Peter Connan

ISO Invariance

24 posts in this topic

Posted (edited)

This is a term that caught my attention recently, and I think the possible ramifications are quite interesting.

 

Effectively, the theory is that with some modern sensors, the noise added as a result of increasing the ISO when taking the photo is similar to the result of lifting exposure in post-processing afterwards.

 

I recently read an article where the writer claimed that, for star photography, it is better to shoot under-exposed at a lower ISO, because at lower ISO's the camera captures more dynamic range and the resulting noise will be the same as the shutter speed and aperture are basically set in stone anyway.

 

I recently experimented with this a little bit. The following images were taken at the same shutter speed and aperture, under similar if not exactly the same lighting conditions (some of the foreground lighting was a campfire), one at ISO 8000 and the other at ISO800, and had the exposure lifted by 3 1/3 stops. The results look as follows:

 

ISO8000:

post-24763-0-52391000-1471800581_thumb.jpg

 

ISO800:

post-24763-0-60875200-1471800545_thumb.jpg

 

A crop of the highlight area:

post-24763-0-62795600-1471800630_thumb.jpg

 

post-24763-0-28842100-1471800608_thumb.jpg

 

And the shadow area:

post-24763-0-38666000-1471800672_thumb.jpg

 

post-24763-0-54103300-1471800654_thumb.jpg

 

In short, I can't really see much difference, although maybe the ISO800 image has a little bit more detail in the shadow areas.

Edited by Peter Connan
4 people like this

Share this post


Link to post
Share on other sites

That's a really interesting comparison @@Peter Connan, myself I think the ISO800 shot actually looks better once zoomed, which is surprising to me. One question is what type of camera were you using? It's widely thought that Nikon/Sony sensors have better dynamic range than Canon's, I wonder if the result would be similar across those three types of bodies?

Share this post


Link to post
Share on other sites

...and further to Zubbie's comment.

 

Was there any reason you used a 3 1/3 stop difference? Was the post processing exactly the same for both images (except for the 3 1/3 stop adjustment)?

Share this post


Link to post
Share on other sites

great photos @@Peter Connan

I can't really see the difference either?

 

... But... Does your ISO dial up to 11? ;)

1 person likes this

Share this post


Link to post
Share on other sites

@@Zubbie15, this is a Nikon D750 (with Tamron 15-30mm f2.8 if you're interested).

 

@@Geoff, apart from raw conversion and default adjustments in Lightroom (except that i completely removed the default noise reduction), the ISO8000 image is un-edited. The ISO800 image was treated exactly the same except for the 3 1/3rd-stop exposure increase. The reason for the 3 1/3rd stop difference is that the article i mentioned before recommended shooting Nikons at ISO800. He recommends ISO1600 for the Canons. I wish I had tried a few at ISO100 as well, and will try that at the next opportunity. ISO8000 gave me what i thought was the correct exposure.

 

For reference, if I was really editing the ISO800 image for normal use, i would have lifted the exposure less and then lifted the shadows, which would in theory reduce noise in most areas of the image.

 

@@ZaminOz, my camera's highest "normal" ISO is 12800, and then there is a range of "high" ISO's, which i suspect is useful only to James Bond and the papparazzi.

 

I do need to point out though that not all camera sensors are supposed to be ISO invariant. In fact I believe there are only a handful of contenders.

1 person likes this

Share this post


Link to post
Share on other sites

It's a very interesting (and good) result. Hmmm. Who'd have thought?

 

Is the idea that there is so much more dynamic range in the shadows with the best modorn sensors that it is no longer quite as necessary to expose to the right?

 

Not sure I would trust it, but I suppose that would be the point of trying it.

1 person likes this

Share this post


Link to post
Share on other sites

Posted (edited)

@@pault, now you are getting to the meat of the issue.

 

As stated before, if your sensor is effectively ISO invariant, then the noise will be similar or the same irrespective of whether you lifted exposure or lifted ISO.

 

As to the dynamic range, there is no doubt that this is best at a camera's base ISO. However, you lose DR both by increasing ISO and by lifting exposure, and it seems there is still some argument about which method gives the best result.

 

This single test seems to indicate that results may be better if a low ISO is used, but i would not change methods on the basis of one test.

 

However, if it does turn out that DR is better when shooting at low ISO and lifting exposure later, then it may radically change the way we shoot.

 

In short, it will still be best to get the exposure as close to correct as possible by exposing longer or reducing the aperture, but when this is not possible for whatever reason, it may then be best to change to full manual mode, set the desired aperture and shutter speed and base ISO and ignore exposure completely.

 

The problem with this method at the moment is that firstly it becomes difficult to judge the possible result as the replay on the LCD will be quite dark and you will have no idea of how much noise will be introduced later (and thus whether or not it's even worth taking the shot) and secondly it makes post-processing a bit more laborious...

 

@@Morkel Erasmus, @@Safaridude and especially @@twaffle, i would love to hear your views on this?

