mode with spot metering and half the time in aperture priority with evaluative metering. As Chas and Arthur have stressed, understanding what's going on and then knowing when to use what to get what you want in your image is the key.

Cheers,
Greg Basco

[Steven Kersting]

Steven, nicely explained. Perhaps it is your point, but I would not consider using AutoISO in manual mode 'working manually', in fact I think it's near the same level as using aperture priority - the camera is choosing your exposure by altering ISO.

Yes, it is my point. I seldom ever use full manual because I prefer to change ISO first. But the camera is not choosing my exposure. I am choosing my exposure and allowing the camera to maintain it by altering ISO. If I need to change the exposure the camera is maintaining I'll use EC or a different metering pattern.

My point, which I think many may be missing, is that I am not giving up control to the camera. I am choosing settings so that the camera makes the exact same choices I would make and I don't have to. The more confined the goals/subject the more confined my settings are, up to the point of occasional full manual (no auto ISO).

Let me give another example. I set my base settings to give the desired exposure and because the lighting is pretty flat I add some EC. Now, if I swing into a darker scene (tree line) or a distinctly lighter scene (water/ dry reeds) the auto ISO will automatically keep the same exposure and I don't have to do anything. If you are in full manual you have to change something.
If it is a dark subject/light background then all I have to do is change metering mode. If you are in full manual you have to change something as well, but you also have to estimate the required offset which will probably be less accurate (or also change metering mode).

I am fully admitting that my methods will not always give ideal exposure SOOC, but I also almost never get an unusable exposure. If I have the time to fine tune a manual exposure I also have the time to fine tune an auto exposure (auto ISO in Manual mode probably).

I find my methods work better for me than full manual in most instances. I have never said auto anything is better than full manual. But I do say manual is definitely not "better" than auto settings if auto is used properly. (manual is easier to understand though)

[Desmond Chan]

I set my base settings to give the desired exposure and because the lighting is pretty flat

Say on a cloudy day with a scene that has nothing particularly bright or dark around, Steven? The suggested reading by the exposure meter should work well here without any compensation adjustment needed at all, I'd say, as it sounds like it's an "average" scene here, kind of like shooting a big 18% grey card.

I add some EC.

So you choose to over-expose? All right then, the shadow would appear not as dark and bright areas become brighter.

Now, if I swing into a darker scene (tree line)

Without any EC the darker scene will appears lighter than it should be in the final photo and so, -EC should be used to render the final image darker as it should be.

or a distinctly lighter scene (water/ dry reeds)

Without any EC the lighter scene will appear darker than it should be in the final photo and so + EC should be used to render the final image as bright as the actual scene should look like.

the auto ISO will automatically keep the same exposure

You mean the aperture and the shutter speed stay the same as before? I'd agree as the camera is now adjusting the ISO instead. I don't think the exposure -- exposure value -- is the same though; the exposure value of, say, f2.8 1/250s ISO 400 is not the same as f2.8 1/250s ISO 800. It's one stop difference there.

Also, as you have add some EC -- I assume you mean + EC -- and you does not adjust the EC afterward even though you are facing two different scenes, i.e., one darker and one brighter. I'd say only one of the resulting photos would have the exposure closer to the "correct" one and the other shot would have its exposure off by quite a bit.

I am fully admitting that my methods will not always give ideal exposure SOOC,

If the above describe how you actually work, it appears to me -- I'm guessing -- more than half of the time you'd have your exposure off

but I also almost never get an unusable exposure.

Well, these days, a bit off in exposure could be rescued in post-processing. Actually, it has always been like that only that back then, you fixed that when developing the film and during printing in the dark-room.

[Charles Glatzer]

Steven,

With the best of intentions....

My final suggestion is to wipe the slate clean and completely start over with a class on exposure fundamentals. I find your understanding, rational, and workflow flawed and beyond bizarre.

Warm Regards, and good images

Chas

[WIlliam Maroldo]

That might work in Florida or southern California. In the Northeast in winter and shade, I doubt it. There's just not enough light. Even shooting at the maximum ISO I can plausibly use on my camera (800), I pretty much need to be at wide open aperture (f/4 or f/5.6 depending on lens) to hope to get *up* to the desired shutter speed. Coming down to that speed is not an option. Shutter speeds range from about 1/60 to 1/800. If the sun actually deigns to come out, and I'm not shooting under trees, then I'll drop the ISO to 400, and maybe then if I'm still getting shutter speeds over 1/1000 I'll consider manual exposure.

At more southerly latitudes, or in summer, it's not hard to get enough sun to shoot at 1/1250 and f/11. There you're right that you do have a lot more room to play with the shutter speed. I don't meter the same way in southern Texas I do in southern New York, especially not in winter. But 9-10 months out of the year up here in New York the only plausible aperture is wide open, especially for handheld cameras or moving subjects. That value is essentially fixed no matter what metering mode you use. And at that maximum aperture, I'm taking the best shutter speed I can get, not picking the one I want. :-(

Elliot, an interesting point, and one I was thinking about since it occurred to me that the lighting conditions I shoot under might be what make manual exposure particularly effective and reliable.
As you have explained it, ISO is your limiting factor.

I may live in a southern climate, but I limit my photography to cloudy, overcast, and even rainy days if I can keep the camera dry. I avoid sunny cloudless days, some would say obsessively. This is because the advantages of low contrast light (overcast, etc) outweigh, by a considerable margin IMO, any advantages of shooting in high contrast light (bright sunny days). . However, as we decrease contrast we are also working with less light, and need higher ISOs.


My working ISO limit is ISO 3200, and I'd guess 20% of the time I use it. Mostly ISO 1000-1600. And this has little to do with the camera I am using, which is by no means a high ISO superstar (Sony A850).
None of this would be possible without the use of "exposing to the right" or pushing exposure as I usually refer to it, and since the ability is to push exposure, and not clip the highlights, corresponds directly to the contrast of the light, the lower the contrast, the higher the ISO that will produce acceptable results. .
I guarantee that if I do not push exposure at ISO 3200 the resulting image will be garbage, and I suspect that many who have visible proof that their camera has a useable ISO limit of lets say ISO 800 when they expose normally, have not tried to "expose to the right".
Again, the key is the contrast of the light.
As I read this it occurred to me I didn't explain why exposing to the right works to reduce digital noise.
In a nutshell, digital noise occurs mainly in the darker parts of an image. If you avoid capturing the darker parts of an image by increasing exposure (exposing to the right), then restore to the correct exposure in post-processing, you reduce digital noise.
regards~Bill

[Geoff Warnock]

Hi Bill, interesting reading from you too, but the fact you can talk about exposing to the right and purposely using overcast days suggests to me that your overcast is some peoples 'blindingly bright'!

From my personal perspective, I'm frequently amazed by the quoted exposures given here on BPN, while people say it was the last of the evening light, or a very overcast day. Here in Belgium this winter we had about 4 months of almost non-stop overcast weather. I was lucky if I could get 1/200th at f5.6 and iso 3200 for a dull exposure, so exposing the right would take me to very low shutter speeds for small birds at 400mm. Even now with great (for people at least) sunny weather, my exposure is around 1/500th at f8 iso400. The realms of shutter speeds 1/2000th and shorter are very rarely found at these latitudes, unfortunately.

Anyway, I'm straying away from the discussion about manual mode, though I can say I used full manual mode, with manual focus, for some remote camera shots of my shy garden birds!

[Roger Clark]

As I read this it occurred to me I didn't explain why exposing to the right works to reduce digital noise.
In a nutshell, digital noise occurs mainly in the darker parts of an image. If you avoid capturing the darker parts of an image by increasing exposure (exposing to the right), then restore to the correct exposure in post-processing, you reduce digital noise.
regards~Bill

Bill,
Actually as one exposes to the right noise increases! But signal increases more. So our perception is that noise is reduced. We perceive signal-to-noise ratio. Most of the noise we see in our digital camera images is due to photon noise (Poisson counting statistics from counting photons over the short time interval-our exposure time). Photon noise is the square root of the photons collected.

As we increase ISO we are not changing the actual sensitivity of the sensor, only a post sensor amplification. Above about ISO 400 to 800 one is digitizing a small enough photon range that the A/D converters are digitizing pretty much all the real signal there is to digitize. So one can shoot at say ISO 800 (or 1600) and set the exposure one wants (thus "underexposing") and bring the brightness up in post processing. The advantage here is that one has a greater dynamic range at the lower ISO. The disadvantage is the LCD display is a little dark.

Expose to the right is important only for lower ISOs where one needs to maximize signal-to-noise ratio and dynamic range.

For example, say you determine the correct exposure is 1/500 second at f/4, ISO 3200. Well, image at 1/500 second, f/4, ISO 1600 (or 1/500 second, f/4, ISO 800). You get a whole stop (or 2 stops) of headroom and dynamic range. In each exposure (1/55 at f/4, ISO 800, 1600, 3200) the signal-to-noise ratio on the subject is the same.

The above is true for newer cameras (less than about 3 years old that have good low level response without a lot of fixed pattern noise.

Roger

[Steven Kersting]

Desmond.

Everything you are saying is correct. Except for the exposure being "off". That comes down to whether you want a dark scene to be dark ("correct") and a light scene to be light (again "correct") based upon an average metering, or if you would rather have a dark scene lightened and a light scene toned down. I would generally choose the latter.

[Steven Kersting]

Steven,

With the best of intentions....

My final suggestion is to wipe the slate clean and completely start over with a class on exposure fundamentals. I find your understanding, rational, and workflow flawed and beyond bizarre.

Warm Regards, and good images

Chas

Ok, here's my "class on the exposure triangle".

http://skguitar.com/exposureTriangle.html

I created this based upon the "sunny 16 rule". (Max the light level out for "sunny day" ISO 100, f/16, 1/125 will give correct exposure)

[Roger Clark]

the auto ISO will automatically keep the same exposure

You mean the aperture and the shutter speed stay the same as before? I'd agree as the camera is now adjusting the ISO instead. I don't think the exposure -- exposure value -- is the same though; the exposure value of, say, f2.8 1/250s ISO 400 is not the same as f2.8 1/250s ISO 800. It's one stop difference there.

Yes, Steven is correct. Changing ISO when f/ratio and exposure time are held constant does not change exposure. ISO does not change the amount of light captured by the sensor, it only changes the level the A/D digitizes. See my other post in this thread from a few minutes ago.

Once one is above about ISO 800, there really is no difference (on newer good cameras) on the recorded image. Say one needs ISO 3200 at 1/500 second, f/4. One can set manual mode ISO 800, 1/500 second, f/4 and image forever (as long as the light doesn't get too bright) and brighten the image in post processing. The advantage is one has a higher dynamic range (2 stops in the example given). The disadvantage is the image will be dim in the LCD.

Roger

[Peter Hawrylyshyn]

Roger -
You've lost me. I see a white egret on a dull overcast day. By checking the histogram to make sure i'm protecting my whites : determine optimal exposure at ISO 3200 , 1/500 at F4. Why would I ever want to drop to ISO 800 1/500 F4 thereby shifting my image to the left?

I thought the idea was to shift to the right to optimize the amount of digital information you'll capture whether in Manual or Priority modes

[Desmond Chan]

Yes, Steven is correct. Changing ISO when f/ratio and exposure time are held constant does not change exposure. ISO does not change the amount of light captured by the sensor, it only changes the level the A/D digitizes. See my other post in this thread from a few minutes ago.

