Does high iso capability really matter in RAW

[unclened]

I was thinking this through and wandered whether high ISO really mattered from a pure IQ perspective when shooting in RAW? To the best of my knowledge, ISO is an digital exposure boost applied by the camera using a proprietary algorithm. Whats to say 1600 on a 5DMKII pushed two stops in LR isn't any worse than 5DMKIII at 6400?

 

[woof woof]

If you have a camera and can shoot RAW why not try it and see?

 

[KIPAX]

there is more noise in RAW file than a JPG ....well OK same amount BUT... a JPG is compressed.. therefore it shows less noise

 

[arad85]

Yes it matters. Higer ISO reduces dynamic range and decreases signal to noise ratio among other things. Look at it this way, if you capture a number of photons and then amplify the signal, you are also amplifying the noise.

The old adage always works. Crud in = crud out.

 

[Johnd2000]

there is more noise in RAW file than a JPG ....well OK same amount BUT... a JPG is compressed.. therefore it shows less noise

Primarily, the jpg will be less noisy because the camera will apply NR. Unless the user switches in-camera NR off, of course.

To the OP, I'm not sure if you meant to compare the 5D2 and 5D3, but generally I don't think you gain anything by underexposing and pushing in post. I've done tests on a couple of my cameras and the amount and appearance of the noise generated by pushing is generally the same as you'd get shooting at the higher ISO originally. NR software improves over time, but the "amplification" used to boost exposure adds noise in a very linear/predictable way.

 

[Cagey75]

A Jpeg is a file processed by the camera. This often includes noise reduction/smoothening. Fine for snap shots of yer mates in the pub, but for your serious images, RAW will always be much better. Do your own controlled noise reduction and it's always going to be better, better control over all - shadows, highs, mids ...

I would rather expose the image right at the time than rely on pushing exposure levels later in post. Your not going to get the same detail in shadows shooting under exposed for a start. You can't bring something back when it's not there to begin with.

 

[ernesto]

Not sure where the talk of JPEGs has come from? I read the question as is there a difference in choosing ISO 3200 in camera or shooting at ISO 800 and using LR to get to 3200 equivalent.
Does the camera produce more noise physically giving ISO 3200 than software does to get to same ISO or vice versa?

 

[KIPAX]

Even without noise reduction.. the simple fact that a jpg file is compressed would mean it shows less noise then a RAW file.. nothing to do with NR ta

 

[maninsuitcase]

Yes it matters. Higer ISO reduces dynamic range and decreases signal to noise ratio among other things. Look at it this way, if you capture a number of photons and then amplify the signal, you are also amplifying the noise.

The old adage always works. Crud in = crud out.

But, you take 2 exposures both the same shutter speed and aperture one at 1600 one at 6400. The same amount of light was collected. Correct. Both raw.

Now with the 6400 it was the camera that boosted it, but with the 1600 you do it +2ev in lightroom.

With out trying it (as I am at work) what's the difference, both had the same signal (based on assumptions on how I understand ISO to work in digital), the difference is what did the boosting.

Essentially the equivalent of a push process if my terminology is correct.

 

[unclened]

Even without noise reduction.. the simple fact that a jpg file is compressed would mean it shows less noise then a RAW file.. nothing to do with NR ta

what about RAW to JPG in post? same thing but better control?

 

[unclened]

Primarily, the jpg will be less noisy because the camera will apply NR. Unless the user switches in-camera NR off, of course.

To the OP, I'm not sure if you meant to compare the 5D2 and 5D3, but generally I don't think you gain anything by underexposing and pushing in post. I've done tests on a couple of my cameras and the amount and appearance of the noise generated by pushing is generally the same as you'd get shooting at the higher ISO originally. NR software improves over time, but the "amplification" used to boost exposure adds noise in a very linear/predictable way.

yeah the comparison is a bit of a misnomer. A like for like comparison sounds more reasonable!

 

[ausemmao]

I was thinking this through and wandered whether high ISO really mattered from a pure IQ perspective when shooting in RAW? To the best of my knowledge, ISO is an digital exposure boost applied by the camera using a proprietary algorithm. Whats to say 1600 on a 5DMKII pushed two stops in LR isn't any worse than 5DMKIII at 6400?

