D70 resolution: 3008x2000 - what's the 8 all about

The D70 resolution is 3008x2000 pixels.

What is the extra 8 pixels all about?

Not that it matters I suppose, just curious.

 

3000 isn't divisible by 16, 3008 is.

The answer to your next question is to allow lossless rotation of JPEGs

--
John Bean

The most likely way for the world to be destroyed, most experts agree, is by
accident. That's where we come in; we're computer professionals. We cause
accidents (Nathaniel Borenstein)

 

Why would you care? The Sony sensor used in the D70 doesn't have the 3:2
ratio anyway. A 35mm film is 36mmx24mm. Sony's chip is 23.7mmx15.6mm and
that makes the crop factor comes out to somewhere between 1.519x and
1.538x. Just for comparison, Canon's sensor is 22.5mmx15mm which is
exactly 1.6x.

 

And MY next question is "How does 3008x2000 equate to the advertised 6.1
megapixels?" Seems to me it should be 6.016 megapixels.

Merritt

 

One presumes that marketing are allowed to round anything up!

Actually 3008 x 2000 is:-

6.016 million pixels

5875 KP (if K = 1024)

5.74MP (if M = 1024 x 1024)

So if the computing definition of mega were used, the camera is only
5.74MP!

(Some people complained a while back that not all "128MB" CF cards had the
same capacity, and indeed it seems that some manufacturers use the MB =
1000000 bytes and other use MB = 1024 * 1024 bytes).

Cheers,
David

 

You'll have to ask the whoever writes the advertising copy ;-)

FWIW my Pentax DS (with the same sensor) delivers 3008x2008 raw files, but
crops JPEGs to 3008x2000.

--
John Bean

In all large corporations, there is a pervasive fear that someone, somewhere
is having fun with a computer on company time. Networks help alleviate that
fear (John C. Dvorak)

 

Why would you care? The Sony sensor used in the D70 doesn't have the 3:2
ratio anyway. A 35mm film is 36mmx24mm. Sony's chip is 23.7mmx15.6mm and
that makes the crop factor comes out to somewhere between 1.519x and
1.538x. Just for comparison, Canon's sensor is 22.5mmx15mm which is
exactly 1.6x.[/QUOTE]

why post an answer if you are totally clueless?

Lourens

 

Lossless rotation can be maintained by merely keeping the number of pixels the
same, eg: 3000 x 1500 -> 1500 x 3000 (or whatever dimesnions you please).

The 3008/16 is, as you imply, simply adopting the digital 'rounding' boundary
that takes place with most digital devices.

Cheers,
Alan

 

In message <>,

There is a border of RAW image data around the edges that most cameras
making JPEGs or RAW converters will toss away in the process. There are
also areas of pixels that are not exposed to light at all.
--

 

Alan Browne wrote:
[]

Alan, I think you will find that lossless rotation requires that the image
dimensions be a multiple of 16 (and not 8 as I incorrectly stated
earlier).

Cheers,
David

 

I fail to understand why. A matrix of 7 x 11 has the same quantity of
information as a matrix of 11 x 7. Assuming the rotation is correctly done (and
after conversion of individual R,G,B sensor to RGB pixels) then it makes no
matter at all.

Cheers,
Alan

 

If the image is JPEG, then the indivdual blocks are 16 x 16. The size
quantisation restriction only applies to JPEG images, but of course almost
all cameras do allow JPEG images as output.

If the image is not JPEG then it can, of course, be losslessly rotated at
any size - as you say.

Cheers,
David

 

<rant>
It seems mega has two different meanings, in storage it's 1,024x1,024. But
in pixels it's the actual number. Same math my ISP uses for throughput speed
I also would prefer CF cards speeds be in mbps not 40X, 80X etc..
</rant>

 

Are the photosites in the sensor grid perfect squares? a 7x11 grid of
rectangles is not the same 7x11 as it would be at 11x7

 

Why not. As the ratio indicates, there is more width info than height in
the sensor (0.3mm wider). It might even translate into 8 pixels...

 

Darrell wrote:
[]

You will be delighted to hear that both my SanDisk Ultra II CF and my
wife's Kingmax Platinum SD cards do indeed carry a read and write MB/s
rating....

David

 

Thank you David, exactly correct

 

The correct answer is already given. 3008 is possible to divide by 16.
Thats it - nothing more to see here - please move along.


/Roland

 

Now I'm really confused. Nothing prevents me from editing a JPG image to 1
pixel in either dimension. I can easilly come out with a 111 x 213 pixel image
in JPG. From there, I can rotate it to 213 x 111 and it contains the same info.
I can then save it at full quality. I've lost nothing.

Getting back on track, the reason we end up with 3008x2000, IMO, (D70, Maxxum
7d, et al) is that the hardware and software can be designed in even boundaries
of 16 (or 8) and that is convenient to the engineers in terms of performace/$.

 

On the sensor in the D70 and Maxxum 7d I'm not sure they are really perfect
squares. The data on dpreview suggests perhaps not. (pixel dimensions v.
sensor dimensions. It's not clear what is 'cropped' by the camera when reading
the device).

In the Canon 20D they do appear to be square..