Once one is above about ISO 800, there really is no difference (on newer good cameras) on the recorded image. Say one needs ISO 3200 at 1/500 second, f/4. One can set manual mode ISO 800, 1/500 second, f/4 and image forever (as long as the light doesn't get too bright) and brighten the image in post processing. The advantage is one has a higher dynamic range (2 stops in the example given). The disadvantage is the image will be dim in the LCD.

Roger

So, the exposure value of ISO 3200 1/500s f4 is the same as ISO 800 1/500s f4 and also is the same as ISO 200 1/500s f4?

OK, it's something new here, Roger. So, what is exposure?

Traditionally, the above settings all give different exposure as we the mortals or most of us would understand it. If we look at the resulting photos straight off the camera (film or sensor), I believe we can all see the different among them: they don't look the same. Based on what you say, it seems to suggest that as long as they can be fixed in post-processing and get to the same or closely the same look, they actually all have the same exposure? And the fact that it can be done shows that they have the same exposure?

I could agree though that a bit off in exposure can be fixed in post-processing (or development and printing in the old days). But then what I just said is wrong, too, since, based on what you said, there is no such thing as a bit off in exposure at all !? Or are we just playing with words here?

[Desmond Chan]

Desmond.

Everything you are saying is correct. Except for the exposure being "off". That comes down to whether you want a dark scene to be dark ("correct") and a light scene to be light (again "correct") based upon an average metering, or if you would rather have a dark scene lightened and a light scene toned down. I would generally choose the latter.

And I'm referring to what most people would want in general, i.e., not turning a night scene photo into daylight shot

[Geoff Warnock]

I'm intrigued by what you're saying now Roger, and I can immediately see a use (choose a higher shutter speed to control motion, therby underexposing in camera, but brightening in processing).

My understanding up to now was that brightening an image in post would increase the perception of noise - is this no longer true?

Presumably at some point you do lose information to black once you drop out of the dynamic range of the sensor - so is about a 2-stop underexposure the sensible limit, or is this highly sensor (camera) dependent?

[Roger Clark]

My understanding up to now was that brightening an image in post would increase the perception of noise - is this no longer true?

Geoff,
Above about ISO 800 this has never been true. The problem with older cameras (more than about 3 years) there was larger fixed pattern noise and larger read noise. Fixed pattern noise still exists, but is greatly reduced as is read noise. So once you are up to about ISO 800, there is effectively no difference in signal-to-noise ratio from highlights to deepest shadows.

Presumably at some point you do lose information to black once you drop out of the dynamic range of the sensor - so is about a 2-stop underexposure the sensible limit, or is this highly sensor (camera) dependent?

With a good camera once you get above about ISO 800, you can boost in post processing. For example, ISO 1600 versus 100,000: you'l get the same results if the f/ratio and exposure times are the same. People tend to think of higher ISO delivering shorter exposure time, which means less light captured, so apparent noise goes up. But it is the less exposure time, not the ISO affecting the noise. So if you keep exposure time and f/ratio the same, it does not matter what the iso is above about ISO 800.

Detail sensor perfoemance at:
http://www.clarkvision.com/articles/...mance.summary/

Roger

[Mike Milicia]

Roger, I understand that there will be no difference regarding the amount of light that is collected at the sensor for the following two exposure settings : 1/500, f/5.6 ISO 800 and 1/500 f/5.6 ISO 3200. The only difference between these two settings is the amount of hardware amplification that will be done to the image data in the camera AFTER the shutter closes. Since the amount of light collected is the same in both cases, the amount of inherent noise in the initially captured data will be identical. With the ISO 800 image, you will have to boost the exposure by 2 stops in the RAW converter to get the image to look the same as the ISO 3200 image. Which ISO is "better" comes down to who does a better job of maximizing the signal-to-noise ratio during the post-capture amplification, the camera's hardware amplifier or the the RAW converter software. The "Expose to the Right" methodology is about maximizing the amount of light collected at the sensor. This is only affected by the aperture and shutter speed, so leaving those two things the same and just increasing the ISO to "expose to the right" is not really buying you anything. Lowering the ISO will generally reduce noise but only because you also then use a slower shutter speed or larger aperture which will mean that the sensor will collect more light. Increasing the ISO will generally increase noise but most of the increase in noise is due to the fact that you also then use a faster shutter speed or smaller aperture will mean that the sensor will collect less light.

Assuming the above is correct (probably a big assumption so please elaborate or correct as needed), I still have a couple of questions about your recent posts :

1 - Are you saying that on the latest cameras, ISO 800 is the point at which the signal-to-noise ratio of the end result becomes the same between increasing it further in camera or boosting the exposure in the RAW converter? Is this true of all the latest cameras? Does it matter what RAW converter you use? How did you arrive at ISO 800?

2 - You state in your example that reducing the ISO to 800 (vs. 3200) and underexposing by two stops gives you two more stops of headroom and dynamic range? How is that the case if I have to boost the exposure in the ISO 800 image by 2 stops in RAW conversion anyway?

Thanks for your help.

P.S.
Even thought the amount of light collected at the sensor at image capture time is the same in both cases, I would still consider 1/500, f/5.6 ISO 800 and 1/500 f/5.6 ISO 3200 to be two different "exposures" since the resulting RAW file data will be different. Also, since there are situations where ISO 3200 would result in unrecoverable blown highlights and ISO 800 would not, it is hard for me to call them the same "exposure".

[Roger Clark]

Roger -
You've lost me. I see a white egret on a dull overcast day. By checking the histogram to make sure i'm protecting my whites : determine optimal exposure at ISO 3200 , 1/500 at F4. Why would I ever want to drop to ISO 800 1/500 F4 thereby shifting my image to the left?

I thought the idea was to shift to the right to optimize the amount of digital information you'll capture whether in Manual or Priority modes

So, the exposure value of ISO 3200 1/500s f4 is the same as ISO 800 1/500s f4 and also is the same as ISO 200 1/500s f4?

OK, it's something new here, Roger. So, what is exposure?

