From DXOMark: More pixels offsets noise3

Started May 30, 2009 | Discussions thread
Steen Bay Veteran Member • Posts: 7,418

John Sheehy wrote:

plevyadophy wrote:

From a practical perspective it struck me as "hocus pocus mumbo
jumbo". I mean really (!), who on earth buys a high res camera so
that they can print at lower resolution/smaller print sizes so as to,
as they call it, "normalize" the noise/data/image. Really?!!!!
[raised eyebrow]

You don't seem to understand what it is to normalize to the image.

Normalizing to the image simply describes what is contained in the
image, in terms of both resolution and noise. When noise is the only
issue of concern, then one may downsample the image and lose
resolution, but this is not necessary.

Normalization is not something that you need to do, physically, to an
image! It is a point of comparison; not a mandatory action to cash
in on a quality!

That may all sound good to those wishing to justify buying a high
pixel cam, but real life often throws these kinda lab tests out of
the window. For example, I have seen comparison shots taken of the
indentical scene with a Sony a900 (24.6Mpix) -v- Nikon D3 (12Mpix);
the Sony had sharper defined fine detail, however in deap shadow
areas the Nikon CLEARLY outresolved the Sony for one simple reason
............... there was no noise obliterating detail. I am sure the
Sony image could be downsampled (and normalised [yawn]) and look much
cleaner, but the detail would be lost forever.

You've got to be kidding! You didn't even normalize, then, so how
can you criticize the normalized comparison! You just looked at it
at 100%. Try this; take a print from a bubblejet with apparent deep
shadow detail. Now, take a picture of a small area with a macro
lens, and zoom in. Can you still see the detail? No, because the
noise is stronger than the detail at the

I have also seen the Canon marketing hype at work i.e. putting
ISO256K on a camera that clearly shouldn't have that high an ISO. The
result? Whilst it performed well in lab tests, when used for what
most people would use high ISO for (namely taking low light shots,
and in the case of the test under incandescant light) it was
outgunned SIGNIFICANTLY by the Nikon camera that is a true ISO256K
camera (and again, whilst downsampling, and all that "normalising"
mumbo jumbo might make the Canon look as clean as the Nikon, the fine
detail has been zapped forever by noise).

Nonsense. Things never work that way in real life; you are just
projecting from a poor model. If the higher-res camera had less
noise after downsizing, it will also have more usable detail, with
higher resultant pixel contrast, unless the low noise was due to
filtering of noise, rather than avoiding it in the first place. Both
upsampled to the same size, the noise will be lower in the same one
in which it was lower with downsampling. Downsampling is NOT
necessary for noise normalization. It is only necessary when a
coarse DISPLAY DEVICE like a monitor forces it.

The terms and language used in these discussions can sometimes be pretty confusing, at least to me. For example I managed for a long time to misunderstand what you and others meant by the term 'image level'. When I was told that at 'image level' the noise was largely the same, regardless of the pixel count (same sensor-size), then I took it to mean that the total noise would be the same, when comparing images at the same output size (not 100% view), like in big prints.

Later Bob (bobn2, who unfortunately has been permanently banned now..) told me that comparing at 'image level' was comparing images at the same level of detail, meaning that the image with more MP's would have to be downsized/blurred (and thereby lose most of its extra detail) for the noise to be the same, again meaning that there wasn't really any 'free lunch', as i thought at first.

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