ISO - Waste of time when shooting RAW?

Fraser Euan White

Fraser Euan White

Suspended / Banned
Messages
3,062
Name
Fraser White
Edit My Images
Yes
There is an interesting thread on another forum where a contributor is arguing that changing ISO is unnecessary. His argument appears to be very sound in that shooting a picture at the base ISO of the camera usually gives the least noise and the best dynamic range. He then sets his shutter speed and aperture manually to suit the scene he is photographing and doesn't worry what the exposure 'in camera' looks like (most look virtually black and are very under exposed).

The exposure & white balance are then edited in post and look great.

He has faced arguments over noise in shadows and lost details but has proved with examples that this is all myth when shooting RAW as the sensor still captures the details etc.

Thoughts?
 
Be a little difficult photographing moving subjects, I would have thought. So, some of us will need a higher ISO.
 
Last edited:
Fraser Euan White said:
There is an interesting thread on another forum where a contributor is arguing that changing ISO is unnecessary. His argument appears to be very sound in that shooting a picture at the base ISO of the camera usually gives the least noise and the best dynamic range. He then sets his shutter speed and aperture manually to suit the scene he is photographing and doesn't worry what the exposure 'in camera' looks like (most look virtually black and are very under exposed).

The exposure & white balance are then edited in post and look great.

He has faced arguments over noise in shadows and lost details but has proved with examples that this is all myth when shooting RAW as the sensor still captures the details etc.

Thoughts?
Click to expand...

What does he shoot?
 
Totally dependent on the sensor.
It wouldn’t work with any of my cameras.
 
ecoleman said:
What a load of cobblers. Just by pushing exposure in post he would introduce more noise than he would probably get by shooting with a higher ISO.
Click to expand...
He doesn't and has proved this with examples! Original thread is on the nikonites forum.
 
Any other clues to narrow it down ... or a link?
 
Im a member of Nikonite and cant find this thread anywhere...
 
Its in the Post Processing -> Post Your Before and After Pictures thread, but then moves over to another thread later on. It is very interesting !

His username is J-see and tries it on a couple of cameras (D750, a D810 and a lower spec camera)
 
Last edited:
As Phil V says, this subject is dependent on the camera and sensor.

I won't go into detail as it is very easy to google, just type in 'ISO invariance'. Currently AFAIK only a few sensors are ISO invariant and for some of those only in a smallish range.
 
Is it J-See the user
 
His "proof" is not any kind of print I'm guessing. A little picture on a forum is no proof at all.
 
He has plenty of examples of 'fast moving' subjects, including BIF and dogs !
 
ecoleman said:
What a load of cobblers. Just by pushing exposure in post he would introduce more noise than he would probably get by shooting with a higher ISO.
Click to expand...
As others have said, it depends on the sensor...

Consider the two different workflows - the first from say a Canon 5D Mk III, and the second from say a Fuji XT2

1. Photons -> Photosite -> Amplifier -> AD Converter -> Raw file

2. Photons > Phototsite -> AD Converter -> Raw file (with ISO noted in meta data as a multiplication factor)

In the first example the ISO setting causes the analogue signal to be amplified before it is passed to the Analogue to Digital converter and encoded into the raw file as a fixed value. In the second, all the amplification takes place in the raw file my simply multiplying the recorded number by a suitable factor. In the latter case, it makes no difference whether the calculation takes place in camera on in the PC.

Both methods introduce noise and one is not necessarily better than the other, just different.
 
While it is true that ISO does not change how much light the sensor receives/records, there's also no real benefit to his method. It just increases the post work load, and it prevents you from being able to check your settings/images. The best answer is to expose for the highlights you want to retain and go from there.
 
rookies said:
THEN SHARE US THE LINK pls... We cant find his post we all need evidence
Click to expand...

Calm down - I told you where the thread was, just wasn't sure if we were allowed to cross link to another forum. Here is the link to the first page (of 50 !) :-

https://nikonites.com/post-processing/28233-post-your-before-after-pictures.html#axzz58W95fZPU

...and here is a link to a few pages in which shows a virtually black image and then what he gets out of it from post processing the exposure, etc.

https://nikonites.com/post-processing/28233-post-your-before-after-pictures-12.html#axzz58W95fZPU

Read through the thread for plenty of his examples (the one on Page 1 is not the best)
 
Last edited:
sk66 said:
While it is true that ISO does not change how much light the sensor receives/records, there's also no real benefit to his method. It just increases the post work load, and it prevents you from being able to check your settings/images. The best answer is to expose for the highlights you want to retain and go from there.
Click to expand...

His argument is that the digital sensor has it's best noise/DR performance at the base ISO so this is the best to use for performance.

I just tried it out very quickly:

Settings on Camera - ISO 100, 1/1160 sec. at f6.3 gave this result:


_DSC3882 by Fraser Euan White on Talk Photography

Quick edit with NO noise reduction applied, exposure corrected by 31/3 stops:


_DSC3882 edited by Fraser Euan White on Talk Photography

(For some reason the gallery wouldn't let me save the edited image as the same size?)
 
Last edited:
jonbeeza said:
Be a little difficult photographing moving subjects, I would have thought. So, some of us will need a higher ISO.
Click to expand...

This is the biggest advantage! Set your shutter speed/aperture to a value you want, so if photographing a fast moving subject shoot at a high shutter speed, pick your aperture for DoF and leave ISO at 100 - correct the exposure in post :)
 
Fraser Euan White said:
His argument is that the digital sensor has it's best noise/DR performance at the base ISO so this is the best to use for performance.
Click to expand...
The argument is flawed.
The base ISO generates the lowest noise and the greatest DR *ONLY* because/when it is used for "correct exposure" when there is more light (greater density). The relevant factor is the available light (EV) and not the ISO (amplification).

DR is a factor of signal over noise. And the noise is almost entirely due to a lack of light (photon shot noise). So when you apply amplification of the signal (in post or with ISO) you increase the visible noise level which reduces the DR... there is no benefit to either method per-se.
 
Last edited:
sk66 said:
WTH? I just see a bunch of bad images with poor results presented at small sizes...
Click to expand...
Got to totally agree... I thought we were going to see good examples of ISO invariance, instead poor examples of poorly rescued underexposed images.

ISO invariance is real but not in that thread I'm afraid.
 
For a start he clearly doesn't know how to post process very well in the first place. The Raw files often look better than his processed versions when they aren't horribly under-exposed. He is the first person I have ever heard of to do this and I wouldn't be trusting what he is saying or trying to show based on his work.

100% crops are required as a minimum to make a decent comparison, of many different lighting situations.

Quite clearly someone who is more interested in his gear and technicalities than taking good pictures.
 
Fraser Euan White said:
This is the biggest advantage! Set your shutter speed/aperture to a value you want, so if photographing a fast moving subject shoot at a high shutter speed, pick your aperture for DoF and leave ISO at 100 - correct the exposure in post :)
Click to expand...
How do you know if the SS/Ap you selected was optimal for the result you wanted?
 
What is ISO though.......the guy 'on the other forum' seems to insinuate that it is a software type amplification of the initial 'capture', in which case why does it matter if it is done in the camera or in post processing (although I assume, *IF* it is some software, then Nikon/Canon, etc. would have spent many more millions than Adobe/RaWTherapee, etc.) on the algorithm ??
 
Craigus said:
For a start he clearly doesn't know how to post process very well in the first place. The Raw files often look better than his processed versions when they aren't horribly under-exposed. He is the first person I have ever heard of to do this and I wouldn't be trusting what he is saying or trying to show based on his work.

100% crops are required as a minimum to make a decent comparison, of many different lighting situations.

Quite clearly someone who is more interested in his gear and technicalities than taking good pictures.
Click to expand...


