From: Wolfgang Weisselberg on
Wally <Wally(a)luxx.com> wrote:
> On Thu, 1 Oct 2009 15:09:46 +0200, Wolfgang Weisselberg

>>> What would that show?

>>It would show if the formula was correct. Spot checks say
>>it's off.

> You are not very convincing!

Tell me, what would convince you?

-Wolfgang
From: John Sheehy on
Sorry for the late reply. My ISP discontinued their usenet server
(probably because they don't want to tarnish their image by storing
copyright infringements and under-aged models and such).

Rich <rander3127(a)gmail.com> wrote in
news:00253eed-9843-494a-93a4-f99cb307f96e(a)o36g2000vbl.googlegroups.com:

> Was that posted online somewhere?

Someone linked to it in DPReview, from rapidshare, but I don't think it is
there anymore. There was an 8-column periodic cycle of about 18 14-bit
ADU, IIRC, in the blackframe. The same pattern repeated in other frames,
at other ISOs. It is something that could be fixed if Canon just put a
self-calibration routine in the camera, accessible by the user. Even if it
couldn't be calibrated in the electronics by the user, the RAW data could
easily be re-written to remove the pattern.

I have a 7D pre-ordered from Amazon. I feel like I am at a casino,
gambling on IQ, because of Canon's poor QC. I hate having to erode my
relationship with a vendor to keep returning things to get a good copy.
Canon should bear the brunt of all their sloppiness - then they wouldn't be
sloppy for very long.
From: John Sheehy on
Wolfgang Weisselberg <ozcvgtt02(a)sneakemail.com> wrote in
news:snt7p6-rdf.ln1(a)ID-52418.user.berlin.de:

> It seems a lot of people are, uhm, a bit megapixel crazy.

Other than the increased storage requirements and processing speed, what
is crazy about resolution? You really don't have enough pixels (in terms
of proper sampling) until it takes about 3 pixels exclusive to transition
a sharp black/white subject transient, with your optics. Historical
digital cameras image poorly, and if you print at any good size or crop,
the pixel structure becomes immediately apparent, and all its artifacts.

> A chain being only as strong as the weakest link, I feel that
> resolution is already being limited by most lenses outside
> their best apertures at lower than the present megapixels.
> (I don't think you'll get a noticeably better image from a wide
> open EF 50mm f/1.4 at 18MPix than at 8MPix + sensible upscaling
> (e.g. Lanczos scaling) where needed, for a drastic example.)

Hardly. The noise is finer, and there is less alisasing in the 18MP
version. Low-MP images are fragile things; they only capture some of the
detail at the highest image frequencies they capture, as it depends on
luck of pixel and subject alignment, and then what you do capture at
those highest frequencies are prone to loss with any king of geometric
manipulation, like rotation, perspective correction, CA correction, etc.

>> And that Canon may have concluded that that's where the future is, at
>> least in the near term. Your vigorously defensive reaction syggests
>> to me that you may secretly suspect that I'm right. ;^)
>
> Canon has concluded that they need to sell cameras and to sell
> cameras they need to compete in features and marketing numbers.
> Megapixel record numbers belong there.

Except for some specimen-variable vertical banding patterns at low ISOs
on the 7D (invisible for so-called "proper exposures", for the most
part), the 7D has the best IQ of any APS-C to date, while having the
highest pixel density as well. In fact, I expect the 7D to be more
usable than my 5D2 in low-key images at extremely high ISOs, because the
7D is very lacking in line noise, especially horizontal banding with
nyquist component, which causes color lines (the vertical is lower-
frequency and is more of a luminance noise).

> If suddenly a serious part of the camera buyers(!) would decide
> cameras needed to look like puppys and smell of burned toast ---
> well, Canon would be stupid to not offer them exactly that.
>
> As for *me*, that's not where *my* future lies, yes.

Sometimes the populous is right by accident, which is the case, IMO, with
pixel resolution.

From: John Sheehy on
"DRS" <drs(a)removethis.ihug.com.au> wrote in news:4ac15107$0$12712$5a62ac22
@per-qv1-newsreader-01.iinet.net.au:

> Yes and no. They dropped the megapixel count on the G11.

And they should raise it again, when they figure out how to get the same or
better image-level read noise with the higher densities.

I can tell you this, though. If I'm out on a sunny day shooting scenery,
I'd rather have 14.7MP with a little more noise than 10 MP. I'd only
prefer the G11 if the light got a bit low.

From: Wolfgang Weisselberg on
John Sheehy <JPS(a)no.komm> wrote:
> "DRS" <drs(a)removethis.ihug.com.au> wrote in news:4ac15107$0$12712$5a62ac22

>> Yes and no. They dropped the megapixel count on the G11.

> And they should raise it again, when they figure out how to get the same or
> better image-level read noise with the higher densities.

Yeah, 10.000.000MPix --- the best way to fill memory cards
and hard drives with useless information. And absolutely
needed for showing them on a monitor or printing 4x6. You
need more MPix for 8x12.

> I can tell you this, though. If I'm out on a sunny day shooting scenery,
> I'd rather have 14.7MP with a little more noise than 10 MP. I'd only
> prefer the G11 if the light got a bit low.

So you are often printing posters of house size?

-Wolfgang