Re: Deconvolution of Sensor Anti-Alias filter and length of coherence?
- From: davem@xxxxxxxxx (Dave Martindale)
- Date: Fri, 30 Sep 2005 16:14:57 +0000 (UTC)
Ilya Zakharevich <nospam-abuse@xxxxxxxxx> writes:
>[A complimentary Cc of this posting was sent to
>> You can massively oversample audio, then do the brick-wall filter
>> cheaply digitally. Oversampling images is not practical because of
>> loss of sensitivity (light-collecting area), fabrication problems, and
>> the amount of data that would result.
>Obviously, there is neither loss of sensitivity, nor fabrication
>problems (at least as you go down from 8micron to 2.2microns). And I
>hope that the last problem you mention will disappear in 2-3 years
>too.
We had this argument a couple of months ago. Everyone except yourself
seems to agree that making smaller sensor pixels reduces useful
sensitivity as well as dynamic range. In audio, increasing the
digitizer sample clock does not do that.
If you're building a DSLR sensor, you could use current technology to
build a sensor with 2-3 um pixel pitch instead of 8 um. But if you're
talking about a P&S camera that *already* has a 2-3 um pixel pitch,
you're going to have sever problems reducing that by a factor of 4.
Handling more data could be now, at some increase in cost, if the other
problems were not there and the benefits of oversampling seemed worth
the cost. I really don't think memory or signal processor speed is the
important limitation.
>IMO, this is definitely a factor. However, a larger factor is that
>the price of similar-quality lens (measured via MTF at a given F-stop)
>decreases *enormously* when the sensor size decreases. Thus current
>small-sensor cameras have lenses with MTF one cannot even dream of for
>sub-$100000 budget in 35mm format.
If you measure MTF in terms of line pairs per mm. But the small-sensor
images have to be enlarged further, cancelling much of this advantage.
When comparing different sensors, it makes more sense to compare in
units of line pairs per picture height (or width). Then the lenses on
small-sensor cameras won't look so good in comparison to 35 mm or larger
format lenses.
Dave
.
- References:
- Deconvolution of Sensor Anti-Alias filter?
- From: winhag@xxxxxxxxx
- Re: Deconvolution of Sensor Anti-Alias filter and length of coherence?
- From: David J Taylor
- Deconvolution of Sensor Anti-Alias filter?
- Prev by Date: Re: About Sony memory stick Pro - Magic Gate
- Next by Date: Re: P-880 sample for you
- Previous by thread: Re: Deconvolution of Sensor Anti-Alias filter and length of coherence?
- Next by thread: Re: Deconvolution of Sensor Anti-Alias filter and length of coherence?
- Index(es):
Relevant Pages
|
