Olympus OM-D E-M5

Started Apr 13, 2012 | Discussions thread
CriticalI
Senior MemberPosts: 1,777
Like?
Re: M43 v APSC Dynamic Range
In reply to hellocrowley, Apr 17, 2012

The DxO results are given both for native sensor and downsized images. Downsizing reduces the effective noise level by averaging out the randomness across neighbouring pixels hence a very high MP sensor can apparently achieve the same DR as a much lower MP one with much greater well depth and similar read noise.

But measured natively the real DR is a factor of the effective well capacity and the noise floor, and small pixels always have lower capacity hence lower S/N all else being equal.

The other random factor is QE. With high QE you need less amplification for a given exposure (ISO value) hence lower noise. However most sensors are reaching very similar levels of QE at around 70% so there is not much variation. There is of course also the noise added in signal processing (and removed) which is not insignificant.

So at the end of the day I would expect the Oly to have (at best) 1/2 stop worse noise performance than the Fuji assuming both have good read noise and QE characteristics which they probably have. Looking at high ISO samples from both this seems about right, but without RAWs its very hard to tell.

The Fuji has another advantage which is the higher apparent resolution due to the lack of AA filter which means less aggressive sharpening is required to counteract the NR, producing cleaner files.

I think the Oly is a great little camera and a much better "all rounder" in all probability, but I would be very surprised if it competes on high ISO noise with the Fuji. Most people reckon a stop better than the EP3. Damned good for a smaller sensor and quite close to the D7000/NEX5 in RAW (dpreview) which is a great achievement. More important it holds onto colour information pretty well.

But the high ISO results I have downloaded for the Fuji easily compare with my D700, and the Oly ones most certainly do not, even resized.

hellocrowley wrote:

I'm not convinced that the people who write their own raw converter would be off by 4EV in their test, and that you know better than they do. Your lack of proof doesn't help.

Going by your definition of DR (which is probably correct) low light sensitivity is just 1 part of the equation, the ability to capture high intensity is equally important.

echelon2004 wrote:

hellocrowley wrote:

How do you explain the Nex7's 13.4EV dynamic range and lousy high ISO performance?

Easy. It's nowhere close to that in real dynamic range. I don't know enough about the dxo software to understand why it's do much off. I'd say 9,5-10 ev real world dynamic range. Which is still very good for Sony

But dr is the distance between noise and a clipped signal.

echelon2004 wrote:

The real question is how there can be so many people that things that there's a choice between good dynamic range and good noise performance when they are simply two ways to describe the same qualities.

-- hide signature --

Anders

'It is nice to be important but it is more important to be nice'

-- hide signature --

Regards,
Steve

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
LOLNew
LOLNew
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow