High ISO performance - how important is this to you when choosing a camera?

[Mark Haywood]

I've been considering changing manufacturer and have been looking at the Canon 40D, Nikon D90 and Sony A700. I actually bought a Fuji S5 Pro, but think it's a little too much for me. At present I have Four Thirds cameras and wanted better high ISO performance. At first I thought I got this with the Fuji S5, but at a cost in other areas. And when I used Neat Image, I thought my Four Thirds cameras held up remarkably well. I still have a small hankering for one of the cameras mentioned, but with Olympus glass I'm wondering if I would be better served by upgrading to a better Four Thirds body - either the E30 or soon to be released E620.

After scouring the web, and feeling a little tormented by the competition's well lauded superiority in the noise department I wondered how much better would one camera have to be in its high ISO performance for you to decide to choose that camera over another? How do you compare high ISO performance? And just how much better is the competition at say iso1600?

[richard plumb]

on DSLRs its very important to me. You'd be surprised how many shots in the UK need a relatively high ISO to get a decent shutter speed - we're not the sunniest place in the world :P

Add in affordable lenses which aren't the fastest, and you're often up in the 800 area.

On the positive side, I'd say you could use any currently for sale DSLR up to 1600 ISO without a problem unless you're really fussy. I don't have specific experience of the 4/3 system so don't know if thats any better or worse than other DSLRs.

Worth checking out some pictures taken at 1600 ISO to see how you like them - you might be pleasantly surprised.

[denno75uk]

For me, if I were in the market for an upgrade, I think it may be the most important criteria. That's not to say I wouldn't consider the cameras I was considering as a whole, but comparable DSLRs these days are pretty similarly (and competently) specced.
Even with f2.8 lenses, I find myself regularly shooting at 1600 ISO, indoors especially. Being able to utilise a useable high ISO capability just gives a bit more freedom I think. Noise reduction software goes a fair way to cleaning up an image, but sometimes at the expense of fine detail.
That said, the best high ISO performers are also the most expensive (ie full frame), so money is obv also a factor.
As for Olympus' performance Vs the competition, I don't know, but would suggest that the smaller sensor in the 4/3 system would give it a distinct disadvantage.

[Mark Haywood]

Originally Posted by richard plumb

Worth checking out some pictures taken at 1600 ISO to see how you like them - you might be pleasantly surprised.

Having read most of the well known reviews, I've come to realise that iso1600 may not actually be iso1600. So if one manufacturer's iso1600 setting is reasonably close to its stated setting and another's turns out to be actually iso1125 how does one accurately compare?

It seems that whilst Olympus and Panasonic seem to be fairly accurate, Canon, Nikon, Sony, Pentax and Samsung seem to be quite a ways off in their stated iso settings, at least if DXO's testing suggests what I think it does!

[andy1249]

Having read most of the well known reviews, I've come to realise that iso1600 may not actually be iso1600. So if one manufacturer's iso1600 setting is reasonably close to its stated setting and another's turns out to be actually iso1125 how does one accurately compare?

This is true.

The way it usually works for any particular digital sensor is the point on the gain curve that equates to ISO 100 is arbitrarily chosen, meaning the manufacturer picks the point that it feels is closest to ISO 100.

Then the rest of the points are calculated from there , these points are supposed to match EV values across the entire gain curve , so ideally any particular EV value chosen on any camera should give the same results , however differences in gain curves from different sensors are unavoidable so exact matches between cameras are not possible in some cases.

They should however be very close , and a difference of ISO 575 between cameras set to a value of ISO 1600 as listed above would be very poor indeed.

[Yandros]

If high ISO performance is important to you I'd cut your losses and ditch Olympus.

Canon and Nikon are currently fighting over the high ISO crown. The Nikon D90, D300, Canon 50D and probably 500D are so close it's barey worth talking about, and the 450D not far behind. The Sony sensor is a bit noisier at high ISO, and the Pentax and Olympus ones a LOT worse.

[senu]

In a word, Yes

Where you cannot use a Tripod, Flash or dont have or cant use a fast lens, usable high ISO makes the difference between a shot and a non shot

However even within the same brand I think ISO implemetation are not identical let alone comparing different brands
As an example
Pt the same lens on say a Canon 400D and 40D and take the same shot in challenging light conditions using IS0 1600: the results are nonidentical despite somewhat similar paper specs on the cameras

[Mark Haywood]

Originally Posted by andy1249

This is true.

The way it usually works for any particular digital sensor is the point on the gain curve that equates to ISO 100 is arbitrarily chosen, meaning the manufacturer picks the point that it feels is closest to ISO 100.

Then the rest of the points are calculated from there , these points are supposed to match EV values across the entire gain curve , so ideally any particular EV value chosen on any camera should give the same results , however differences in gain curves from different sensors are unavoidable so exact matches between cameras are not possible in some cases.

They should however be very close , and a difference of ISO 575 between cameras set to a value of ISO 1600 as listed above would be very poor indeed.

Thanks, I didn't know that.

Perhaps I'm simply mis-reading the DXO "Compare Cameras" graph? Wouldn't be the first time I'm way off base!

Originally Posted by Yandros

If high ISO performance is important to you I'd cut your losses and ditch Olympus.