Exposure is the amount of light captured by the sensor. When we change f/ratio or exposure time, we change the amount of light on the sensor. Many, many concepts from film days need to be thrown out the window. ISO is one of them. ISO does not change the amount of light the sensor captures. Digital (electronic) sensors have only one sensitivity (set by the quantum efficiency). (Film is similar, but different--another story.) ISO changes the range over which the signal from the sensor is digitized. Because of the fast readout requirements of photography, the A/D converters only have a dynamic range of 11 to 12 stops. Digital DSLR sensors have 14 or more stops. At low ISO, the noise in the shadows is dominated by noise from the A/D converters, not the sensor. Once gain is increased about 3 stops or more from the minimum ISO, the A/D can cover the range just fine. That is why once we go from ISO 100 to 800 the A/D can now cover the range of data from the sensor. Above that and one is wasting dynamic range while not improving the results from the sensor.

So even though in manual mode, one is not "exposing to the right" above ISO 800, one is not losing any data nor is one making the apparent noise worse.

Traditionally, the above settings all give different exposure as we the mortals or most of us would understand it. If we look at the resulting photos straight off the camera (film or sensor), I believe we can all see the different among them: they don't look the same. Based on what you say, it seems to suggest that as long as they can be fixed in post-processing and get to the same or closely the same look, they actually all have the same exposure? And the fact that it can be done shows that they have the same exposure?

I could agree though that a bit off in exposure can be fixed in post-processing (or development and printing in the old days). But then what I just said is wrong, too, since, based on what you said, there is no such thing as a bit off in exposure at all !? Or are we just playing with words here?

Traditionally, as we boost ISO we get shorter exposures, so less light to the sensor. It is the less light that results in more perceived noise, not that ISO increases noise. But working in manual, where we fix exposure time and f/stop, changing ISO above 800 only changes the brightness of the resulting image on the LCD screen and clips the highlights in the recorded data. It doesn't change the noise floor enough to be perceptible (one can measure a tiny difference in controlled laboratory conditions up to about ISO6400 on a 1DIV, for example, when there is no light on the sensor).

In the most difficult low light photography, astrophotography, amateur astronomers typically use a max ISO of 800 or 1600, because going higher just reduces dynamic range with no additional low light benefit. This has been practiced since the Canon 10D era (when was that, 2003??, or many camera generations ago).

What we need for the field is auto LCD image brightness, so if we underexpose at high ISO, we can still see the image on the screen. Then we can see our images in the field and fix brightness in post.

A next big improvement in DSLRs, I predict, is better A/Ds so we don't need ISO at all. With autoscaling on the LCD brightness, as long as we don't clip at minimum ISO, then all data get recorded from the sensor and we can choose an effective ISO (brightness) in post processing, like we can choose white balance now in post.

But until then, above about ISO 800, we are effectively there in terms of getting everything from the sensor. Just work in manual and cap ISO at 800 or 1600. The problem is reviewing images on the LCD.

Roger

[Roger Clark]

Roger, I understand that there will be no difference regarding the amount of light that is collected at the sensor for the following two exposure settings : 1/500, f/5.6 ISO 800 and 1/500 f/5.6 ISO 3200. The only difference between these two settings is the amount of hardware amplification that will be done to the image data in the camera AFTER the shutter closes. Since the amount of light collected is the same in both cases, the amount of inherent noise in the initially captured data will be identical. With the ISO 800 image, you will have to boost the exposure by 2 stops in the RAW converter to get the image to look the same as the ISO 3200 image. Which ISO is "better" comes down to who does a better job of maximizing the signal-to-noise ratio during the post-capture amplification, the camera's hardware amplifier or the the RAW converter software. The "Expose to the Right" methodology is about maximizing the amount of light collected at the sensor. This is only affected by the aperture and shutter speed, so leaving those two things the same and just increasing the ISO to "expose to the right" is not really buying you anything. Lowering the ISO will generally reduce noise but only because you also then use a slower shutter speed or larger aperture which will mean that the sensor will collect more light. Increasing the ISO will generally increase noise but most of the increase in noise is due to the fact that you also then use a faster shutter speed or smaller aperture will mean that the sensor will collect less light.

Yes, that is correct.

Assuming the above is correct (probably a big assumption so please elaborate or correct as needed), I still have a couple of questions about your recent posts :

1 - Are you saying that on the latest cameras, ISO 800 is the point at which the signal-to-noise ratio of the end result becomes the same between increasing it further in camera or boosting the exposure in the RAW converter? Is this true of all the latest cameras? Does it matter what RAW converter you use? How did you arrive at ISO 800?

It might be dependent on raw converters, but these days most are doing a very good job. For example, it depends on the internal math used. As dynamic range approaches 14+ bits, 16-bit math will be a limitation and will limit results. There will be a need to switch to 32-bit integers or 32-bit floating point. For example, this page:

http://www.clarkvision.com/articles/...shadow.detail/

shows the difference between an older version of photoshop versus a 32-bit floating point raw conversion in ImagesPlus. I need to update the page and see what CS5 delivers.

2 - You state in your example that reducing the ISO to 800 (vs. 3200) and underexposing by two stops gives you two more stops of headroom and dynamic range? How is that the case if I have to boost the exposure in the ISO 800 image by 2 stops in RAW conversion anyway?

See, for example, figure 5b at:
http://www.clarkvision.com/articles/...mance.summary/

look at the 5D Mark IV. At ISO 800 the dynamic range is 11 stops, and then after ISO 1600 is follows a log-linear line down. The A/D converter is digitzing all the way to the sensor noise floor at ISO 1600 and above. So increasing gain is no longer helping to lower the noise floor but is hurting the highlights.

P.S.
Even thought the amount of light collected at the sensor at image capture time is the same in both cases, I would still consider 1/500, f/5.6 ISO 800 and 1/500 f/5.6 ISO 3200 to be two different "exposures" since the resulting RAW file data will be different. Also, since there are situations where ISO 3200 would result in unrecoverable blown highlights and ISO 800 would not, it is hard for me to call them the same "exposure".

