a pixel size physics question

Started Jun 3, 2009 | Discussions thread
DSPographer
Senior MemberPosts: 2,294
Like?
Re: Scaling of CMOS Sensors
In reply to Bill Janes, Jun 14, 2009

Bill Janes wrote:

DSPographer wrote:

From the dismissiveness of your response, I infer that you did not
even read the papers by Wandell. The paper "How Small Should Pixel
Size Be" was written in 2000. The authors naturally were considering
the technology available at that time, but they did look into the
future by discussing improvements that would result from going from
0.35 u to 0.18 u technology.

These authors are professors at one of our top research universities
in the heart of Silicon Valley, and they are fully aware that
progress can be made with microchips. I don't know who you are, but
it would be interesting to compare your knowledge and experience to
theirs before you begin slinging mud. Some of that mud might come
back and cover your own face.

I did quickly read the first paper and I didn't mean to disparage the authors: hindsight is far easier to get right than the type of predictions they were attempting. But, now that progress has been made we should update our interpretations of these papers based on the latest technology that is now available.

Pixel density needed for what? Capturing detail up to the diffraction
limit of the lens or eliminating aliasing? I was speaking about the
former and mentioned the latter only parenthetically. In your
referenced discussion you talk about the latter. Aliasing wreaks
havoc when you are imaging bar charts, but with real world images
that lack regularly repeating patterns, its effects are often not
that harmful. Most medium format digital cameras do not bother with a
blur filter and give excellent results.

An MTF of 10% may be OK for resolving stars in the night sky, but it
is insufficient for photography of terrestrial objects with much
lower contrast. Can you clarify?

I was referring to capturing detail. When there is no aliasing problem a low MTF can be boosted by the use of sharpening to recover details approaching two pixels per Rayleigh spacing. This sharpening does also boost the noise and since the signal has been reduced by the MTF this sharpening is mainly appropriate for low ISO images with plenty of light that have plenty of SNR.

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