DPChallenge: A Digital Photography Contest You are not logged in. (log in or register
 

DPChallenge Forums >> Hardware and Software >> ISO vs Exposure Correction
Pages:  
Showing posts 1 - 17 of 17, (reverse)
AuthorThread
07/30/2006 12:39:14 PM · #1
I was wondering whether i should ever go above ISO 100 on my D200 if i'm shooting in RAW...

What's the difference between boosting the in-camera ISO and doing exposure correction in post-processing on a 16-bit RAW?
07/30/2006 12:45:33 PM · #2
Kirbic can explain this better than I can, but with our sensors the more "to the right" you expose the more information the sensor is processing. So rescuing an underexposed shot by "adding exposure" in RAW processing doesn't work anywhere near as well as correctly exposing in the first place. On the other hand, using RAW "exposure compensation" to bring slightly blown highlights down to a usable value seems to work very well for me, as long as they are not totally out-of-gamut: say a stop and a half or less "overexposed".

For whatever that's worth; this is just what I've found in my fiddling with RAW and exposure.

R.
07/30/2006 02:31:42 PM · #3
Obviously it is possible to achieve a correct exposure whether using iso 100 or iso 200. One should always strive for an exposure that makes use of the maximum amount of image data, that is, the histogram should span the full width of the scale. Shooting conditions may well override the choice, but in my experience, a lower iso will yield lower image noise...to a point. That point is where exposure time begins to cause more noise than using a higher iso setting.
Exposure correction in post processing can usually yield an improvement in image quality, but I try to avoid extreme corrections of one or more full stops.
All that said, Labuda, your photographic skills exceed those of most photographers on this site!
07/30/2006 02:39:49 PM · #4
it's better to use the higher iso and not have to push the picture later in editing.
07/30/2006 03:13:02 PM · #5
Pushing the image will increase the noise, just like pushing film will incrase the grain.
07/30/2006 04:24:21 PM · #6
To clear things up... i know that exposing for longer reduces noise for a given ISO, but my question is for situations where you're limited by shutter speed and your aperture is already wide open. In that case, you can boost your ISO until you have a proper exposure, or keep it at ISO 100 and do exposure compensation later on.

Im actually wondering about how the electronics work just after the ccd. Doas a boost in ISO act on the analog part of the circuit near the CCD, for example by doubling the voltage applied to the pixels? Or does it simply take the signal from the CCD and amplify it just before converting it to a digital signal.

If it only amplifies the signal just before conversion, it's not a big deal to use exposure compensation, all you're basically doing is loosing on bit depth. For example, taking a shot at ISO 100 instead of ISO 200 (with other parameters fixed) would mean that you're only using 15 bits out of the 16 bits available (assuming your camera outputs 16 bits), not a big deal at all. Meanwhile, you're preventing the possibility of clipping your pixels if there is a bright spot in your image.

However, if boosting your ISO can actually reduce some form on inherent CCD electronic noise (*) and improve your signal-to-noise ratio, then i'll have to reconsider my statement.

Please let me know!

(*) edit: or noise which is created by the electronics between your CCD and your analog to digital converter. This noise would surely have less impact on a larger signal (ISO200) than on a small signal (ISO100) once converted to digital.

Message edited by author 2006-07-30 16:28:29.
07/30/2006 04:28:24 PM · #7
So, basically speaking, shooting RAW, it is better to overexpose than underexpose as long as it's not by more than 1 stop or so?
07/30/2006 04:30:23 PM · #8
Originally posted by MrEd:

So, basically speaking, shooting RAW, it is better to overexpose than underexpose as long as it's not by more than 1 stop or so?


Again, i know that over exposure reduces your signal to noise ratio.

But the situation i'm inquiring about is when your aperture and shutter speed are fixed... then comes the decision whether you should bother boosting your iso, or simply boost exposure in post-processing.
07/30/2006 05:18:36 PM · #9
As I understand it, ISO adjusts the sensitivity of the sensor not the gain afterward. That is, the adjustment tells the sensor at 100 ISO that N number of photons is required to record x value. For the sake of example, let's call 100 photons to record at middle grey -- although the actual value is likely quite different.

At ISO 200, the setting tells the sensor it needs half as many photons (50 to continue our example) to record the same value, middle grey. The original 100 photons in the example would then be recorded at 1 stop above middle grey.

The most used analogy I've seen is that each photocell of the sensor is like a bucket and ISO changes the size of the bucket. The higher the ISO the smaller the bucket and the faster it fills up.

To directly answer your question. No, shooting at a lower ISO and boosting the exposure in RAW conversion is not the same as shooting with a higher ISO. To verify this, shoot two images with aperture and shutter speed locked but with 100 and 200 ISO repectively. Open them in a RAW converter and look at them. Since the raw image is the raw sensor data the images would be the same if ISO was added during in-camera post-processing -- as that is not done to a RAW image.

But, instead the ISO 200 image will be brighter and use more of the histogram. And by using more of the histogram the image gains the advantage of using more bits to store the data in. In some cases, the noise gained by higher ISO stored with more bits is less than the noise of a longer shutter speed -- but I'm sure you already knew this last bit.

Originally posted by labuda:

... If it only amplifies the signal just before conversion, it's not a big deal to use exposure compensation, all you're basically doing is loosing on bit depth. For example, taking a shot at ISO 100 instead of ISO 200 (with other parameters fixed) would mean that you're only using 15 bits out of the 16 bits available (assuming your camera outputs 16 bits), not a big deal at all. ...