On the vast majority of cameras it will be worse.
On some it will be essentially identical (D7000, D800, D600, A99).
It'll never be better.

 

[unclened]

On the vast majority of cameras it will be worse.
On some it will be essentially identical (D7000, D800, D600, A99).
It'll never be better.

why worse on some and identical on others?

 

[arad85]

With out trying it (as I am at work) what's the difference, both had the same signal (based on assumptions on how I understand ISO to work in digital), the difference is what did the boosting

It's to do with how the signal is quantised. If everything is held in the shadows, you only have a very limited dynamic range when you push it.

Perhaps I misunderstood the OPs question though.

 

[maninsuitcase]

It's to do with how the signal is quantised. If everything is held in the shadows, you only have a very limited dynamic range when you push it.

Perhaps I misunderstood the OPs question though.

That's true I would assume (perhaps wrongly) that gain is applied before the ADC but certainly before the save to the 12 or 14 bit raw. So the gain could well be working on a higher internal bit rate before saving to a lower bit raw/processed jpg file. Either way you would loose data in the file you have access to that the camera can use to do the push.

 

[HoppyUK]

That's true I would assume (perhaps wrongly) that gain is applied before the ADC but certainly before the save to the 12 or 14 bit raw. So the gain could well be working on a higher internal bit rate before saving to a lower bit raw/processed jpg file. Either way you would loose data in the file you have access to that the camera can use to do the push.

Perhaps rightly

 

[ausemmao]

why worse on some and identical on others?

There's a component of noise that's called read noise. If a sensor has very low read noise that doesn't change much with ISO, then there's little or no benefit. If a sensor has high read noise, then there is a benefit. If it has read noise that falls with increasing ISO, then there is an even bigger benefit (as read noise becomes dominant over shot noise). The reason for this is why the D800/D600/A99 etc have better DR at low ISO but worse at high compared to the D4/1Dx/5D3.

The reasoning behind it isn't particularly interesting unless you enjoy maths.

 

[Phil Young]

I was thinking this through and wandered whether high ISO really mattered from a pure IQ perspective when shooting in RAW? To the best of my knowledge, ISO is an digital exposure boost applied by the camera using a proprietary algorithm. Whats to say 1600 on a 5DMKII pushed two stops in LR isn't any worse than 5DMKIII at 6400?

I started a thread about this a while back and gave examples...the short answer is no...but...

Apparently there is a detail threshold and staying within this is the key.

I shot many shots and a 5 stop pullback from an ISO200 shot looked pretty much identical to the same scene shot at 6400.

It was a very interesting test but the conclusion was: if you shot in manual within your cameras threshold for detail, you can pull back in PP. The only thing is, obviously it won't be spot on in camera.

EDIT: sorry, just read this back and it didn't make a lot of sense. I'll try again...

A few months back we were discussing ISO in another thread and somebody posted a link to a camera rest on a D7000 which showed something ABOUT DSLRs that if you shoot within the camera's detail threshold (can't remember the exact terminology) that the body simply applies, like you said, a digital boost to the signal. This was proved that shooting below a certain level (for the D7000 it's ISO800) will give you the same results as shooting at the correct ISO to begin with.

I was intrigued at this new knowledge to me and took a few shots indoors at ISO's below what they should have been and also another shot at the correct ISO.

The results were bar a little contrast reduction in the pulled back exposure, they looked the same.

There shots went all the way up to 25600.

Conclusion for me was:

In theory, you could do a wedding by sticking your camera at ISO400 and just leave it there in manual all day. The results when pulled back in PP will be the same.

Obviously, that's not an efficient way of doing it but what it does prove is not to be afraid of pulling back and IQ loss when you didn't nail the ev. I tried shooting and ISO3200 to start with and then boosting and the results were considerably worse.

So my simple answer is no it doesn't matter (unless your ISO is higher to begin with) but for obvious reasons, it's best to up the ISO rather than shoot low ISO and boost in PP all the time.