On Mon, 01 Jun 2009 18:24:28 -0400, nospam
wrote:

In article , Eric Stevens
wrote:

The "RAW data file" is merely a file containing the
camera raw data. The only part of the file that relates
to the image is the data it contains. Which is to say
that we *are* talking about the "RAW data", even if you
want to call if a "RAW data file". It's the same data
either way.

Nope. What comes out of the sensor is not what is saved in the RAW
file. There is a transformation involved.

what type of transformation and from what to what?

There is no single answer.

that's ok, just one example is fine.

It depends on the camera and the
manufacturer. I suppose there are cameras which save a RAW file which
is nothing but a bit-map of the sensor

actually, just about every camera.

but generally the RAW file is
modified in some way.

nothing that would affect bayer interpolation. there may be some noise
reduction or in the case of the nikon d2x, white balance.

In the case of the Nikon D300 there are several
choices as to how the RAW file may be saved:

Compressed, lossless compression or lossy compression.

compression is not modification. it's there to save space. users like
being able to put more photos on a card so smaller raw files is a plus.


now, the lossless compression is a modification, although nikon claims
it's 'visually lossless.' it's also optional and it's a tradeoff for an
even smaller size image.

12 bit or 14 bit.

that's not a modification, that's choosing the resolution of the a/d
converter.

I don't know the details of this but I expect that some of it is
described in the Nikon Software Developers Kit. Whatever it is that
happens, clearly there is a transformation of some kind.

if it's so clear, where's the evidence?

With all these changes to its form, the original data has clearly been
transformed. This does not mean that its meaning has been changed.
Merely that its form has been changed.

It is
interesting that the D300 manual specifically says of the lossy
compression:

"NEF images are compressed using a non-reversible algorithm,
reducing the file size by about 40-55% with almost no effect on
image quality"

This contradicts what I was saying to Floyd: it is not possible to go
back to the original sensor image from the RAW file. There will be a
range of slightly different sensor images which could produce the same
RAW file.

lossy compressed raw is the only case where you don't have an exact
sensor dump (but it's *very* close). fortunately, the d300 offers
uncompressed and lossless compressed.

That sounds like a change in the way raw data from the sensors have
been interpreted and saved to the RAW file.

no it doesn't. the first is embedding the jpeg in addition to the raw
data and the second is how it's compressed. the third is for the size
of the jpeg file. raw files are always full size, with canon's sraw
being an exception (and since this is a nikon d300, not applicable).

See above. The RAW data can be compressed independently of the JPEG
compression.

yea and?

Haw! You really don't understand what Bayer interpolation is all
about.

if anyone doesn't understand it, it's you. nowhere in what *you*
quoted says the data in the raw *file* is interpolated.

I've got to plead guilty to that. In a moment of brain fade I got
sucked into what Floyd was trying to talk about rather than the
original topic which I was trying to talk about. Rereading all this
below I can see that I had become more than somewhat confused.

at least you realize you're confused.

Your turn now. Have a look at what I am trying to say about the
meaning of transformation. 1/4 = 0.25



Eric Stevens