Jump to content

M9 raws vs other FF-sharp but noisy


nugat

Recommended Posts

Advertisement (gone after registration)

CCD-based cameras don't use internal noise-reduction in RAW-files, don't directly compare noise-performance this way (use noise-reduction software for both files and see how well noise & detail hold up in crucial areas - not grey & black cards).

 

I am not sure the noise differences between CCD and CMOS sensors are due to software based reduction in camera. There are electronic controls at the pixel level in CMOS sensors, signal amplification, etc. So, even if your turn off the noise reduction on a CMOS-based camera the noise is much lower at high ISOs than that from CCDs.

Link to post
Share on other sites

Canon cameras do not use internal NR on Raws.

 

At any rate, the result depends a lot on the raw converter. The lastest generation Canon firmware for example produces heavily noise reduced JPEGs which cannot be used for any comparison. ACR on the other hand was known not to be able to handle high ISO Sony A900 files properly.

Link to post
Share on other sites

All CMOS based cameras use black frame subtraction, a form of noise reduction. This is only possible on CMOS because the noise signature is relatively constant, on a CCD it is much more random.

True, but then the kind of fixed-pattern noise typical for CMOS sensors is not an issue with CCDs to begin with. It is caused by non-uniformities in the amplifier circuits for each pixel; in CCDs these circuits simply don’t exist.

Link to post
Share on other sites

Interesting. I personally can't believe how mushy those other, more mature, cameras are. None of them are apparently as high definition as the M9!

 

Since we know this not to be the case (don't we?) there must be a flaw in the methodology.

 

Um, no kidding.

 

This is perhaps the dumbest comparison I've seen yet.

 

Does anyone--and I mean anyone--know what that "milk maid" is supposed to look like? What it's actual printed resolution is?

 

It looks like she has been printed with half-tone reproduction, which makes it utterly useless for noise comparisons. What we are seeing here for noise might also be partly moiré based on the Leica being able to resolve the actual dots in the printing. What a horrible choice of test subject.

 

About all I can take away from this is that the 18MP of the Leica seem to more than match the 21/24MP cameras for resolution, quite astounding, but also expected. The other obvious fact is how horrible the Canon looks, absolutely terrible sharpness from 21MP. I don't know why you don't think this is par for the course, Jamie, it matches my expectations and experience. The 1Ds3 is a 16MP camera, effectively (and the 1Ds2 before it was a 12MP camera). And if the focus of the Canon is off, well, I don't see how that is an excuse for the Canon. Rather, it is an indictment.

 

I know that you can make great shots with a Canon, but my god, those cameras are unattractive to me, ever since I had a 5D. The AA filter and the excessive processing of the images just speaks spades.

 

A far more useful real-world test would be a subject which isn't subject to moiré, and all noise reduction turned off. Then pipe the images through NoiseWare or similar, and see where they stand, sharpness- and noise-wise.

Link to post
Share on other sites

Advertisement (gone after registration)

True, but then the kind of fixed-pattern noise typical for CMOS sensors is not an issue with CCDs to begin with. It is caused by non-uniformities in the amplifier circuits for each pixel; in CCDs these circuits simply don’t exist.

 

CMOS and CCDs have noise with different composition.

 

So, Is it easier to correct the noise on a RAW image if it comes from a CMOS sensor? (I mean the residual noise finally present on the saved RAW file).

Link to post
Share on other sites

CMOS and CCDs have noise with different composition.

 

So, Is it easier to correct the noise on a RAW image if it comes from a CMOS sensor? (I mean the residual noise finally present on the saved RAW file).

No, not really. As you said, noise is somewhat different, but it is the relative importance of different sources of noise that is different, not the quality of the noise.

Link to post
Share on other sites

Valdemar wrote: "I do like that the Leica doesn't have an anti-alias filter in front of the sensor, since they claim it relies on software algorithms to remove Moire, like the 8 year older Kodak 14n. This, to my eye, makes a far crisper and higher resolution image file. I wish the Japanese makers would follow suit.

 

Here's a couple of recent Kodak 14n files shot last week. Albeit lower iso range, but with a similar Kodak sensor to the Leica stills gives anything "modern" a run for the money, including the M9. (This was with a cheap, 35-70mm Nikkor "kit" lens)"

 

These shots look like something I would have gotten with my M8. It's a crying shame Kodak could not get its act together, I still prefer the CCD sensor to the CMOS.

Link to post
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...