Velocity Reviews > Re: Pixel size of individual Pixel

Re: Pixel size of individual Pixel

Tom Thackrey
Guest
Posts: n/a

 09-14-2003

On 13-Sep-2003, Jim Townsend <(E-Mail Removed)> wrote:

> Pdelpriore wrote:
>
> > Anyone know the dimension of one pixel in the CCD array of theNikon
> > Coolpix
> > 885?
> >
> > Paul

>
> There are no 'pixels' on the CCD array..
>
> The pixels are assembled from the output of a grid of sensors on the
> surface of
> the array. In order to create a pixel, there is one sensor for detecting
> red,
> one for blue and two sensors for detecting green. This is known as a
> Bayer
> filter.
>
> See http://www.pinnaclevision.co.uk/wpbaycol.htm for a brief description
> of
> how this filter works
>
> The camera contains circuitry that measures the analog output of each of
> the
> sensors, then digitizes the result. The processor combines the raw
> digitized
> info from the four sensors to form a single pixel.
>
> However, you *can* determine the number of pixels created per inch of
> sensor
> width/height and from there the size.
>
> The camera produces 2048 pixels across
> The sensor is 7.18mm wide (Or .2827 inches)
>
> The camera produces 1536 pixels vertically
> The sensor is 5.32mm high (Or .2094 inches)
>
> dividing .2827 by 2048 = .00013 inches wide
> dividing .2094 by 1536 = .00013 inches high
>
> So.. each pixel is .00013 inches wide and .00013 inches high. It's pretty
> amazing when you realize that they squeeze four sensors into this area.
>
> This works out to 7692.3 pixels per inch. (1 divided by .00013)
>
> If you had a scanner with this resoluton and scanned a 4x6 photo, you'd
> wind up
> with a 1.4 gigapixel image.

You're wrong about the digital sensor in most cameras. It's true that the
sensor has a Bayer filter which covers groups of four photosites with an
RGBG filter pattern. Your mistake is that the software in the camera (or the
raw converter software) interpolates these four photosites to produce four
RGB pixels, not one. So, each photosite is actually a pixel and the OP's
question was not as foolish as you claim.

In addition, there is a significant gap between photosites, so simple
division will tell you how much of the chip is devoted to each photosite,
but not the actual size of the photosite.

reference:
http://www.bythom.com/ccds.htm

--
Tom Thackrey
www.creative-light.com

Jim Townsend
Guest
Posts: n/a

 09-14-2003

> On 13-Sep-2003, Jim Townsend <(E-Mail Removed)> wrote:

>> There are no 'pixels' on the CCD array..
>>
>> The pixels are assembled from the output of a grid of sensors on the
>> surface of
>> the array. In order to create a pixel, there is one sensor for detecting
>> red,
>> one for blue and two sensors for detecting green. This is known as a
>> Bayer
>> filter.

> You're wrong about the digital sensor in most cameras. It's true that the
> sensor has a Bayer filter which covers groups of four photosites with an
> RGBG filter pattern. Your mistake is that the software in the camera (or the
> raw converter software) interpolates these four photosites to produce four
> RGB pixels, not one.

No.. I never said that.. But I'll agree that I wasn't clear. I should have
said 'A' pixel is created by combining the output of four sensors. Red, Green,
Green and Blue.

> So, each photosite is actually a pixel and the OP's
> question was not as foolish as you claim.

I never said he was foolish But I'll still maintain a photosite isn't a
pixel. It's a group of four sensors who's output is combined to CREATE a
pixel. It may be splitting hairs, but it is a more accurate descripton.. The
sensor isn't made of pixels, it creates them.

This is relatively important since a photosite has size, but the resulting
pixel has no size. It's only a digital representation of the site.

> In addition, there is a significant gap between photosites, so simple
> division will tell you how much of the chip is devoted to each photosite,
> but not the actual size of the photosite.

I did divide and came up with a site size of .00013 inches.. Which was what
the OP wanted to know in the first place

JPS@no.komm
Guest
Posts: n/a

 09-14-2003
In message <(E-Mail Removed)>,
Jim Townsend <(E-Mail Removed)> wrote:

>No.. I never said that.. But I'll agree that I wasn't clear. I should have
>said 'A' pixel is created by combining the output of four sensors. Red, Green,
>Green and Blue.

Not at all. There are as many pixels output as there are photosites.
Each of the three color channels from the bayer pattern:

r r r r

r r r r

r r r r

r r r r

b b b b

b b b b

b b b b

b b b b

g g g g
g g g g
g g g g
g g g g
g g g g
g g g g

are interpolated or resampled to fill in the spaces with solid color for
that channel:

rrrrrrrr
rrrrrrrr
rrrrrrrr
rrrrrrrr
rrrrrrrr
rrrrrrrr

bbbbbbbb
bbbbbbbb
bbbbbbbb
bbbbbbbb
bbbbbbbb
bbbbbbbb

gggggggg
gggggggg
gggggggg
gggggggg
gggggggg
gggggggg

If what you are saying were true, there would be only a 1mp file from a
4mp sensor, and there would be color shifts that resemble chromatic
abberation, only they would shift in rectangular cordinates rather than
from the center of the image.
--

<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<>
John P Sheehy <(E-Mail Removed)>
><<> <>>< <>>< ><<> <>>< ><<> ><<> <>><