It really isn't about taking good pictures though - it is about whether ISO has any relevance when shooting RAW.
 
Richard Breland said:
Errrrr....you still need a good picture at the end of it though :p
Click to expand...

Is it me am I the only one thinking his images are crap they look noisy or not too clean?
 
Richard Breland said:
What is ISO though.......the guy 'on the other forum' seems to insinuate that it is a software type amplification of the initial 'capture', in which case why does it matter if it is done in the camera or in post processing (although I assume, *IF* it is some software, then Nikon/Canon, etc. would have spent many more millions than Adobe/RaWTherapee, etc.) on the algorithm ??
Click to expand...
There is analog gain (amplification) and there is digital gain, both applied prior to writing the raw file. And then there is digital gain applied in post (exposure/DR). The benefit of one method over the other depends on the technology in use... i.e. how much amplification noise is added by the camera, and how much of that is read noise. And yes, if recovering in post then the algorithm used makes a big difference (i.e. Nikon's software does better than ACR for a lot of Nikon files).
 
Richard Breland said:
What is ISO though.......the guy 'on the other forum' seems to insinuate that it is a software type amplification of the initial 'capture', in which case why does it matter if it is done in the camera or in post processing (although I assume, *IF* it is some software, then Nikon/Canon, etc. would have spent many more millions than Adobe/RaWTherapee, etc.) on the algorithm ??
Click to expand...

It would appear that with early digital sensors the ISO was 'burned into' the RAW data and was applied during the Capture which lead to increased noise in shadows when increasing the exposure in post, however some of latest sensors perform all amplification post capture and record the image at the base ISO; like you say Richard - who do you think spends the most money on the post production - Nikon/Canon etc or Adobe etc? (I know who my money is on!)
 
Fraser Euan White said:
It really isn't about taking good pictures though - it is about whether ISO has any relevance when shooting RAW.
Click to expand...
Depends on the camera. If the technology in use does a very good job with very little read/write noise added, then it is said to be ISO invariant... it doesn't have much relevance. However, if the camera has more read/write noise at base ISO the results may benefit by amplification of the received signal above the camera generated noise floor prior to writing the file.
 
Last edited:
sk66 said:
There is analog gain (amplification) and there is digital gain, both applied prior to writing the raw file.
Click to expand...

Apparently not so on the very latest (Sony) sensors; everything is applied post?

Either way I find it quite interesting and think I will try it; unless I experiment I will never know. I started photography with film and ASA mattered then but I am aware that what we are taught because 'it's the way we have always done it' sometimes isn't correct.
 
Last edited:
Just tried it on my D7100 and it works up to a point. Reckon you can get away with 4 or 5 stops of upping the exposure in LR, but any more than that and you get some pretty bad noise/banding, which leads me to believe that this will be present at lower stops as well, but just not as noticeable ? Further investigation required, methinks......
 
Fraser Euan White said:
Apparently not so on the very latest (Sony) sensors; everything is applied post?
Click to expand...
Where does this information come from?
The latest Sony sensors (in certain cameras) have very low camera generated noise, so it takes very little received signal (light) for it to be discernible above the noise floor... i.e. ISO invariant and a large DR. If ISO was only applied in post (as an exif/meta tag) then you could also go the other way... Overexpose an image using ISO to where most everything clips to pure white in the raw file; you should be able to recover the color/detail, but you can't.
 
Last edited:
Richard Breland said:
Just tried it on my D7100 and it works up to a point. Reckon you can get away with 4 or 5 stops of upping the exposure in LR, but any more than that and you get some pretty bad noise/banding, which leads me to believe that this will be present at lower stops as well, but just not as noticeable ? Further investigation required, methinks......
Click to expand...
You will get better results in bright/strong light (higher light density/EV) than you will in low light (low density/high shot noise)... this is the same reason a sensor has it's best performance in bright light situations (where base ISO would typically be used).
 
Last edited:
You must log in or register to reply here.
Back
Top