The loss of that single bit cuts the available storage values in half. Granted at 16-bits there would still be several thousand values left, but it is still a huge loss.

David
07/30/2006 05:48:18 PM · #10
Originally posted by David.C:

As I understand it, ISO adjusts the sensitivity of the sensor not the gain afterward. That is, the adjustment tells the sensor at 100 ISO that N number of photons is required to record x value. For the sake of example, let's call 100 photons to record at middle grey -- although the actual value is likely quite different.

At ISO 200, the setting tells the sensor it needs half as many photons (50 to continue our example) to record the same value, middle grey. The original 100 photons in the example would then be recorded at 1 stop above middle grey.

The most used analogy I've seen is that each photocell of the sensor is like a bucket and ISO changes the size of the bucket. The higher the ISO the smaller the bucket and the faster it fills up.

David


Partly right, David. But with exception of the charge transfer electronics (CCD and CMOS 'bucket brigade' or CAM devices), the sensor is entirely a passive device and it's sensitivity is not adjustable. ISO is adjusted in post sensor electronics. There is no built in mechanism to reduce or increase the number of electrons produced by incident photons. There is a mechanism to 'sweep' or clear the pixel sites (transfer zero charge in serial fashion), and this is usually done just before the shutter is opened.
There is a third type of device marketed a few years ago by CIDTEC and also by GE. That device operated in depletion mode rather than accumulation mode, that is, incident photons removed charge from the pixel sites rather than increasing charge. The advantage was that it would not 'bloom' under intense light conditions.
07/30/2006 06:14:16 PM · #11
Originally posted by labuda:

To clear things up... i know that exposing for longer reduces noise for a given ISO

Actually the shorter the exposure the lower the noise.

edit: that's assuming the exposure was the same at that ISO.

Message edited by author 2006-07-30 18:14:55.
07/30/2006 06:25:33 PM · #12
Originally posted by kyebosh:

Originally posted by labuda:

To clear things up... i know that exposing for longer reduces noise for a given ISO

Actually the shorter the exposure the lower the noise.

edit: that's assuming the exposure was the same at that ISO.


Of course, but the signal-to-noise ratio, which is all that matters in the end, increases considerably with increased exposure length for a given ISO (the lower the ISO, the better this works)

When i say "noise", i actually mean the "noise/signal"

Message edited by author 2006-07-30 18:26:23.
07/30/2006 07:45:19 PM · #13
i'm almost completely sure that you're not either saying it clearly enough for me to understand or not correct (or i'm wrong :-D )

If you have two photos with the same exact exposure, with different shutter lenghts, and varying the aperture to compensate the one with the shorter exposure time should have a better s/n ratio.

example:
shot 1: AV F8 shutter speed 1/500
shot 2: AV F5.6 shutter speed 1/1000

shot 2 should have a better s/n ratio.

Message edited by author 2006-07-30 19:45:41.
07/30/2006 10:49:20 PM · #14
Originally posted by kyebosh:

i'm almost completely sure that you're not either saying it clearly enough for me to understand or not correct (or i'm wrong :-D )

If you have two photos with the same exact exposure, with different shutter lenghts, and varying the aperture to compensate the one with the shorter exposure time should have a better s/n ratio.

example:
shot 1: AV F8 shutter speed 1/500
shot 2: AV F5.6 shutter speed 1/1000

shot 2 should have a better s/n ratio.


Well, that's because I never mentioned anything about my aperture changing. So that's where the misunderstanding is.

You're right shot 2 would be better, but i think that the effect would be small even for high ISO, and virtually inexistant for ISO 100.

I started this thread with a fixed aperture and shutter speed in mind. I want to know the effect that ISO has on the noise of the image, vs exposure compensation in post-processing...
07/30/2006 11:08:32 PM · #15
If you're talking about going from ISO 100, I would definitely boost ISO rather than underexposing and doing EV Comp later. On my D70 I can barely tell the difference between ISO 200 and 500, and I'm sure the D200 is even better. And there's always NeatImage to remove that little bit of noise.
07/30/2006 11:09:12 PM · #16
To address Aleks's original question, it's quite a bit better to boost the ISO than to underxpose. The bit-depth problem alone is a substantial one. When we turn up the ISO, we're increasing the gain of the analog amplifier section. Although we're amplifying noise as well as signal, at least we are boosting the signal at the earliest possible stage, thus limiting signal degradation at later stages.
Any astrophotographer who shoots digital will tell you that their experience confirms the above. It's often much better to set ISO for 400 or 800 and obtain a greater signal than to set it for 100 and stretch the historgam severely in post.
07/31/2006 03:00:59 PM · #17
AFIK, setting the ISO adjuststs the amplification between the photosite and the ADC.

Remember that 17% grey is percieved at about halfway between black and white, so that about 17% of the levels the ADC outputs are below "average" brighness. (We see luminosity in an exponential manner, but the ADC sees it linearly.)

By setting the ISO higher, you will have more levels between 0 and 17% grey. If you rely on postprocessing to do this, you will not have as many levels to expand. Also, the ADC will put a lot of the dimmer sections into its lowest level (it will "clip" those levels).

Pages:  
Current Server Time: 01/02/2026 11:59:04 PM

Please log in or register to post to the forums.


Home - Challenges - Community - League - Photos - Cameras - Lenses - Learn - Help - Terms of Use - Privacy - Top ^
DPChallenge, and website content and design, Copyright © 2001-2026 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 01/02/2026 11:59:04 PM EST.