Re: Mega Pixel Myth

On Mon, 28 May 2007 13:48:01 GMT, John Sheehy wrote:

Have you seen Simon Joinson's article "Compact Camera High ISO
modes: Separating the facts from the hype" posted several days ago?
I suspect that you haven't.

I suspect that I have, and it isn't anything I haven't seen before, and
don't see any evidence of major RAW differences. It is all explained by
binning (binning does *not* increase the detail to noise ratio unless there
is not detail near the nyquist to begin with) and NR. The Fujis make some
bold assumptions about what is noise, and create cartoon-like effects
(points and smal clusters removed) with sharpened remaining edges, edges
that are in the wrong place, like aliased imaging, but with a more random
pattern of displacement.

If you really did see Joinson's article come out and say so
instead of couching it to make it appear that you did when your
following statements make it appear that you obviously haven't. The
images that were binned were identified as such, and they *all*
looked terrible. Are you claiming that the Fuji images were binned?
I identified the ones that were, and the Fuji images were not among
them. Despite talk of "bold assumptions" and "cartoon-like effects"
there's no getting around the fact that where the high ISO images
were concerned, the Fuji images were clearly superior to all of the
others in detail and lack of noise except for the ones taken by the
30D, and the F30 produced results almost as good as the 30D, at
least for the cropped images displayed. If the complete image is
downloaded, the superiority of the 30D's image is easier to see, but
by the same token, it's much easier to see that the F30's images are
*much* better than those taken by any of the other P&S cameras. Try
downloading the ISO 800 images taken by the D30, F30 and Panasonic's
LX2. The 8mp LX2 has detail comparable to the 6mp F30, but if you
look closely at the little robot, the only noticeable fault in the
F30's image is the slight stairstepping of the words "TIN LIGHT" on
the robots chest. But in all other respects the F30's image is far
superior, lacking the gross noise that obliterates much of the
robot's detail, as well as the text on the bottom of the Bailey's
Irish Cream label. Looking at the 30D's image you can make out the
word "DUBLIN" quite easily. It's discernable in the F30's image,
although not very distinct. In the LX2's image, it's a pure mess of
noise, completely unreadable, and not at all obvious that there's
any text in that part of the label.

The reasons for the F30's reputation aren't due to the assumptions
you've made, but to the fact that the Fuji F30 sensor is larger and
has fewer pixels, making them unusually large and inherently better
for reducing noise at high ISOs. If you bother to reply and want to
be taken seriously, please explicitly state whether you've actually
examined Joinson's images. In the real, as opposed to the
theoretical world, the F10/F30 has been particularly popular with
DSLR owners. Some have noted that they aren't the best performers
in bright daylight, but I have yet to hear any of them complain that
the F10/F30's high ISO performance in low light levels wasn't up to
the claims made for them.

.

Relevant Pages

  • Re: Nikon D80/200 - Canon 30d
    ... than an el cheapo Point&Shoot camera with a tiny teeny-weeny CCD ... from a manufaturer that its CMOS sensor has less noise than a CCD ... just look at the images (not the ...
    (rec.photo.digital.slr-systems)
  • Re: Digital Rebel XT vs. E-300 - need opinions, sooner better than later :>
    ... I test a lot of cameras each year and I have to say that I find very few ... the quality of images produced that most people will notice or be bothered ... >> There is too much obsession about sensor noise. ... > the tendency to render blues with purple tints, ...
    (rec.photo.digital)
  • Re: Last Question...Please
    ... but the noise at ISO 400 is absolutely horrible. ... Take a look at these two images of the ... The Canon's are popular in astrophotography for their low noise ... performance even at high ISO setting, ...
    (sci.astro.amateur)
  • background filament networks (Murray mesh) in deep sky photos-- noise artifacts or early cosmic stru
    ... very reasonable interpretation that the faint background filaments noticable ... shifts are possibly just noise. ... as in the case of the recent Hubble images of Abell 1689. ... if an image has 8 bits of depth per pixel ...
    (sci.physics)
  • Re: Nikon D80/200 - Canon 30d
    ... unequivocally that CMOS sensors generate *more* noise than CCDs, ... physics and statistics --- you don't even try to understand ... that they are not measurements, but claim that nobody does have ... Basically, they take two images shot one right after the other, then ...
    (rec.photo.digital.slr-systems)
Loading