Canon and Nikon are currently fighting over the high ISO crown. The Nikon D90, D300, Canon 50D and probably 500D are so close it's barey worth talking about, and the 450D not far behind. The Sony sensor is a bit noisier at high ISO, and the Pentax and Olympus ones a LOT worse.

I've read all of the major reviews for those cameras. What do you base your opinion on, other than those reviews, though?

Originally Posted by senu

In a word, Yes

Where you cannot use a Tripod, Flash or dont have or cant use a fast lens, usable high ISO makes the difference between a shot and a non shot

However even within the same brand I think ISO implemetation are not identical let alone comparing different brands
As an example
Pt the same lens on say a Canon 400D and 40D and take the same shot in challenging light conditions using IS0 1600: the results are nonidentical despite somewhat similar paper specs on the cameras

If that's the case, then how do we compare like for like iso settings?

I've been looking here and at the SNR 18% graph... Am I just missing something?

[senu]

Originally Posted by Mark Haywood

If that's the case, then how do we compare like for like iso settings?
I've been looking here and at the SNR 18% graph... Am I just missing something?

In a sense , It would seem that objectively the differences are not chalk and cheese
What seems to make a difference (in my limited experience) is what you RAW process with for NR and how much quality is left from the Post NR RAW output
DXO results dont go that far as there are too many variables
So. paper specs aside, in use some cameras are that much " worse" in higher ISO Noise output than others
Im sure someone wih a lot more knowledge ( than I ) in these matters will shed furter light on this

[Mike.P®]

Since moving back to the UK from Spain last Sunday it's a lot more important than it used to be.
Talk about crap weather

[Mark Haywood]

Originally Posted by senu

In a sense , It would seem that objectively the differences are not chalk and cheese
What seems to make a difference (in my limited experience) is what you RAW process with for NR and how much quality is left from the Post NR RAW output
DXO results dont go that far as there are too many variables
So. paper specs aside, in use some cameras are that much " worse" in higher ISO Noise output than others
Im sure someone wih a lot more knowledge ( than I ) in these matters will shed furter light on this

But, presumably, the lower the iso setting the greater the dynamic range, which should lead to a little more leeway as far as raw processing goes. So if the iso setting isn't accurate - ie. it is actually less than stated - there would be a little more adjustability in processing the raw image.

All this is moot of course if when I look at an iso1600 shot from a Nikon D90 and compare it to an iso1600 shot from the Olympus E30 I am actually comparing similar isos! It just looks from the DXO site that this may not be the case, but I'm probably missing something obvious! Interestingly though, DPReview have often mentioned that Olympus cameras have less raw headroom adjustment than other cameras and I wondered if this was because Olympus' iso settings were more accurate?

[imcarm]

High ISO performance is a must for me. Take for example your out shooting an action shot - perhaps someone mountain biking in the woods for example, it's an overcast day and the light is really poor due to the above tree cover, even with a fast F2.8 telephoto your going to struggle to stop the action - Using flash is one way to overcome this (not always an option though) or pushing up the ISO and risking some noise is the other option, ISO noise can actually add to the shot sometimes

[Yandros]

Mark, mostly from lots of reviews and test shots in various forums. Senu is probably the best person to compare Canon/Nikon like for like as he uses both on a day to day basis.

The main thing I'd say is that you do have to take test shots in reviews with a pinch of salt. High end Nikons always undersharpen by default vs Canon, so will always look soft. Canon applies a heavy handed general noise reduction, which has traditionally been labelled 'plasticy' by Nikon shooters. Nikon on the other hand kills chroma noise in preference, leaving a more film like grain - more detail, but more obvious 'noise'. Sony use pretty much the same sensor as Nikon, but don't take out the chroma noise quite as well. Having said all this, I'm sure if you took a RAW file from a Canon, Sony or Nikon 12MP+ camera and processed it through 3rd party NR software, you'd end up with very similar results. With the 50D vs D300, you also have a resolution difference, so although the D300 has the edge at high ISO, by the time to factor in the 3MP difference, the advantage disappears.

The high ISO test shots from the K20 and the E30 at 1600 ISO just look bad I'm afraid - not just slightly worse, that could be compensated for in post - a LOT worse. I've seen some amazing shots at A3 size from a friends K10 though, but I don't think he's pushing the ISO too hard.

[tracymae]

It wasn't something I really considered...until I took some pictures at my daughter's christmas party, 90% unusuable from the panasonic bridge camera I think it has to be considered unless you plan on using only outdoor and in good light.
I havent noticed a problem with noise at all with my Olympus but maybe I have just been lucky.
Is there any chance we could do a wee comparison test of our own? If so how to go about it?

[Mike.P®]

Originally Posted by Yandros

The high ISO test shots from the K20 and the E30 at 1600 ISO just look bad I'm afraid - not just slightly worse, that could be compensated for in post - a LOT worse. I've seen some amazing shots at A3 size from a friends K10 though, but I don't think he's pushing the ISO too hard.

Well I dont know where you have been looking but do a search on Flickr for K20D and ISO 1600 as I just did and I can see some excellent shots.
There's even a couple for the E30.