Could you actually see photos made from RAW files?

wrote:
I am sorry if this topic may have been discussed too many times.
However, I still have difficulties dealing with the concept of RAW
files. Someone suggested that RAW files are like negatives, while-as
JPEG files are like prints.
My question is whether we can physically see a RAW file... I mean
without placing it in the mercy of a software to open it as a JPEG
file (and in the mean time, the software is doing the processing and
converting it into JPEG using their own algorithm to produce what they
consider to be the best JPEG. I agree that perhaps people should
create both RAW and JPEG files when they take pictures.


The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter.

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

I've done it, it works.

Doug McDonald

"mcdonaldREMOVE TO ACTUALLY REACH wrote:
The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter.

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

I've done it, it works.

Doug McDonald

Most of the responses to the OP's questions are more or
less correct. This one is different, as virtually nothing
said above is correct. (Not even the comment about
software for Canon, as no it is not Public Domain at all.)

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

Floyd L. Davidson wrote:
"mcdonaldREMOVE TO ACTUALLY REACH wrote:
The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter.

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

I've done it, it works.

Doug McDonald

Most of the responses to the OP's questions are more or
less correct. This one is different, as virtually nothing
said above is correct. (Not even the comment about
software for Canon, as no it is not Public Domain at all.)


Of course it is correct. Try it yourself. In response to your
troll, I will post on my web site tomorrow an example.

There **is** public domain software for Canon dSLRs, in DCRAW.exe.
You can get source code and look at it for yourself.

You need to tell dcraw to make absolutely plain
(-d or -D, I forget which) TIFFS (-T) and, if cyou so desire,
even gamma = 1.0.

I did it this afternoon. You can see the difference colored
pixels (different colors in the Bayer filter) in the
monochrome (grayscale) TIFF file.

And in the past I actually HAVE done the display that way, using
software I wrote myself. (Convert TIFF to plain genuine "raw"
bitmap using ImageMagick, then write a program to convert that
to 24 bit color as I described above.)

Could you exmplain how you could be so wrong?

Caveat: the dcraw code is dense. I have no 100% veridird what it does
by looking at the code. Don't attack me on this point
unless you have successfully understood the dcraw code which is
executed using -D or -d.


Doug McDonald

There **is** public domain software for Canon dSLRs, in DCRAW.exe.
You can get source code and look at it for yourself.

... CLIP ...

Caveat: the dcraw code is dense. I have no 100% veridird what it does
by looking at the code. Don't attack me on this point
unless you have successfully understood the dcraw code which is
executed using -D or -d.

I also downloaded the dcraw software a few months ago and looked it over
briefly, couldn't make sense of the code in short order so I have up on
it. It was just a curiosity to me so it didn't matter much. I hate how
some code can be so hard to read. Some programmers truly have a gift
there And those that can read it are somewhat blessed.

Doug McDonald wrote:
Floyd L. Davidson wrote:
"mcdonaldREMOVE TO ACTUALLY REACH wrote:
The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter.

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

I've done it, it works.

Doug McDonald
Most of the responses to the OP's questions are more or
less correct. This one is different, as virtually nothing
said above is correct. (Not even the comment about
software for Canon, as no it is not Public Domain at all.)


Of course it is correct. Try it yourself. In response to your
troll, I will post on my web site tomorrow an example.

There **is** public domain software for Canon dSLRs, in DCRAW.exe.
You can get source code and look at it for yourself.

I've been looking at Dave Coffin's dcraw.c for years.
The first thing you find is a copyright notice, because
it is *not* Public Domain software and never has been.

You need to tell dcraw to make absolutely plain
(-d or -D, I forget which) TIFFS (-T) and, if cyou so desire,
even gamma = 1.0.

I did it this afternoon. You can see the difference colored
pixels (different colors in the Bayer filter) in the
monochrome (grayscale) TIFF file.

What's your point? The -d option generates a greyscale
image using individual sensor data without
interpolation. (As noted too, it might be useful for
documents, but not for color imagery.)

And in the past I actually HAVE done the display that way, using
software I wrote myself. (Convert TIFF to plain genuine "raw"
bitmap using ImageMagick, then write a program to convert that
to 24 bit color as I described above.)

Could you exmplain how you could be so wrong?

You get an image that way and it is indeed generated
from the camera's data. It is not a viable alternative
to using an appropriate raw conversion process.

Caveat: the dcraw code is dense. I have no 100% veridird what it does
by looking at the code. Don't attack me on this point
unless you have successfully understood the dcraw code which is
executed using -D or -d.

I'll grant that Coffin does not write easily readable
code. However, it is also true that perhaps a majority
of third party RAW converters do in fact use the code
from dcraw.c. That includes UFRAW, which I use. (And I
wouldn't say the code for UFRAW is any easier to read
than ufraw.c...)

My point was that your comment about Public Domain
software was wrong, and your description of how to
generate an image might well produce something that is
recognizable, but it is not a useful way to generate
images from camera data.

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

In message , Floyd L. Davidson
writes
"mcdonaldREMOVE TO ACTUALLY REACH wrote:
The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter.

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

I've done it, it works.

Doug McDonald

Most of the responses to the OP's questions are more or
less correct. This one is different, as virtually nothing
said above is correct. (Not even the comment about
software for Canon, as no it is not Public Domain at all.)

I think we have a confusion over terminology

RAW files are the RAW sensor data with a meta file and usually an
embedded Jpg for viewing.

As pointed out the actual format of the RAW files depends on the
sensors etc and is normally specific to a manufacturer or a range of its
cameras.

