Sorry Don I tried to take bites out of the tomato but couldn't quite get it right.

Christine,

The effect you have achieved is very interesting if you look at the blue. A bit like a thick liquid. Very effective if magnified. Almost like the blue of the tumbler is melting.

Don

Corrected it the best I could.

OK Christine, I think we've spotted the dummy. Nice. ....................... but nicking the sauce, thats sneeky. A very clever job on the bottle. Still retains the plastic look.:)

Don

I thought the grub was a nice touch. LOL!

Well I got that one wrong. Missed the grub. Should have photographed the sarnie cut in half for that.:D

Don

Don

A couple of things are happening here. Your file contains image data and what is called meta data. Lets deal with the easy bit first. Meta data is mainly text that contains such things as your camera Exif info, the time and date when the file was edited, the names of the programs that have edited the file and so on. In recent versions of programs data is included to provide evidence ( for legal purposes) that an image has been tampered with by editing. So each time you save a bit more info is tagged on.

The other bit of file size change relates to the jpeg encoding and decoding. A lot of current technology is based on mathematics developed 200 yrs ago. A group of mathematicians in france ( namely fourier and laplace ) developed the concept of transforms. Without getting too involved with the complexities lets convert something that is difficult to work with into something that is simpler to work with.

e.g if we take a heap of iron ore and process it ( transform ) into uniform steel rods. We can work with the steel rods. Then if nature takes it's course the rods will revert back ( transform ) to the heap of iron ore.

In the case of jpeg the concept is somewhat more difficult to grasp if your are not from these disciplines. Fourier came up with the concept that any waveform ( think of a sound wave of music ) can be made of a number of component sine waves of different frequencies and amplitudes.

So the theory is a sound can be dismantled and reassembled using these components.

In the case of a picture we are converting data from a space domain into a time domain. If that statement has just made your brain blow a fuse no worries it's not an easy subject.

The point is that calculations are performed on the image to encoded and decode the data. The result will not be an exact fit into one of the 256 binary values in the image. This is one of the minor drawbacks of the digital world is the result has to be rounded up or down to fit.

e.g 106.86568 will be rounded to 107
this rounding produces what is know as digital noise.

This effect is one of the factors that contribute to image noise.
The other being produced by the camera sensor due to heat (thermal shot noise) and you could include gamma rays from outter space !!

This low level of noise added to the image effectively increases the number variations in image data and thereby reduces the compression efficiency.

The highest efficiency is achived if the whole image contained the same data value. A very boring image at that :D

One of the major benefits of image noise reduction programs is the smaller file sizes produced.

I hope that was not too heavy for you folks but I did try to make it simple.

Phew, thanks Rob,

I need to read this a few times.

I really first noticed the file size bit while we were doing stuff on the ' in camera processing ' thread.

Using NC to convert a raw to jpeg gave a file size twice that of in camera jpeg. Could not understand, so assumed that the NC converted must contain more info. Did my crazy pixel by pixel analysis and could not find it. :confused:

Thought I would come back to that later, and went no further.

Now as mr. average ( no specialist knowledge of the workings of computers ) I would assume that in digital photography, bigger file size = more available actual image information in that file. Obviously now an incorrect assumption.

As this info relates to two threads, I will pose a question that relates to this thread and one that I could not previously investigate due to lack of processing power. I will post a link to your info in the other thread.

Now I have my image, what advantage is there in upsizing the file ??

This comes from looking at DPR where some guys are upsizing 4.1mp D2Hs files as a matter of course.

Don

Strickly speaking this assumption is true. It depends what you think "actual image information" as. A endoder cannot tell the difference between noise and wanted image data. Mind you some encoders did attempt to spot low level rapid changes and smooth them to improve the compression of data.

With experience you should be able to look at an image and gauge how well it will compress. Is it worth starting a dummies guide to data compression seeing so many places impose a file size limit ?

Rob,

I think thats a good idea.

This is something I would be interested in. Thought about it today and am going to do a test using XP, 2000, and 98SE. to see if the modern program writes more info to the file.

Personal example ; We moved to Norfolk, and are now 5 hours one way from my mother. I now write using word ( her eyesight is not what it was ), and add pictures to make the letters more interesting. When we joined the RSPB and started visiting Tichwell, I took lots of pictures, some became joiners. I downsized and compressed these, as that letter was to have around 10 pics and the printer stuggled with that. Those with large areas of blue skies showed the dreaded banding. The result of overcompression.

If I was new to this forum and posting a gallery pic, and did not understand compression at all, I could be quite frustrated posting that glorious summer scene and looking at the final result in the gallery and not understanding why.

Don