Iliah Borg
Forum Pro
I do it the other way around, I set the exposure, and increase ISO to get the lightness I want. What follows is that with the increase of ISO perceived noise decreases.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
ISO Invariance: an experiment
- Thread starter Tom Axford
- Start date
FingerPainter
Forum Pro
- Messages
- 12,593
- Solutions
- 37
- Reaction score
- 13,643
The context is given here. It mentions nothing about the model applying only to a constant exposure or a constant lightness. It merely say you can use many combinations of the three settings to achieve the same exposure, not that the model only applies if you do.This site is correct to say that ISO speed affects image noise because this is in the context of the exposure triangle....
Here's an example from the popular (but frequently incorrect) site Cambridge in Colour:
In the diagram, the "ISO Speed" is shown as directly affecting (only) image noise. In the text, the claim is that "ISO speed affects image noise", and that "ISO Speed: controls the sensitivity of your camera's sensor to a given amount of light".
First, increasing the ISO while increasing the shutter speed or narrowing the aperture doesn't produce the same exposure, so the text you cite is incorrect.
Second. the diagram and text don't say that it applies only to a constant exposure, or more properly a constant lightness. It says that changing ISO changes noise, full stop.
Third, increasing ISO does not cause an increase in noise, even in the context of constant exposure. If you increase ISO while holding exposure constant, you will get the same or less noisiness.
Last edited:
FingerPainter
Forum Pro
- Messages
- 12,593
- Solutions
- 37
- Reaction score
- 13,643
Read it again, Tom. I didn't cut out any of your comment. I merely placed my comment in the middle of yours.
You are making an assumption about dependence that is not stated by the model: that an autoexpoosure mode is being used.
BTW I do hope you realize that my orignal post ITT was in no way a criticism of your experiment. Rather it was a comment that your experiment correctly shows what is going on, which is in contrast to what the "exposure triangle" model (as usually presented) shows.
FingerPainter
Forum Pro
- Messages
- 12,593
- Solutions
- 37
- Reaction score
- 13,643
Agreed. Now where does the exposure triangle model as presented by CiC or most other popularizers of the model even mention "final image brightness"?
Noisiness (NSR) is an inverse function of the ISO, and an inverse function of the exposure. As either ISO or exposure increases, the nosiness decreases. The relationship between ISO and shot noise is non-causal, so shot noise is not a function of ISO.
Actually, where I learned functions (as opposed to relations) stating that one variable was a function of another implied a causal relationship.
Yes.
Perfectly possible, but not stated in the usual presentation of the "exposure triangle" model.
Iliah Borg
Forum Pro
Tom's experiment is very useful, because among other things it shows the "hidden" problem of not enough precision in the shadows (too heavy rounding), leading to visible non-linearity and channel mismatch after the push. With 14 bit raw it is much less of a problem, by the way. ADC should better be used optimally, full scale.
tinternaut
Veteran Member
I’ve only really experimented in terms of dynamic range, rather than testing invariance. I was using a Sony A7II, so the older Sony sensor. See:
But how do we define sufficient? For full frame, I tend to think in terms of two stops based on the best equipment I have (a modest A7II - an A7RII or Nikon D750/810 owner may think in terms of three or more). So:
I’m not sure absolutely invariant (so called ISOless) sensors exist. That said, a degree of invariance is useful to have when it comes to exposing to protect the highlights.
Loading…
www.dpreview.com
But how do we define sufficient? For full frame, I tend to think in terms of two stops based on the best equipment I have (a modest A7II - an A7RII or Nikon D750/810 owner may think in terms of three or more). So:
- Can I under expose by two stops, at base ISO, with no practical noise penalty compared to shooting at base ISO x 4?
- Does the same hold true if, for example, a low light, hand held shot requires ISO 3200, but I instead shoot at ISO 800 and under expose by two stops (nice to have if you shoot night/Street with very strong highlights)?
I’m not sure absolutely invariant (so called ISOless) sensors exist. That said, a degree of invariance is useful to have when it comes to exposing to protect the highlights.
But why does one apparently get a decrease in noise in the XT2 when going from ISO400 to ISO800?
gollywop
Veteran Member
Agreed.
D
Donald B
Guest
why don't you now show us the same shot that is correctly exposed at base ISO ?
Don
Don
Tom Axford
Forum Pro
Because I didn't take that shot. I thought it would just complicate matters. As you can see there has been quite a lot of discussion as it is. I'm glad I didn't post a third shot!
FingerPainter
Forum Pro
- Messages
- 12,593
- Solutions
- 37
- Reaction score
- 13,643
I'm not sure one actually gets a decrease in noise if one increases ISO from 400 to 800 while dropping exposure one stop. However, one does get less of an increase in noise than one would otherwise expect, and that seems to be because the X-T2 has a dual conversion gain sensor in which the conversion gain switches between ISO 640 and ISO 800.