I agree this is the traditional view, but it is incorrect. It is incorrect because the amount of light (true exposure) collected is the same. It is no different than changing exposure in the raw converter. If you boost the brightness in the raw conversion by one stop, do you then report a stop higher ISO?

Roger

[Mike Milicia]

For example, this page:

http://www.clarkvision.com/articles/...shadow.detail/

shows the difference between an older version of photoshop versus a 32-bit floating point raw conversion in ImagesPlus. I need to update the page and see what CS5 delivers.

See, for example, figure 5b at:
http://www.clarkvision.com/articles/...mance.summary/

Thanks, Roger! I guess I have some reading to do.

I agree this is the traditional view, but it is incorrect. It is incorrect because the amount of light (true exposure) collected is the same. It is no different than changing exposure in the raw converter. If you boost the brightness in the raw conversion by one stop, do you then report a stop higher ISO?

No, I don't typically indicate changes to the Exposure in RAW conversion at all but it would make good sense to do so and we probably should. You make a good point that reporting it as a higher or lower "effective" ISO would be a sensible way to do it but it might be easier to just indicate the change independently as a fourth parameter.

[Desmond Chan]

Digital (electronic) sensors have only one sensitivity (set by the quantum efficiency).

All right.

So even though in manual mode, one is not "exposing to the right" above ISO 800, one is not losing any data nor is one making the apparent noise worse.

So exposing to the right does not exist even though we "over expose", or in other words there's no what we traditionally called over-exposure at all after "ISO" 800 given the same f stop and shutter speed?

In the most difficult low light photography, astrophotography, amateur astronomers typically use a max ISO of 800 or 1600, because going higher just reduces dynamic range with no additional low light benefit. This has been practiced since the Canon 10D era (when was that, 2003??, or many camera generations ago).

Is that true for camera of other brands, too?

[Desmond Chan]

I agree this is the traditional view, but it is incorrect. It is incorrect because the amount of light (true exposure) collected is the same.

I could agree technically it's incorrect given what you said.

It is no different than changing exposure in the raw converter. If you boost the brightness in the raw conversion by one stop, do you then report a stop higher ISO?

Although most don't report it, but wouldn't that be essentially the same as in doing it through the camera? That's what I've been thinking anyway.

And, how do you teach and communicate with others about exposure in photography? Do you still use the concepts of stops and exposure value? Do f2.8 1/500s ISO 400 still have different exposure values than f2.8 1/500s ISO 1600?

[Roger Clark]

Although most don't report it, but wouldn't that be essentially the same as in doing it through the camera? That's what I've been thinking anyway.

And, how do you teach and communicate with others about exposure in photography? Do you still use the concepts of stops and exposure value? Do f2.8 1/500s ISO 400 still have different exposure values than f2.8 1/500s ISO 1600?

Hi Desmond,

Yes, post processing increasing exposure (above ISO 800) is effectively the same as doing it in camera. But in post, one can decide how much so watch for clipping highlights. Thus one has more control.

"Do f2.8 1/500s ISO 400 still have different exposure values than f2.8 1/500s ISO 1600?"

No, they are the same exposure, as you delivered the same amount of light to the sensor. Over most of the dynamic range, the signal-to-noise ratio will be the same. In the stop closest to zero intensity, the ISO 400 image will have a little more noise because the A/D converter noise will be a factor. In the best cameras, that would be difficult to see in practice.

Roger

[Elliotte Rusty Harold]

Very interesting. Let's see if we can boil this down to practical field rules.

First, which cameras are "modern" cameras? I assume the 7D, 60D, and 1D Mark IV all qualify. How about the 50D, 5D Mark II, and 1D Mark III? Any Rebels?

Second, assuming one has such a modern camera, you are recommending that one set the ISO to 800 and leave it there, correct?

Third, now suppose my camera is at ISO 800 and f/4. How do I determine the correct shutter speed? Do I aim at a neutral gray target and then underexpose two stops from what the camera thinks is the correct exposure?

[Roger Clark]

Very interesting. Let's see if we can boil this down to practical field rules.

First, which cameras are "modern" cameras? I assume the 7D, 60D, and 1D Mark IV all qualify. How about the 50D, 5D Mark II, and 1D Mark III? Any Rebels?

Hi Elliotte,
Yes, the 7D, 60D, and 1dIV qualify, as does the 5DII and T2i. The T1i and 50D maybe, but I need to see the sensor data to be sure. Nikons of similar vintage also qualify. Probably also Sony's (as Nikon uses Sony sensors).

Second, assuming one has such a modern camera, you are recommending that one set the ISO to 800 and leave it there, correct?

Not at all. What I was saying is that if you work in manual and you need an exposure time that would meter at a higher ISO than 800, then you really do not need to increase the ISO. Cap it at ISO 800 or 1600 and if underexposed, correct in post processing.

Third, now suppose my camera is at ISO 800 and f/4. How do I determine the correct shutter speed? Do I aim at a neutral gray target and then underexpose two stops from what the camera thinks is the correct exposure?

Let's try this example. Say you are at your fastest opening (f/4) and you want a shutter speed of 1/1000 second to freeze the action. Imaging a gray card, or similar, your light meter says you need ISO 6400 to get that 1/1000 second. I'm saying that if you are working in manual, set the camera to f/4, 1/1000 second and cap ISO to no more than about 1600. That gives the same exposure and same noise as imaging at ISO 6400, f/4, 1/1000 second, but gives you 2 stops of headroom.

Now if you can get your 1/1000 second f/4 exposure at a lower ISO (say the meter says you get this at ISO 100), then by all means use the lower ISO because the higher ISO (e.g. 800) would cause too much saturation. If you only need 1/1000 second to freeze action, don't go faster as you capture less light resulting in lower signal-to-noise ratios.

