Some observations on "high" ISO

Joined
Aug 4, 2012
Messages
1,793
Location
SW England
Real Name
Rob
I thought I would share something I have only recently realised myself, about shooting with high(ish) ISO.
Disclaimer: I have only tested this with my E-M10ii, shooting raw. My conclusions from my observations may be erroneous.

If you shoot a given scene, correctly exposed, at ISO3200, then you seem to get basically identical "IQ" (noise, colour, details etc) if you were to shoot the *same* shutter speed and aperture at a lower ISO (right down to 400) and then "boost" the exposure in post (say, in Lightroom). This does not hold though if you underexpose down to ISO 200, as the colours then go "off" for some reason.

To get the expected better IQ at a lower ISO, you have to shoot correctly exposed, so using a slower SS for example. So the improvement comes from having more light, not from the ISO setting per se. I imagine this implies most of the visible noise is shot noise, not noise related to the performance of the electronics etc.

I used to think, for example when photographing birds where I want fast SS, that I would be better shooting at a lower ISO (underexposed) and then pushing exposure in post, as if there was something intrinsically "better" about lower ISO shots.
It turns out this seems to be a fallacy. You can't get the "benefit" of the lower ISO without basically taking in more light. It seems I'm better off sticking to, say, 3200, to get faster SS (at least with birds). Obviously shooting in brighter light potentially gives a real IQ boost but I can't necessarily control that.

I realise that some people reading this will already know it or think it's obvious. I imagine some people may think it's wrong. But I just thought some might think it interesting, or useful, or worth debating.

I've read a lot about the meaning of "ISO" in digital photography over the years, and most of it I found baffling. Recently, looking into astrophotography I read about ISO invariance, which was also baffling, maybe more so.

Anyway, the pics below hopefully illustrate my point. They are 100% crops of a shady spot outside under a shrub, taken a few seconds apart, on a tripod, with light conditions not really varying (bright sun blue sky). I'd be interested to know what others think. Also - I am fully aware that these kinds of home experiments are not really very scientific….

There are some caveats. For example I assume (from all the graphs you can find on the web) that stuff like dynamic range might vary with ISO setting, which might matter in some situations.

Just to be clear - I've shot m43 for nine years and I'm very happy with it - I'm not about to change system or sensor size. I just offer this here as an observation about the apparent behaviour of my kit.😀

My "take-home", for myself, is that if I need a particular aperture and SS, then there's no real intrinsic downside in higher ISO for many types of picture - so maybe I should start using auto-ISO, which I've not really used up to now. Finally, I imagine there is an ISO value above which this might fall apart, so for now I'll probably stick to 3200 as a general limit.

OM287431.jpg
Subscribe to see EXIF info for this image (if available)

ISO 200 as shot
OM287436.jpg
Subscribe to see EXIF info for this image (if available)

ISO 3200 as shot
OM287437.jpg
Subscribe to see EXIF info for this image (if available)

ISO 1600 exposure lifted +1 in LR
OM287438.jpg
Subscribe to see EXIF info for this image (if available)

ISO 800 exposure lifted +2 in LR
OM287439.jpg
Subscribe to see EXIF info for this image (if available)

ISO 400 exposure lifted +3 in LR
 
Joined
Nov 6, 2017
Messages
1,235
Location
Thessaly, Greece
It may be some slight aperture mechanism inaccuracy so that the selected aperture difference of f/5 instead of f/6.3 does not correctly coincide with the selected ss of 1/4s vs 1/6.25s so you basically get a bit underexposed (1/4? stop) at ISO200 vs the rest of the shots, hence the difference in color rendition by the final jpeg engine renditions we see here - or the lens behavior changes ever so slightly with the slightest selected aperture difference or so…
 
Joined
Aug 4, 2012
Messages
1,793
Location
SW England
Real Name
Rob
It may be some slight aperture mechanism inaccuracy so that the selected aperture difference of f/5 instead of f/6.3 does not correctly coincide with the selected ss of 1/4s vs 1/6.25s so you basically get a bit underexposed (1/4? stop) at ISO200 vs the rest of the shots, hence the difference in color rendition by the final jpeg engine renditions we see here - or the lens behavior changes ever so slightly with the slightest selected aperture difference or so…
Yes - I think I changed the aperture on the ISO200 shot by accident - annoyingly didn't realise until afterwards...
 

BDR-529

Mu-43 Top Veteran
Joined
Jun 27, 2020
Messages
980
The fun fact here is that there is no such thing as ISO in digital photography. It's meaningless number and kept just for the old times sake.

This is not to say that the setting called ISO does not do anything but what it actually does will not have any impact whatsoever on sensor sensitivity to incoming photons (as it did in the analog film world)

No matter how much you turn the ISO knob, photosites will capture exactly the same amount of incoming photons and turn them into charge just the same. Only aperture and exposure time will have impact on this actual data photosites can collect.

ISO setting will, however, have impact on post-processing because it will set the analog amplification when charge is collected from photosites and fed to A/D converter.

Some very high ISO settings will eventually saturate this analog pre-amp and they are "simulated" which means that ISO value is just written on the RAW file as an instruction to image viewer to show this file brighter than it actually is.