D
Donald B
Guest
you know why you didn't post one ? because this whole discussion would be a load crap
because I just shot my own experiment with 3 images and guess which one had the lowest noise ?Don
--
Olympus EM5, EM5mk2 my toys.
http://www.dpreview.com/galleries/9412035244
past toys. k100d, k10d,k7,fz5,fz150,500uz,canon G9, Olympus xz1
Last edited:
gollywop
Veteran Member
Someone who fails to see the nature of crap often steps in it. Congratulations.you know why you didn't post one ? because this whole discussion would be a load crap
Naturally, if one is able to achieve ETTR (maximal exposure at base ISO just shy of desired clipping), one will have achieved the best results possible. The purpose, however, of exploiting ISO-invariance deals with shooting situations where, for whatever reason, ETTR is not possible. When this situation occurs, the question arises, once has maximized exposure subject to shooting constraints, whether it is better to enact the necessary additional brightening using in-camera ISO or during raw processing. The ISO-behavior of one's camera answers this question.
Thus there was no need for Tom to include an ETTR shot in his test, because the value of ISO-invariance is necessarily at an exposure less than ETTR. The ETTR exposure is an irrelevant comparison.
Tom's problems were that his test extended over an ISO-variant range of his E-M10 II and that he was definitely straining the ability of most cameras in applying a 5 EV push. I have been able to push my E-M5 II 3-1/2 EV with good success, but I prefer to keep at 3 EV or below.
Meanwhile, it would be nice if you would clean your shoes before entering again.
--
gollywop
I am not a moderator or an official of dpr. My views do not represent, or necessarily reflect, those of dpr.
http://g4.img-dpreview.com/D8A95C7DB3724EC094214B212FB1F2AF.jpg
Last edited:
mamallama
Forum Pro
But one for comparison might have been nice.Someone who fails to see the nature of crap often steps in it. Congratulations.you know why you didn't post one ? because this whole discussion would be a load crap
Naturally, if one is able to achieve ETTR (maximal exposure at base ISO just shy of desired clipping), one will have achieved the best results possible. The purpose, however, of exploiting ISO-invariance deals with shooting situations where, for whatever reason, ETTR is not possible. When this situation occurs, the question arises, once has maximized exposure subject to shooting constraints, whether it is better to enact the necessary additional brightening using in-camera ISO or during raw processing. The ISO-behavior of one's camera answers this question.
Thus there was no need for Tom to include an ETTR shot in his test, because the value of ISO-invariance is necessarily at an exposure less than ETTR. The ETTR exposure is an irrelevant comparison.
When is it not possible?Someone who fails to see the nature of crap often steps in it. Congratulations.you know why you didn't post one ? because this whole discussion would be a load crap
Naturally, if one is able to achieve ETTR (maximal exposure at base ISO just shy of desired clipping), one will have achieved the best results possible. The purpose, however, of exploiting ISO-invariance deals with shooting situations where, for whatever reason, ETTR is not possible.
Hmm, perhaps the better way to state it, yes.
When the shutter speed necessary to stop motion blur or camera shake is too fast to allow enough light on the sensor to ETTR or when the aperture necessary to achieve desired depth of field is too small to allow enough light on the sensor to ETTR.When is it not possible?Someone who fails to see the nature of crap often steps in it. Congratulations.you know why you didn't post one ? because this whole discussion would be a load crap
Naturally, if one is able to achieve ETTR (maximal exposure at base ISO just shy of desired clipping), one will have achieved the best results possible. The purpose, however, of exploiting ISO-invariance deals with shooting situations where, for whatever reason, ETTR is not possible.
Doug Haag
Senior Member
I was the first respondent to this thread and have therefore been automatically alerted to all the subsequent postings without additional input from myself. I have read all these additional posts.
But now that the kooks have arrived (not you Golly), I think I'll click on the setting to eliminate the alerts of further posts.
But now that the kooks have arrived (not you Golly), I think I'll click on the setting to eliminate the alerts of further posts.
gollywop
Veteran Member
Yes, irrelevancies of all sorts, even posts, are always nice. ;-)But one for comparison might have been nice.Someone who fails to see the nature of crap often steps in it. Congratulations.you know why you didn't post one ? because this whole discussion would be a load crap
Naturally, if one is able to achieve ETTR (maximal exposure at base ISO just shy of desired clipping), one will have achieved the best results possible. The purpose, however, of exploiting ISO-invariance deals with shooting situations where, for whatever reason, ETTR is not possible. When this situation occurs, the question arises, once has maximized exposure subject to shooting constraints, whether it is better to enact the necessary additional brightening using in-camera ISO or during raw processing. The ISO-behavior of one's camera answers this question.
Thus there was no need for Tom to include an ETTR shot in his test, because the value of ISO-invariance is necessarily at an exposure less than ETTR. The ETTR exposure is an irrelevant comparison.
confused circle
Senior Member
- Messages
- 1,932
- Reaction score
- 1,554
I would think bif in certain conditions or evening/indoor sports would be great examples.When the shutter speed necessary to stop motion blur or camera shake is too fast to allow enough light on the sensor to ETTR or when the aperture necessary to achieve desired depth of field is too small to allow enough light on the sensor to ETTR.When is it not possible?Someone who fails to see the nature of crap often steps in it. Congratulations.you know why you didn't post one ? because this whole discussion would be a load crap
Naturally, if one is able to achieve ETTR (maximal exposure at base ISO just shy of desired clipping), one will have achieved the best results possible. The purpose, however, of exploiting ISO-invariance deals with shooting situations where, for whatever reason, ETTR is not possible.
Similar threads
Keyboard shortcuts
- f
- Forum