Expose to the right is really about longest exposure time and/or fastest f/ratio to deliver the most light, not ISO.

Roger

[Roger Clark]

All right.

So exposing to the right does not exist even though we "over expose", or in other words there's no what we traditionally called over-exposure at all after "ISO" 800 given the same f stop and shutter speed?

Desmond, yes that is correct, given the same f/stop and shutter speed.

Is that true for camera of other brands, too?

Yes. However, some data for the latest Nikons seem to show that the Nikon A/Ds have better performance than the Canons. I say seem to show, because I have not seen the definitive data to prove this (e.g. Nikon raw data may not be as "raw" as one might hope). If the performance is true, then with some Nikons the ISO where one is digitizing the sensor read noise well is below ISO 800, perhaps even ISO 400 (this is good).

Roger

[Elliotte Rusty Harold]

Hi Elliotte,
Let's try this example. Say you are at your fastest opening (f/4) and you want a shutter speed of 1/1000 second to freeze the action. Imaging a gray card, or similar, your light meter says you need ISO 6400 to get that 1/1000 second. I'm saying that if you are working in manual, set the camera to f/4, 1/1000 second and cap ISO to no more than about 1600. That gives the same exposure and same noise as imaging at ISO 6400, f/4, 1/1000 second, but gives you 2 stops of headroom.

OK. Assume I'm imaging at f/4, 1/1000 second and ISO 1600 when the meter suggests 6400. The "2 stops of headroom" mean what exactly? Simply that my raw converter needs to add two stops? Or is there something more?

If I understand it, in this situation the JPEG preview will look way underexposed, but the final image after mechanically adding +2 in the raw converter should be as good or better as the image had I shot at 6400 in the first place.

[Roger Clark]

OK. Assume I'm imaging at f/4, 1/1000 second and ISO 1600 when the meter suggests 6400. The "2 stops of headroom" mean what exactly? Simply that my raw converter needs to add two stops? Or is there something more?

That is correct. But the advantage is that in case there was a highlight that you wanted you could add only 1.5 stps in the raw converter and not saturate those highlights which would have happened if you had set the camera to ISO 6400.

If I understand it, in this situation the JPEG preview will look way underexposed, but the final image after mechanically adding +2 in the raw converter should be as good or better as the image had I shot at 6400 in the first place.

It will not be better regarding signal-to-noise ratio, but if you had highlights, you can get the best "exposure" in post processing and not saturate those highlights. In that case, you can get a better image. If you sim[ply add the two stops, then the image will effectively be the same as if you imaged at ISO 6400.

Here is a situation where I always use this method: night scenes, and in particular, night scenes with stars. For example, with a 30 second exposure of a scene with stars, I'll never go above ISO 1600. The advantage is that more of the brighter stars are not saturated, so retain their colors. In post processing I do not use levels; I use curves to increase brightness. Often I image such scenes at ISO 400 and fix brightness in post.

The disadvantage is that the jpeg and LCD preview is dark. So I may boost ISO to take a test image so I can see I have composition like I want it, then go back to ISO 1600 for the main images.

Another example: low light images at a party. With lights in the room, light levels vary a lot, but if the metering says it will always be above ISO 1600 for a fast enough shutter speed, I go to manual, set aperture and shutter speed, ISO 1600 and image away.

Roger

[Steven Kersting]

Yes, Steven is correct. Changing ISO when f/ratio and exposure time are held constant does not change exposure. ISO does not change the amount of light captured by the sensor, it only changes the level the A/D digitizes.
Roger

LOL, while this is technically correct, it is not what I was trying to say. I was saying that when using auto ISO an overall dark scene will be lightened and an overall light scene will be darkened which is generally what is desired.

Yesterday I went and shot a "car show". Lighting was everything from bright midday sunlight to dark under tent. Light pink Thunderbirds in the sunlight and black Cadillacs under tent. The only setting I changed was switching from CH to CL so I could snap single shots. I took about 40 images and only about 4 were "bad". Those 4 were due to either getting too low a SS that I couldn't handhold (no tripod, not worth going above ISO 1600...suppose I could have taken the CPL off ) and one in which the dynamic range was just excessive (no tripod, not worth HDR/image blending).
(I had many others I am not so happy with either due to BG or reflections (the cars were on top of each other and people were everywhere) which I recognized at the time but I was hoping the "fix" would be easier.) Basically, I was able to spend the afternoon focused on composition/lighting/adjusting the CPL and not worrying about my exposure.

As for the ISO discussion:
*I* though the only "real" ISO for a camera was it's native/base (200 in my case) and not "up to approximately 800." I also understand that for absolute best image quality you should stay with "base ISO" and use full stop increments up to about ISO 800/1600. Not that the camera is changing the amount of light actually captured up to that point, but that it does a great job of adjusting the exposure up to that point.

Given the need I will use higher ISO's in brighter light (1600+), and lower ISO in lower light (1600 and under). In average lighting I use 200-800 without concern.

[Peter Hawrylyshyn]

Roger
In Pane#94 you gave the example of the brown bear against the white water. You set up in Manual mode checking against the white water.

If the brown bear now moved onto a dark grey stone shore against dark green grass (ie - only dark and midtones left in the image), and the lighting & sun angle have not changed - would you adjust the exposure settings (F stop or SS) on your camera to improve the "noise" in darkest parts of the image before taking the image? Most "guru's" say the answer should be yes, but after reading the above, no longer so sure.

How dependent is the answer on your ISO setting?

Could you achieve the same effect in post processing by just adjusting the exposure setting in RAW convertor? Asked another way - how critical is the exposure compensation Artie, Charles, and others recommend while taking the photo given today's sensors and RAW converters?

[Desmond Chan]

Exposure is the amount of light captured by the sensor. When we change f/ratio or exposure time, we change the amount of light on the sensor. Many, many concepts from film days need to be thrown out the window. ISO is one of them. ISO does not change the amount of light the sensor captures.

The amount of light that reaches the sensor or film is always controlled by the size of the aperture and how long the shutter stays open. The sensor or the film just receives light passively. As far as I know, film ISO number is a way to tell us how sensitive the film is to light. With sensor, it's post sensor amplification (is it still "amplification" when we lower the ISO?). So, although the way we understand ISO number may need to be changed, but the use of ISO number could still be useful as a way to communicate to most of us "something has changed" ...at least up to a certain extent

[Steven Kersting]

The amount of light that reaches the sensor or film is always controlled by the size of the aperture and how long the shutter stays open. The sensor or the film just receives light passively. As far as I know, film ISO number is a way to tell us how sensitive the film is to light. With sensor, it's post sensor amplification (is it still "amplification" when we lower the ISO?). So, although the way we understand ISO number may need to be changed, but the use of ISO number could still be useful as a way to communicate to most of us "something has changed" ...at least up to a certain extent

My take on it is it's not amplification if you are at the base ISO, above that yes. Additionally, full stops of ISO only use the A/D converter while "in between" ISO's rely on the processor for "fuzzy math". (Not always that great and no way to prevent it when using auto ISO.)

I do think ISO is relevant information, certainly up to the point of max benefit and further as information to evaluate improper use of ISO or poor exposure SOOC due to ISO selection.

It is odd, but I am much more comfortable using high ISO in good light and lower ISO in poorer light. But I will generally sacrifice SS in poorer light; I haven't found a post solution that really works significantly better than increasing ISO in-camera. Maybe image stacking for certain subjects (not that I've looked very hard).

[Desmond Chan]

I do think ISO is relevant information, certainly up to the point of max benefit and further as information to evaluate improper use of ISO or poor exposure SOOC due to ISO selection.

It is odd, but I am much more comfortable using high ISO in good light and lower ISO in poorer light. But I will generally sacrifice SS in poorer light; I haven't found a post solution that really works significantly better than increasing ISO in-camera. Maybe image stacking for certain subjects (not that I've looked very hard).

"Adjusting" exposure in post-processing is not new. Those who have played around in Photoshop likely knew about it. I suppose the better you're with Photoshop, the better will the results be (isn't that obvious?). Here's an example from Artie (scroll down to Sometimes I Even Impress Myself):

http://www.birdsasart-blog.com/page/2/


A lot of the time, when something happens and I have no time to change the settings (I use manual 99.9% of the time), I'll just shoot (or move the dial fast and hope for the best) and fix it in post. It works for me quite a few times. I guess it's a way to use the sensor's ability to its utmost (like push/pull development of films in the past and all those dodging and burning to squeeze out everything a printing paper can offer). Anyhow, I likely will do some experimentation and see if the result is really as good as Roger suggested.

[WIlliam Maroldo]

I appreciate the feedback from my posts in this thread by all concerned, and apologize to have drifted off-topic, though the subject of ISO seems to have been prominent. It seems that I have made errors in certain assumptions based on interpretation of images I've taken, and it might be useful to explain why (and of course, admit that I was wrong).
My primary errors involve the effect of ISO on digital noise, and the idea that with the same amount of light , a higher ISO with optimization of the S/N ratio can have less noise that an image at a lower ISO without S/N optimization.
I know I wasn't getting this across very well in my posts, but it was the basis for my thinking. The responses to the idea seemed to suggest that it may be a zero sum gain, and although I thought this might be the case, I started wondering if it may actually be worse than that, and the higher ISO would be a poorer choice.
So I tested the hypothesis. I used a tripod mounted Sony A850 with a Sony 70-400G.
The light remained constant.
Two photographs were taken. The first image was at ISO 400, 1/1000 sec at F5.6 that resulted in a histogram that was mostly showing midrange .
With the second image I raised the ISO 2 stops to ISO 1600, with the same 1/000 sec at F5.6. I made sure there was no highlight clipping or blocked blacks in either image and all other variables remained constant. The histogram showed the 2 stop shift to the right and it read mostly in the high range.
In ACR the exposure of the second image was decreased 2 stops, so the final histograms were roughly the same between the 2 images, although the 2nd image had a somewhat less dynamic range as expected.
So did the 2 images have similar amounts of digital noise?
Two factors worth considering; If you increase from ISO 400 to ISO 1600 there will increase in the amount of digtal noise since the signal is being amplified more, yet if the histogram indicates a more advantageous S/N ratio with the higher ISO, as opposed to the lower ISO, there maybe a decrease in digital noise. Would the amount of noise be similar between the two ISOs? This is assuming that S/N ratios increase from left to right on the histogram, and the closer it is to the right the more it is optimized. Then again, you know the saying about assuming?
The results: They optimized S/N ratio does seem to counteract digital noise, but not enough, and there are other factors. There is also the possibility I am misinterpretting the results.
The higher ISO, after exposure correction, showed more noise than the ISO 400 image, though not alot more and certainly less than ISO 1600 where the histogram didn't show the right shift. This would suggest that S/N optimization had an effect. Nonetheless the image did show more noise. As to be expected noise was most noticeable in darker parts of either image. There was also an increase of detail visible in the darker parts of the higher ISO image. So shadows were opened up a bit. Although more detail was visible it was "noisey detail". However the detail resolution in the lighter parts of the image was considerably worse with the higher ISO image. It is my opinion that this alone completely offsets any advatages of the higher ISO.
Increasing the ISO to create a histogram with a optimized S/N ratio , and correcting exposure in ACR, has a small advantage in that it opens up shadows a bit more, and although the digital noise isn't a whole lot worse, there is an obvious loss of detail in other parts of the higher ISO image.
Verdict; I was wrong and it is overall worse than a zero sum gain.
Second question. Roger made a suggestion concerning underexposure and fixing it in post at high ISOs.
Again, the light source remained constant.
I was wondering what would happen if I underexposed an image by 2 stops at ISO 6400, shot the same image with normal exposure, and one with 2 stops overexposure and then compared the images after the underexposed and overexposed images were corrected in ACR. Same light, same subject, everything identical except the exposure. I was not surprised by the results.
The digital noise of the corrected underexposed image is substantially worse than the normally exposed image, and the corrected overexposed image shows considerably less noise than the normally exposed image,
Sony A850 (same sensor as Nikon D3X)
regards~Bill

[Arthur Morris]

Bill, There is an error in your shutter speeds above; you wrote 1/000 sec. Not sure if that would be really fast or really slow. Heck, didn't they always teach us that you cannot divide by zero. A typo for sure.