Camera does actually force a *lot* of computational improvements on true "raw" data it gets from photosites when RAW file is created so ISO setting might produce an illusion that it does something even when values are already in the simulated range.
 
Last edited:

archaeopteryx

Gambian sidling bush
Joined
Feb 25, 2017
Messages
1,720
I've read a lot about the meaning of "ISO" in digital photography over the years, and most of it I found baffling. Recently, looking into astrophotography I read about ISO invariance, which was also baffling, maybe more so.
What you're describing in your observations is ISO invariance, both the invariant range and the limits at which invariance no longer applies.
 

PakkyT

Mu-43 Hall of Famer
Joined
Jun 20, 2015
Messages
4,682
Location
Massachusetts, USA
To get the expected better IQ at a lower ISO, you have to shoot correctly exposed, so using a slower SS for example. So the improvement comes from having more light, not from the ISO setting per se.
Yes, the moral of your story is the more actual light you capture with the sensor itself the better the image will be regardless of the ISO used. Watch your histogram.
  • If you are clipping on the right (highlights) and if you are not using the base ISO, turn the ISO down first. You didn't need to set the ISO that high to begin with.
  • If you are clipping on the left (shadows) then first adjust your aperture or shutter to push the histogram right. ISO should be the last thing to adjust up after you have maximized the shutter and aperture you can use*.
  • If (when you first lift the camera and haven't done either of the two steps above) everything is within the histogram, if you are not using the base ISO the first thing to check is can you turn down the ISO (shift histogram left) and if needed compensate back to the right with shutter/aperture without clipping. Repeat as much as you can until you can no longer shift the histogram right with shutter and aperture*.
* The above statements imply adjusting the aperture/shutter as much as you can while also keeping in mind the depth of field or the subject motion & camera shake tolerances desired by the photographer. e.g. yes the camera can be set to 8 seconds of shutter but not if you are hand holding it so a setting of 1s might be the practical limit of what you "can" set it for.
 

BPCS

Mu-43 Regular
Joined
Oct 15, 2017
Messages
181
I have followed these sort of discussions for a few years, and for our current m43 sensors, have noted some common observations that I find noteworthy:

1. A picture at 400 ISO and lifted a few stops in processing, appears to look better than if the camera was set at 200 ISO and lifted a few stops (the 200 ISO picture will show more noise and banding in the shadows).
2. A camera setting of higher than 1600 ISO will look more noisy if lifted in processing than if the setting is between 400 and 1600 ISO... so potentially the ISO on auto should be set to float between 400 and 1600 ISO.
3. If you know you are going to get noticeable underexposure, say due to the need for very high shutter speeds, limit the ISO setting to 1600...but only if the exposure (lightness or darkness of image preview) looks correct at up to ISO 6400... if the image still looks dark at ISO 6400, then use ISO 6400 and expect a very noisy picture regardless, but better than lifting the ISO 1600 shot in processing.
4. Shooting at 250 ISO and exposing to the right without clipping highlights, will give the best quality file.
5. Setting the camera to between 400 and 1250 ISO and exposing to the right without clipping highlights, can give more highlight detail than setting the camera to below 400 ISO (at the expense of a flatter and duller looking picture).

Having observed all of the above, I generally don't worry too much about any of it as I find DXO PL4 Elite and RAW files give me a great shot most of the time, provided I don't blow highlights badly or expect more than 2 or 3 stops lifting of the shadows... sunrise/ sunset shots at ISO 400 and a bias towards the correct exposure of the highlights, does look nicer than at below ISO 400. I also use ISO 400 if shooting rapidly and not able to watch highlight clipping closely enough. Using even ISO 64 or 100 works fine if the scene is flatter or duller in contrast range. Astro at 1600 ISO and then lifting exposure in processing, if necessary, is best.
 

archaeopteryx

Gambian sidling bush
Joined
Feb 25, 2017
Messages
1,720
I have followed these sort of discussions for a few years, and for our current m43 sensors, have noted some common observations that I find noteworthy
These appear to be descriptions of dynamic range shadow improvement (1, 2, 3) and dynamic range (4, 5). I see support for the shadow improvement claims for certain bodies in the available data but am failing to identify support for the dynamic range claims.

Could you be more specific?
 

BPCS

Mu-43 Regular
Joined
Oct 15, 2017
Messages
181
These appear to be descriptions of dynamic range shadow improvement (1, 2, 3) and dynamic range (4, 5). I see support for the shadow improvement claims for certain bodies in the available data but am failing to identify support for the dynamic range claims.

Could you be more specific?
I can't really... my observation is practical, along with various demonstrations I have seen done by others. In practice, the comments at the end that I made are where I sometimes remember to try a different ISO setting and sometimes I see a better result... less banding in shadows or better highlight detail. EM1.2 and X.
 

Latest threads

Links on this page may be to our affiliates. Sales through affiliate links may benefit this site.
Mu-43 is a fan site and not associated with Olympus, Panasonic, or other manufacturers mentioned on this site.
Forum post reactions by Twemoji: https://github.com/twitter/twemoji
Forum GIFs powered by GIPHY: https://giphy.com/
Copyright © Amin Forums, LLC
Top Bottom