The information on the formats is not Open Source or Public Domain but
the source and or libraries and or formats ARE freely available. I have
the information and software for the Ninkon Raw format.

However whilst it is free to download, just as with GNU, Open Source and
any other software there is a licence.

This is why there are many programs under many licences and costs that
will handle RAW files from most manufacturers.


With the RAW data you can apply the settings that were in the camera
when the picture was taken or change them. This gives you the
opportunity to "re-take" the photo on the computer +/- 3 stops
(depending on your RAW processor) Some such as DxO can even adjust for
lens characteristics.

So you can from a RAW file produce a series of jpegs with different F
stops and white balance settings as though you were still where you took
the picture with the camera.

As such it is more versatile than a conventional negative

If you process the RAW file and turn out a TIFF which is loss less that
TIFF is more like a film negative. Than you can apply "dark room "
techniques on the TIFF to produce prints in Photoshop which is the
equivalent of a darkroom.

The RAW processor is a NEW step in digital photography that gives more
flexibility







--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/

In message , Floyd L. Davidson
writes
Doug McDonald wrote:
Floyd L. Davidson wrote:
"mcdonaldREMOVE TO ACTUALLY REACH wrote:
The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter.

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

I've done it, it works.

Doug McDonald
Most of the responses to the OP's questions are more or
less correct. This one is different, as virtually nothing
said above is correct. (Not even the comment about
software for Canon, as no it is not Public Domain at all.)


Of course it is correct. Try it yourself. In response to your
troll, I will post on my web site tomorrow an example.

There **is** public domain software for Canon dSLRs, in DCRAW.exe.
You can get source code and look at it for yourself.

I've been looking at Dave Coffin's dcraw.c for years.
The first thing you find is a copyright notice, because
it is *not* Public Domain software and never has been.

I think we have a problem with definitions here.

What do you define as "Public Domain"?

Most public domain software has a copy right or license of one sort or
another. In fact I don't know of any that does not.

--
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/

Chris H wrote:
In message , Floyd L. Davidson
writes
Doug McDonald wrote:
Floyd L. Davidson wrote:
"mcdonaldREMOVE TO ACTUALLY REACH wrote:
The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter.

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

I've done it, it works.

Doug McDonald
Most of the responses to the OP's questions are more or
less correct. This one is different, as virtually nothing
said above is correct. (Not even the comment about
software for Canon, as no it is not Public Domain at all.)


Of course it is correct. Try it yourself. In response to your
troll, I will post on my web site tomorrow an example.

There **is** public domain software for Canon dSLRs, in DCRAW.exe.
You can get source code and look at it for yourself.

I've been looking at Dave Coffin's dcraw.c for years.
The first thing you find is a copyright notice, because
it is *not* Public Domain software and never has been.

I think we have a problem with definitions here.

What do you define as "Public Domain"?

Most public domain software has a copy right or license of one sort or
another. In fact I don't know of any that does not.

"Public Domain" means "The total absence of copyright
protection."

In fact, you cannot know of any Public Domain software
that is copyrighted or requires a license to use.

Of course the laws regarding how a work becomes Public
Domain vary from country to country. In the US today it
is relatively difficult for a private individual to
produce Public Domain software! Years ago it merely
meant not explicitly copyrighting something, but we
changed our laws to make copyright automatic, and at the
same time we made it so that it is putting a work in the
Public Domain that requires an explicit (written I
believe, but don't quote me on that) statement.

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

Doug McDonald wrote:
Floyd L. Davidson wrote:
"mcdonaldREMOVE TO ACTUALLY REACH wrote:
The answer is, at least for Canon, yes. There is public
domain software available that will convert the data, which is
intensity data at each pixel, into a pixel-for-pixel file, that is,
8 or 12 or 14 bits per pixel. This is not full color, it
is filtered by the Bayer filter

Here is such a file, converted to TIFF by dcraw and then to 24 bit .bmp
by Photoshop. You can look at it. Blow it up to 200% or larger
and you will see the Bayer matrix.

http://www.scs.uiuc.edu/~mcdonald/IMG_2706.bmp

note: Unix, so it is case-sensitive

You can then take that file and make a 24 bit file from it, by
moving red pixel values to the red byte, blue ones to the
blue byte, and green ones to the green byte, leaving the other
two bytes of each 24 bit number zero. This can then
be displayed on you computer. To get the color right you need to
scale the
R, G, and B numbers correctly. It will display as
a color image, albeit rather dark since each pixel will
be mostly black.

and here is this:

http://www.scs.uiuc.edu/~mcdonald/rgb_2706.bmp


It does not display well because it is so dark.


I've done it, it works.


And finally, I used the Photoshop blur function on that to
do a VERY CRUDE conversion to what Davidson would call an "image"
and here it is:

http://www.scs.uiuc.edu/~mcdonald/blur_2706.bmp

Note that this was lightened up and color-temperature corrected in Photoshop.
It does not look as "peppy" and colored as the default Photoshop
or Canon Digital Photo Professional conversion does, but it looks
remarkably like what you get if Digital Photo Professional
is set to "faithful", including those grayish green leaves.


Doug McDonald

Floyd L. Davidson wrote:


My point was that your comment about Public Domain
software was wrong,

That is true ... the words should be changed to "free".


and your description of how to
generate an image might well produce something that is
recognizable, but it is not a useful way to generate
images from camera data.


Oh, agreed. I was tyring to explain things!

See my other post where I actually did it and present the results.
Yes, they are of course not "presentable" but are good examples
of how it works.

Doug McDonald