[WIlliam Maroldo]

Thanks Arthur, fixed the typo. Should be 1/1000. regards ~Bill

[Roger Clark]

I was wondering what would happen if I underexposed an image by 2 stops at ISO 6400, shot the same image with normal exposure, and one with 2 stops overexposure and then compared the images after the underexposed and overexposed images were corrected in ACR. Same light, same subject, everything identical except the exposure. I was not surprised by the results.
The digital noise of the corrected underexposed image is substantially worse than the normally exposed image, and the corrected overexposed image shows considerably less noise than the normally exposed image,

Bill,

A couple of questions. When you changed the settings in ACR to make the images equal, what sliders did you move?

Then, regarding the above quote from your 6400 test, you say you changed exposure. What did you change (shutter speed, f/stop)? What were the settings (f/ratio and shutter speed)?

Roger

[WIlliam Maroldo]

At IS0 6400, obviously in manual mode, I changed the shutter-speed until the meter in the VF showed -2, o, and +2. Actually I didn't keep track of the shutter-speeds, and I thought I had kept the files, but it seems that the overexposed one, which was at 1/5 sec, was all I could find. The f stop was kept at F5.Easy enough to repeat the test, this time keeping track of the shutter speeds. Exposure was adjusted in ACR with the exposure slider only. regards~Bill

[Roger Clark]

At IS0 6400, obviously in manual mode, I changed the shutter-speed until the meter in the VF showed -2, o, and +2. Actually I didn't keep track of the shutter-speeds, and I thought I had kept the files, but it seems that the overexposed one, which was at 1/5 sec, was all I could find. The f stop was kept at F5.Easy enough to repeat the test, this time keeping track of the shutter speeds. Exposure was adjusted in ACR with the exposure slider only. regards~Bill

Hi Bill,
As you changed the shutter speed, you changed the amount of light captured by the sensor, so you changed the signal-to-noise ratio in your iso 6400 test.

In the other test, the results should have been very close to the same IF the ACR exposure slider is linear. I looked at it and it appears close to linear, but seemed slightly non linear. In one test I tried, the green and red histogram peaks changed relative position at I moved the slider, indicating some non-linearity. That might explain the small differences you observed.

Roger

[Roger Clark]

Roger
In Pane#94 you gave the example of the brown bear against the white water. You set up in Manual mode checking against the white water.

If the brown bear now moved onto a dark grey stone shore against dark green grass (ie - only dark and midtones left in the image), and the lighting & sun angle have not changed - would you adjust the exposure settings (F stop or SS) on your camera to improve the "noise" in darkest parts of the image before taking the image? Most "guru's" say the answer should be yes, but after reading the above, no longer so sure.

How dependent is the answer on your ISO setting?

Could you achieve the same effect in post processing by just adjusting the exposure setting in RAW convertor? Asked another way - how critical is the exposure compensation Artie, Charles, and others recommend while taking the photo given today's sensors and RAW converters?

Peter,
Sorry for not answering sooner--too many posts in this thread so I missed it.

It is an interesting question and my answer would be dependent on my intention for the image.

If I wanted the dark bear to appear dark, I would leave the exposure the same.

If I wanted to maximize signal-to-noise ratio, I would lengthen the exposure and expose to the right, assuming the exposure remained short enough to freeze action that I wanted frozen. Then in post processing I could darken the image to show the dark bear, and have better control of noise.

Roger

[Steven Kersting]

As I said earlier, I haven't found (nor really looked for) a method in post that produced significantly better post recovery over just increasing ISO (other than image stacking for certain subjects) I'd love to learn the trick...

The main advantage I see from what Roger noted is the ability to keep a sufficient SS and recover in post, assuming I've hit excessive ISO amplification.....I haven't really explored this yet, but it seems a factor you have "negated"...of course, the significance of this is subject dependent.

My experience is there is much more room for recovery at the overexposed end as opposed to underexposed when working in RAW. And dark areas are easier to "fix" acceptably...

What you are saying tends to support my observations....but I definitely think that camera capabilities really play in here...(many of my methods work for crap w/ a D90/D200)

I'm watching for answers....

[Steven Kersting]

Peter,
Sorry for not answering sooner--too many posts in this thread so I missed it.

It is an interesting question and my answer would be dependent on my intention for the image.

If I wanted the dark bear to appear dark, I would leave the exposure the same.

If I wanted to maximize signal-to-noise ratio, I would lengthen the exposure and expose to the right, assuming the exposure remained short enough to freeze action that I wanted frozen. Then in post processing I could darken the image to show the dark bear, and have better control of noise.

Roger

Aw c'mon, almost no one would keep dark dark (or light light) unless going for an artistic interpretation (high/low key).

I would let auto ISO "fix" the situation for me. EC is only necessary if there is meter confusion or trying to "expose to the right"...choose the right metering mode and EC is not needed.

One thing I can't seem to get across..if I am in A (Av) or S (Tv) modes with the ISO locked down I am still in "full manual" mode. Even if I am in one of those modes with auto ISO settings, if the lighting/scene is constant I am still in "full manual" control. It's when things change unexpectedly that the difference becomes apparent (as in the example given).

ISO gain is always a consideration. And, as I noted earlier, different cameras have different capabilities in dealing with difficult situations