Edited by Peter Connan
2 people like this

Share this post


Link to post
Share on other sites

@@Peter Connan I haven't had a chance to consider your tests very well but I have to say that in very difficult lighting I will underexpose by 1 or 2 stops to get the ss and aperture that I want knowing that modern post production can work wonders. I think it depends a lot on the camera you have as well. The 1DX handles under exposure pretty well, my old 7D was dreadful.

 

I think the key is to know your cameras capabilities at extremes and also know how to post process very well which includes good knowledge of the various tools available.

 

Good discussion and testing though. I'll try to look closer at your photo tests.

1 person likes this

Share this post


Link to post
Share on other sites

Thank you very much @@twaffle.

1 person likes this

Share this post


Link to post
Share on other sites

Very interesting topic, @@Peter Connan ! Would you be so kind and post a link to the original article, plus the list of (Nikon) cameras that supposed to be ISO invariant? Thanks!

Share this post


Link to post
Share on other sites

@@xelas, that was remiss of me.

 

Here is the original star-scape article: http://www.clarkvision.com/articles/nightscapes/

 

However, he does not mention ISO Invariance. That came from a couple of articles on DXOMark, which I can't seem to find right now.

 

I don't know if there is such a list, but I suspect that at the very least the D610, D750, D810 and D5 are on it.

 

I also nbelieve the D7200, D500, D4s and Df must be very close if not invariant.

Share this post


Link to post
Share on other sites

Thanks Peter!

1 person likes this

Share this post


Link to post
Share on other sites

This is Thom Hogan's take on it: http://www.dslrbodies.com/cameras/the-d5d500-blog/iso-variance.html

 

I'm not absolutely sure he is talking about exactly the same thing but I think he is. His discussion is about Nikon but I wonder if there is something in his suggestion of using full ISO 'increments' (100,200,400,800 etc) rather than any intermediate points?

On the other hand we seem to be looking at fairly extreme photographic demands and certainly for me there are lots of other issues of technique to sort before I need woryy about this :)

1 person likes this

Share this post


Link to post
Share on other sites

Interesting thoughts and concept. I have read about it but I do prefer the look of the Milky Way in camera at higher ISOs...I would just shoot ISO-3200, that is somewhere between your 800 and 8000 parameter :)

I do agree with modern sensors it's better to underexpose and recover later (to a degree). Some photo tour operators teach their clients to shoot at -1.7 EV all the time and protect highlights, but I can't recommend following such a blanket approach for all exposure/lighting situations and all wildlife sightings eh. I mostly shoot -0.3 EV and still use auto ISO very effectively in manual mode, not seeing any issue with the intermediate ISO values on my Nikons.

3 people like this

Share this post


Link to post
Share on other sites

Thanks, @@KaingU Lodge for the link, And @@Morkel Erasmus thank you also to put my mind at rest! -0.3 EV underexposed is what we are using also. With AutoISO.

Share this post


Link to post
Share on other sites

Thanks for all the input guys.

 

@@Morkel Erasmus, I read somewhere that AutoISO negatively affects frame rate on the faster cameras, but have not noticed a difference on my (relatively slower) cameras. Have you perhaps noticed a difference with the faster cameras?

1 person likes this

Share this post


Link to post
Share on other sites

@@Peter Connan nope - but then again I don't focus on counting the frames when firing my D3s at 9fps :)

1 person likes this

Share this post


Link to post
Share on other sites

Posted (edited)

@@Morkel Erasmus your recent blog post (which I just saw on Facebook) seems to be a very good example of utilizing ISO invariance. Even if you don't call it such :) The ability to pull up the shadows in that elephant/sun photo without introducing a lot of noise (by starting out with a low ISO value, in this case 250) is exactly the point. I think! And this is something that can be done to good effect only with certain sensors (not sure which camera you used for these, as there is no EXIF.) But with the Nikon D800/810 and now the D500 it works very well. (Probably other models too...have no idea about Canon.)

 

http://blog.morkelerasmus.com/2016/08/trust-your-histogram.html

Edited by janzin
2 people like this

Share this post


Link to post
Share on other sites

Posted (edited)

Here's an article that lists some of the cameras that are best at this. Notice that no Canon's are listed.

 

http://improvephotography.com/34818/iso-invariance/

Edited by janzin
2 people like this

Share this post


Link to post
Share on other sites

@janzin Not a full list (so maybe po' Canon users have something now) but good find.

1 person likes this

Share this post


Link to post
Share on other sites

Apparently the latest canon cameras have different sensor tech. No idea if that will allow them to be ISO invariant, but the 80D, 1DXII and now 5DMKIV are apparently with this new technology allowing more dynamic range than previous.

Share this post


Link to post
Share on other sites

@janzin Not a full list (so maybe po' Canon users have something now) but good find.

Yes, this article is a year old so newer cameras aren't included. The Nikon D500 should be on the YES list (although I've yet to really try it out, I've read that it supports it.)

Share this post


Link to post
Share on other sites

@@janzin cheers - I actually used a D800, will amend the post to reflect that :)

1 person likes this

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now

  • Recently Browsing   0 members

    No registered users viewing this page.


© 2006 - 2017 www.safaritalk.net - Talking Safaris and African Wildlife Conservation since 2006. Passionate about Africa.