Got 140 bit DR Image?

BradGuth wrote:
On Jan 29, 12:45*am, (Floyd L. Davidson) wrote:
"Mr.T" MrT@home wrote:
"J. Clarke" wrote in message
...
Then there's the "". *My old Coolpix 990 has considerably 140 dB
and why one would want a camera with 140 bits is a mystery to me.

Gee even the best SLR's only do ~14 bits per channel, or 42 bits per pixel
atm.

That is *not* 14 bits per "channel", as there are no
channels. *It's simply 14 bits per "pixel". *(It isn't
really a pixel either, it's 14 bits per sensor
location.)

To actually capture a 140 dB dynamic range requires a
minimum of 23 bits per sensor data sample.

But there still seem to be a lot of people who do want (or will happily
accept) "140 bits DR".

*:-)

We don't want to hold our breath waiting for a 24 bit ADC?

--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska) * * * * * *

Perhaps no breath holding is necessary.

Using 2 or 4 quick scans per image could make the 16 bit ADC more
viable for the 140 db cmos imager. There's at least one commercial/
science camera that can selectively average up to 256 scans per output
image.

~ BG

People are commonly using HDL with modern DSLR's to
achieve greater dynamic range. In theory a Nikon D3 can
get about 12 fstops, so two images would suffice. In
practice of course it takes 3.

--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska)

On Thu, 29 Jan 2009 04:30:04 -0800 (PST), BradGuth
wrote:

On Jan 29, 12:39*am, "Mr.T" MrT@home wrote:
"BradGuth" wrote in message

...

BTW the incorrect/ad-hoc transposition of "bits" and "DB" makes the OP
both
wrong and unreadable in any case!
My dyslexic mind isn't what it used to be, so sue me. *In spite of
myself, you know exactly what I'd intended to convey.

Not so, or we wouldn't be arguing where you are technically wrong and/or
technically obtuse.

The A/D bits of
digital imaging was similar enough to the range of DR in db.

Not as an equivalent figure they are not, 1 bit =~6dB and therefore dB is
not the same as the number of stops either.

Most
scientific cameras are A/D limited as to creating 16 bit image data,
although the 140 db worth of CMOS imager DR itself is simply pretty
hard to ignore.

There are sample shifting techniques that can use 16 bit A/D's to provide 32
bit data, more than enough for 140dB sensors.

MrT.

Thanks for that informative feedback, as I certainly didn't understand
what it would take in order to fully realize the maximum 140 db
imaging capability. This should mean our consumer video products
should soon incorporate at least this much or better DR, and thus
images of Sirius shouldn't be over saturated to the point of losing a
good look-see at Sirius C, especially if certain optical bandpass
elements are utilized.

The link you provided gives several hints about how they achieve wide
dynamic range-
Page 2: "...the CMOS imaging approach also provides the choice between
linear and logarithmic pixel readout characteristics, random pixel
access..."
Page 1, Fig 2, (5) "Locally adaptive - 128 * 128 pixels, logarithmic
readout, on-chip local brightness adaptation""
Page 2, Fig 3: "...CMOS camera with individual exposure time
selection"

If they can control both exposure time and sensitivity (ISO)
individually for small blocks of pixels, and read them
logarithmically, the wide irradiance range is less astonishing.
--
John

"BradGuth" wrote in message
...
Thanks for that informative feedback, as I certainly didn't understand
what it would take in order to fully realize the maximum 140 db
imaging capability. This should mean our consumer video products
should soon incorporate at least this much or better DR

I wouldn't want to hold my breathe if I were you :-)
But I do expect we will get to true 16 bit capture before too long, at which
point most people will be more than happy.

MrT.

"BradGuth" wrote in message
...
Your 6 bits per cmos imager DB even sounds terrific.

Thats 6 dB per bit which is simple mathematics, nothing more.

I suppose what's
next will be the 256 db imager and 21 bit A/D processor,

Would need at least 2 in that case.

along with a
terabyte in media storage capacity, plus 100:1 zoom optics including
many nifty bandpass filter options, all for under a few grand.

You are aware of the concept of market demand?
*IF* you think there is sufficient demand for the above, it may even happen.
But maybe not in our lifetime :-)

MrT.

"BradGuth" wrote in message
...
Perhaps no breath holding is necessary.

Using 2 or 4 quick scans per image could make the 16 bit ADC more
viable for the 140 db cmos imager. There's at least one commercial/
science camera that can selectively average up to 256 scans per output
image.

Multiple scans are only necessary for sensor limitations, not A/D
limitations. If the imager *can* provide 140dB DNR in one pass, then no
extra scans are necessary to capture more than 16bits of data.

MrT.

"BradGuth" wrote in message
...
And as you've suggested, a pair of 16 bit captures could also take
full advantage of the 140 db worth of DR, or even up to 192 db should
a given cmos imager offer as much.

No, I said extra scans are unnecessary *if* your sensor provides 140dB in
one scan. Simply use 2 or more A/D's.

In other words, at 192 db of
dynamic range (similar to the best human eye),

The best human eye cannot manage that range in one pass either. It can only
do so by adjusting the pupil diameter. But then the problem of the eyes
color accuracy at extreme sensitivities goes to hell in any case.

Just as it is now possible to record audio with more accuracy than human
hearing, it should be possible to record images with more accuracy than
human vision in the foreseeable future, IMO. The only thing missing at the
moment is suitable sensors AFAICT. Possibly not even that :-)

Of course you do realise such an image can never be printed with even the
slightest accuracy?

MrT.

Bhogi wrote in
:

No need for floating point. I don't think there's a dslr that can
collect more than 65000 electrons per pixel. Smaller pixels can hold
even less. All you need is a 16bit A/D at unity sensitvity (1 electron
= 1 numerically).

The Canon 5D and 1D2 have a maximum of about 80,000, at "ISO 50".

Pixel density has move higher, without well depth increasing with the newer
models, so new FWCs are much lower. A breakthrough in well depth will be
needed before FWCs start climbing again.

On Sat, 31 Jan 2009 02:29:42 -0800 (PST), Vance
wrote:

On Jan 29, 2:51*pm, BradGuth wrote:
On Jan 29, 12:43*pm, "Mr.T" MrT@home wrote:

"BradGuth" wrote in message

...

And as you've suggested, a pair of 16 bit captures could also take
full advantage of the 140 db worth of DR, or even up to 192 db should
a given cmos imager offer as much.

No, I said extra scans are unnecessary *if* your sensor provides 140dB in
one scan. Simply use 2 or more A/D's.

That's even better, because 12 bit ADCs are relatively dirt cheap, and
16 bit can't be all that much worse off. *Fast terabyte media storage
could get a little spendy.



In other words, at 192 db of
dynamic range (similar to the best human eye),

The best human eye cannot manage that range in one pass either. It can only
do so by adjusting the pupil diameter. But then the problem of the eyes
color accuracy at extreme sensitivities goes to hell in any case.

I agree, the 140 db of the cmos imager would always far outperform the
human eye, not to mention at ten thousand fold faster response.



Just as it is now possible to record audio with more accuracy than human
hearing, it should be possible to record images with more accuracy than
human vision in the foreseeable future, IMO. The only thing missing at the
moment is suitable sensors AFAICT. Possibly not even that :-)

Of course you do realise such an image can never be printed with even the
slightest accuracy?

MrT.

Image accuracy is always the eye candy that's in the eye of the
beholder. *I've called it observationology, and it's entirely
subjective as well as in need of deductively interpreting no matter
how good or bad the image technology is. *If you can't deductively
think and thus interpret for yourself, you're kind of screwed into
accepting whatever others (your peers) have to say. *Unlike most, I
still like to think for myself, and I never saw one stinking image of
anything looking WMD worthy, but then I'm not as smart as our SEC and
Madoff either.

*~ BG

Do you actually pictures?

Do you actually English?

--
John

John O'Flaherty wrote:

Do you actually English?

The usenet editor position was abolished a long time ago.

On Sat, 31 Jan 2009 13:56:54 -0500, Alan Browne
wrote:

John O'Flaherty wrote:

Do you actually English?

The usenet editor position was abolished a long time ago.

About the time the censor was dismissed?
--
John