Canon Line Offset Noise

Started Oct 20, 2009 | Discussions thread
Mrs Reality Regular Member • Posts: 366
Re: Your big 'wrong'...

Mark H wrote:

John Sheehy wrote:

Mark H wrote:

Your description makes it sound simple - but the underlying 'problem' probably isn't.

I seriously doubt that Canon are simply neglecting to do something that basic and simple.

As I'm sure you know, Canon's sensor technologies are quite sophisticated - supposedly with both 'fixed pattern noise' and 'random noise' reduction mechanisms built in to their CMOS sensor designs, on a per-pixel basis.

http://www.usa.canon.com/uploadedimages/FCK/Image/White%20Papers/Canon_CMOS_WP.pdf

You seem to have a quite a faith in corporate cohesiveness and integrity.

I don't need 'faith' - I've 'been there & done that'...

I've worked 24 years for a 'high technology' company, which does business with many other high tech' companies like, and including, Canon and Nikon, etc, etc. I've worked alongside R&D, design, patents, production, production test, quality assurance, returns and repairs, sales and marketing, customer support - I've seen 'a lot' (both the good and the bad).

... and yes there are, in general, highly commendable levels of 'cohesiveness and integrity', often especially so from our Japanese counterparts - which no doubt contributes greatly to why so much Japanese consumer technology and automobiles get sold to 'the west' and not so much vice versa.

Yeah, Canon's "integrity" really shows in the 1D3 fiasco.

Canon consistently says things which are not correct in their white-papers.

You really are 'full of it'...

For example, they said that the bigger pixels in the 5D and 1D2 make for better high-ISO performance than the 20D pixels, because the former, being bigger. collect more photons. The fact is, all three collect the same number of photons with the same exposure.

...nonsense, your "For example..." is completely wrong.

The bigger pixels do collect more light (photons) - and that is why they do have better high-ISO performance.

For any given exposure value the 'constant' at the sensor is the 'light per unit area' - a larger pixel size (area) therefore equates to more light collected per pixel.

For a given exposure value - the larger the pixel size, the more light collected per pixel - the larger the sensor size, the more light collected by the sensor.

You very recently asserted that you were "very intelligent" - unfortunately you're not doing a very good job of demonstrating it.

Perhaps 'the problem' here is quite simply that there is a fault and/or quality issue.

No kidding. And I solved it.

Wow - I guess you really should be working for Canon then, as you obviously know better than Canon's entire design team does.

You point out yourself that the 'masked' pixels in the top '~32 rows' don't work as an effective 'mask' for subtraction - there probably is an explanation for this, but we will probably never know it.

What difference does it make? 100 rows from the bottom works very well. If Ii were to have these pixels painted black in my camera, I could do this with every image. Why doesn't Canon do this right when anyone else can?

Well maybe you should just go right ahead and start painting your bottom pixels black - (that's sarcasm by the way, so perhaps don't actually).

Canon is either stupid, apathetic, or malicious. There is no noble excuse.

It seems more evident that you are the 'stupid' and 'malicious' one - not Canon.

Developing that idea though (purely hypothetically) - possibly, a subtraction has already been carried out on the main unmasked area, and you might therefore be trying to double subtract - but at this stage I/we are just extrapolating on speculation.

It doesn't matter what has already been done - the fact is, it wasn't done right, if there is still simple pattern left. The only mystery with the top border is how Canon managed to mess it up so badly.

I feel it's a waste of time talking to someone of your disposition and conceit.

You have no idea as to whether or not the 'top border' is being used, or how - so you are not in any position to claim that Canon have 'messed it up'.

There appear to be plenty of Canon DSLR cameras out there that do not need any additional noise subtraction process.

If someone thinks that they have a bad camera that does not perform to an acceptable standard, then they should either get a replacement, or repair, or if they prefer return it and buy a different brand.

The thing is, fewer people would have to return or exchange their cameras or send them in for repair if Canon would either design them better or have better quality control. There should be no such thing as a "bad copy" and no one should have to worry about getting a bad copy, or have to test copies to find a good one. It seems to me that this goes to the point that John and others are trying to make.

I can't fathom the mentality of someone who sees such solid evidence as this, and says there must be some insurmountable problem as an explanation. Get with Reality; Canon doesn't care - and they're not going to care if all their toadies keep defending them.

I don't defend Canon anymore than any balanced and fair minded person would defend any other company.

That's a laugh. A balanced person accepts reality.

For what it's worth, this "Canon toady" (and supposedly paid "Canon shill", as you've accused me of in your previous ranting) currently shoots with a Nikon D90.

You come across as a Canon shill.

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