Canon 20D, final verdict?

In message . com,
wrote:

I have not had a single problem with my 20D and I have had it in use
since Spetember. Some people have had various problems, but the problem
traffic at www.dpreview.com has died down to an insignificant level. It
is my guess that 98% of everybody who bought this camera has had
absolutely no problems with it. I find my 20D an excellent dSLR.

It depends on your threshold for complaint. The banding issue is
nothing short of pathetic, IMO. There is no excuse for Canon letting
the camera out of the gates without at least addressing blackpoint
line-by-line in its own software. It gives me the feeling that the
company is a bunch of disconnected departments, and the the part that
makes the excellent sensor has never explained to the part that makes
the readout circuitry or the JPEG-writing firmware that one single
blackpoint value does not work well for the entire image.
--


John P Sheehy

wrote:
In message . com,
wrote:


I have not had a single problem with my 20D and I have had it in use
since Spetember. Some people have had various problems, but the problem
traffic at www.dpreview.com has died down to an insignificant level. It
is my guess that 98% of everybody who bought this camera has had
absolutely no problems with it. I find my 20D an excellent dSLR.


It depends on your threshold for complaint. The banding issue is
nothing short of pathetic, IMO. There is no excuse for Canon letting
the camera out of the gates without at least addressing blackpoint
line-by-line in its own software. It gives me the feeling that the
company is a bunch of disconnected departments, and the the part that
makes the excellent sensor has never explained to the part that makes
the readout circuitry or the JPEG-writing firmware that one single
blackpoint value does not work well for the entire image.


The banding occurs in 1DMKII and 1DsMKII as well. I think this is
pushing the limit of the CMOS and we're seeing noise, rather than
signal. It should happen only when pushing very high ISO though.

In message . net,
leo wrote:

The banding occurs in 1DMKII and 1DsMKII as well. I think this is
pushing the limit of the CMOS and we're seeing noise, rather than
signal. It should happen only when pushing very high ISO though.

Not at all. This is not noise; it is *modulation* of the noise, caused
by the lack of an insanely simple arithmetic adjustment that needs to be
applied on a line-by-line basis. I've seen the solution in action,
already, in a doctored version of DCRAW. It is not a sensor problem,
per se. It is a hardware problem reading the RAW data off the sensor,
with a workable software solution. The solution is as simple as adding
1 to all the RAW values on line 3, subtracting 3 from all RAW values on
line 7, etc. It's really that simple, but it got by Canon completely.

The sensors in these cameras are much better than what the camera's
JPEGs and external RAW conversions are currently churning out. Nothing
draws one's attention more to noise than simple geometric modulations of
it. The random noise of the sensor, when properly read, is only strong
when the image is viewed at magnification. The banding patterns do
*not* disappear with minor downsampling.
--


John P Sheehy

In message . net,
leo wrote:

The banding occurs in 1DMKII and 1DsMKII as well. I think this is
pushing the limit of the CMOS and we're seeing noise, rather than
signal. It should happen only when pushing very high ISO though.

Not at all. This is not noise; it is *modulation* of the noise, caused
by the lack of an insanely simple arithmetic adjustment that needs to be
applied on a line-by-line basis. I've seen the solution in action,
already, in a doctored version of DCRAW. It is not a sensor problem,
per se. It is a hardware problem reading the RAW data off the sensor,
with a workable software solution. The solution is as simple as adding
1 to all the RAW values on line 3, subtracting 3 from all RAW values on
line 7, etc. It's really that simple, but it got by Canon completely.

The sensors in these cameras are much better than what the camera's
JPEGs and external RAW conversions are currently churning out. Nothing
draws one's attention more to noise than simple geometric modulations of
it. The random noise of the sensor, when properly read, is only strong
when the image is viewed at magnification. The banding patterns do
*not* disappear with minor downsampling.
--


John P Sheehy

In message . net,
leo wrote:

The banding occurs in 1DMKII and 1DsMKII as well. I think this is
pushing the limit of the CMOS and we're seeing noise, rather than
signal. It should happen only when pushing very high ISO though.

Not at all. This is not noise; it is *modulation* of the noise, caused
by the lack of an insanely simple arithmetic adjustment that needs to be
applied on a line-by-line basis. I've seen the solution in action,
already, in a doctored version of DCRAW. It is not a sensor problem,
per se. It is a hardware problem reading the RAW data off the sensor,
with a workable software solution. The solution is as simple as adding
1 to all the RAW values on line 3, subtracting 3 from all RAW values on
line 7, etc. It's really that simple, but it got by Canon completely.

The sensors in these cameras are much better than what the camera's
JPEGs and external RAW conversions are currently churning out. Nothing
draws one's attention more to noise than simple geometric modulations of
it. The random noise of the sensor, when properly read, is only strong
when the image is viewed at magnification. The banding patterns do
*not* disappear with minor downsampling.
--


John P Sheehy

wrote:

In message . net,
leo wrote:


The banding occurs in 1DMKII and 1DsMKII as well. I think this is
pushing the limit of the CMOS and we're seeing noise, rather than
signal. It should happen only when pushing very high ISO though.


Not at all. This is not noise; it is *modulation* of the noise, caused
by the lack of an insanely simple arithmetic adjustment that needs to be
applied on a line-by-line basis. I've seen the solution in action,
already, in a doctored version of DCRAW.

Any chance of a link to this?


Rob.
--
Images http://www.pbase.com/mapleglen/

wrote:

In message . net,
leo wrote:


The banding occurs in 1DMKII and 1DsMKII as well. I think this is
pushing the limit of the CMOS and we're seeing noise, rather than
signal. It should happen only when pushing very high ISO though.


Not at all. This is not noise; it is *modulation* of the noise, caused
by the lack of an insanely simple arithmetic adjustment that needs to be
applied on a line-by-line basis. I've seen the solution in action,
already, in a doctored version of DCRAW.

Any chance of a link to this?


Rob.
--
Images http://www.pbase.com/mapleglen/

"Confused" wrote in message
news
In message
"Si" wrote:

...

So, rather than spend a grand plus on a 20D and grip,
I'm going to be rash and get a 1D Mk II

If it's that easy dollar wise, consider a 1Ds Mark II. When you want
to simulate a 1D Mark II sports machine reduce the captured file to
JPEG quality 8 at 3600x2400. More frames before the buffer fills (ie
faster operation) similar to the 1D Mark II. (It might even out
perform it but no one has compared the two that I know of.)

Extra benefits include a 100% sensor AND 100% viewfinder, 17.2
megapixels, and probably other stuff too. :^)

Jeff

I'd love a 1Ds Mark II, but UK price for the 1D Mark II is around £2700,
whereas it's another 3 grand for the 's'...

That's some chunk of cash.

Si.

"Confused" wrote in message
news
In message
"Si" wrote:

...

So, rather than spend a grand plus on a 20D and grip,
I'm going to be rash and get a 1D Mk II

If it's that easy dollar wise, consider a 1Ds Mark II. When you want
to simulate a 1D Mark II sports machine reduce the captured file to
JPEG quality 8 at 3600x2400. More frames before the buffer fills (ie
faster operation) similar to the 1D Mark II. (It might even out
perform it but no one has compared the two that I know of.)

Extra benefits include a 100% sensor AND 100% viewfinder, 17.2
megapixels, and probably other stuff too. :^)

Jeff

I'd love a 1Ds Mark II, but UK price for the 1D Mark II is around £2700,
whereas it's another 3 grand for the 's'...

That's some chunk of cash.

Si.

In message ,
Rob Davison wrote:

Any chance of a link to this?

http://forums.dpreview.com/forums/re...ssage=12040120
--


John P Sheehy