PS CS2 compression

hello,

when i down load a large jpg from my rebel xt the file size is around
2.x mb. In PS, if i save it on the 12 setting the file size gets
larger. What is happening to make a jpg bigger and is there any benefit
to the pic? Also in raw (from the camera) it is 8.x mb, and ps raw
gives the option to save as 25mb+. Same question, what is the software
doing to make a small file bigger and is there any benefit?

thanks

ps: any words why the jpgs are so small from the rebel xt. I had a
older canon that the jpg file sizes were twice as big. when I purchased
the xt i figured with the 8 meg sensor the jpgs would have been bigger.

On Thu, 04 Jan 2007 18:31:31 -0600, Dave wrote:

hello,

when i down load a large jpg from my rebel xt the file size is around
2.x mb. In PS, if i save it on the 12 setting the file size gets
larger. What is happening to make a jpg bigger and is there any benefit
to the pic?

It is being saved with a higher 'quality' that it was saved with before.
There is no benefit because lost data cannot be retrieved.

Also in raw (from the camera) it is 8.x mb, and ps raw
gives the option to save as 25mb+. Same question, what is the software
doing to make a small file bigger and is there any benefit?

On my Kodak P850 (5mp EVF) raw files are 8mb, because the raw data is
compressed - a tiff is 15mb. I expect you are seeing something similar. If
the raw data were uncompressed, I'd be expecting about 22mb - 4mp x 4.5
bytes/pixel since raw data is usually 12 bits per channel i.e. 4.5 bytes.


thanks

ps: any words why the jpgs are so small from the rebel xt. I had a
older canon that the jpg file sizes were twice as big. when I purchased
the xt i figured with the 8 meg sensor the jpgs would have been bigger.

Probably more compression. Do you have the picture quality set to 'fine'?

In article ray writes:
$On Thu, 04 Jan 2007 18:31:31 -0600, Dave wrote:
$Also in raw (from the camera) it is 8.x mb, and ps raw
$ gives the option to save as 25mb+. Same question, what is the software
$ doing to make a small file bigger and is there any benefit?
$
$On my Kodak P850 (5mp EVF) raw files are 8mb, because the raw data is
$compressed - a tiff is 15mb. I expect you are seeing something similar. If
$the raw data were uncompressed, I'd be expecting about 22mb - 4mp x 4.5
$bytes/pixel since raw data is usually 12 bits per channel i.e. 4.5 bytes.

12 bits per channel is 1.5 bytes. Don't forget that a RAW file
has data right off the sensor, and each photosite on the sensor
works with only *one* colour, not *three*. So for an 8 megapixel
camera, that's about 12 megabytes for an uncompressed RAW file.
Canon's RAW files are losslessly compressed, so they're of
variable size (depending on how well the data from a particular
image compress; this also explains why higher ISOs result in larger
RAW files, as the extra noise doesn't compress well) and typically
modestly smaller than the 12 megabytes of information they
contain. 7-9 megabytes would be typical sizes for a Rebel XT's RAW
files.

Converting the RAW file applies Bayer interpolation, resulting
in three colours per pixel - so for an 8 megapixel image, that's
three bytes per pixel if you're working at 8 bits, yielding an
uncompressed file size of around 24 megabytes.
--
Stephen M. Dunn
---------------- http://www.stevedunn.ca/ ----------------
------------------------------------------------------------------
Say hi to my cat -- http://www.stevedunn.ca/photos/toby/

ray wrote:
On my Kodak P850 (5mp EVF) raw files are 8mb, because the raw data is
compressed - a tiff is 15mb. I expect you are seeing something similar. If
the raw data were uncompressed, I'd be expecting about 22mb - 4mp x 4.5
bytes/pixel since raw data is usually 12 bits per channel i.e. 4.5 bytes.


Hello. Raw data is indeed 12 bits per channel, but in fact each pixel
only contains information about one channel: thus, 12 bits=1.5 bytes
per pixel, or 7.5mb for an uncompressed file from a 5mp camera (and a
bit more for an embedded jpeg).

The reason there is information about only one channel per pixel is
because it works like this:
http://www.cambridgeincolour.com/tutorials/sensors.htm
(I must admit to not having read it, but the explanations on that site
are usually very careful). In short, there is a filter over the sensor
that allows only one of R, G or B colours to pass for each pixel. Thus
the colour resolution is lower than the luminance resolution (and in
fact there are twice as many green as red or blue detectors), but this
isn't a problem (convert a photograph in photoshop to LAB mode and
smooth the L, then one of the other two channels, and see how much
difference there is).

Anyway. Hope this helps.

On Thu, 04 Jan 2007 19:03:45 -0800, acl wrote:

ray wrote:
On my Kodak P850 (5mp EVF) raw files are 8mb, because the raw data is
compressed - a tiff is 15mb. I expect you are seeing something similar. If
the raw data were uncompressed, I'd be expecting about 22mb - 4mp x 4.5
bytes/pixel since raw data is usually 12 bits per channel i.e. 4.5 bytes.


Hello. Raw data is indeed 12 bits per channel, but in fact each pixel
only contains information about one channel: thus, 12 bits=1.5 bytes
per pixel, or 7.5mb for an uncompressed file from a 5mp camera (and a
bit more for an embedded jpeg).

The reason there is information about only one channel per pixel is
because it works like this:
http://www.cambridgeincolour.com/tutorials/sensors.htm
(I must admit to not having read it, but the explanations on that site
are usually very careful). In short, there is a filter over the sensor
that allows only one of R, G or B colours to pass for each pixel. Thus
the colour resolution is lower than the luminance resolution (and in
fact there are twice as many green as red or blue detectors), but this
isn't a problem (convert a photograph in photoshop to LAB mode and
smooth the L, then one of the other two channels, and see how much
difference there is).

Anyway. Hope this helps.

Thanks to both of you for the clarification.