OK, I am not a techie, so let me try.
Imagine your photo as a matrix of 3000 lines on the screen in front of you,
2000 picture elements wide. Work your way across the first line and evaluate
what you see and think how to tell someone else what you see. You don't go
"white, white, white, pinkish, pinkish, red red, red, red...". Instead you
go "3 white, 2 pinkish, a buncha' red, ..."
jpeg and other compression schemes work that same way. The more compression
you demand of the picture, the more approximate the description. And so, the
more loss in the image when you try to reproduce the photo.
Of course jpeg and others don't just work on one line of pixels at a time,
they look for regions with "blobs" of color that is close enough that you
can say "a blob of xyz type, all red shading to pinkish at the edges." etc.
Every time you can find regularities that you can describe, you will save
room compared to storing the image as a large matrix, 3000 x 2000 x 12 or 16
or 24. If you look at your jpeg file sizes, you will see that images with
exactly the same dimensions (e.g., 640 x 480) with exactly the same level of
compression will have very different file sizes. 'Cause some are more
regular than others.
Stan
on 12/06/03 9:04 PM, Tanya Mayer Photography at [EMAIL PROTECTED]
wrote:
> ok, so i get that bit - it was the red pixel thing that had me confused!
> lol...
>
> BUT (isn't there always a BUT?) i understand this completely using your coin
> example with 100 identical coins making up the image, BUT in an image where
> the entire thing is made up of separate colours, details etc - how can all
> of those colours etc be represented in just one "coin" section iykwim?
>
> tanya, who may be off with the fairies on this one (should i start signing
> "fairygirl" again...?)
>
> ----- Original Message -----
> From: "John Francis" <[EMAIL PROTECTED]>
> To: <[EMAIL PROTECTED]>
> Sent: Sunday, December 07, 2003 12:34 PM
> Subject: Re: JPEG to TIFF or PSD Conversion
>
>
>>>
>>> Woooooooooooosh!
>>>
>>> That was the sound of everything that John just explained going STRAIGHT
>>> over tanya's head......
>>>
>>> tan. (who is completely mind-boggled at this point...)
>>
>> OK. Let's try something different.
>>
>> Imagine a very simple image - a plain white cloth, with 100 identical
>> coins on it, laid out in a 10x10 square.
>>
>> You're looking at that image in front of you, and you want to tell me
>> how to recreate that same scene. You could send me a separate picture
>> of each of the 100 coins. Or you could send me a picture of the 'heads'
>> and 'tails' of just one coin, and just one bit of information for each
>> of the 100 coins in your picture: whether it shows 'heads' or 'tails'.
>>
>> with that information (which will only take up about 1/50th of the
>> space of sending me all 100 detailed pictures of the individual coins)
>> I can recreate an image which will look almost identical to the image
>> you have in front of you.
>>
>> With me so far? Good. Now comes the hard part.
>>
>> The 100 pictures of each of the individual coins corresponds to a
>> TIFF (or PNG, or PSD, ...) image.
>>
>> The two detailed pictures, plus the information as to which of those
>> pictures ('heads' or 'tails') should be used at each position in the
>> 10x10 square, corresponds to a JPEG image.
>>
>> The 'JPEG' will be much smaller than the 'TIFF', but it enables me
>> to recreate an image that is very close to the original image you saw.
>> Not quite identical; your coins probably have some distinguishing wear
>> marks or scratches on them. But for most purposes my recreated image
>> will be equivalent to your original.
>>
>> Only for most purposes, though. There are some questions (example:
>> which of the coins has a small hole drilled through the top of it?)
>> that you, with the original in front of you, can answer, but which
>> I can't (assuming all I have is the 'JPEG'). If I've got the 'TIFF'
>> representation, though, I will be able to answer the question.
>> That's an example of how there is information loss in a JPEG image.
>>
>> Still with me? Then let's complicate it even further.
>>
>> Suppose I want to tell Shel how to recreate my version of the image.
>> I could send him a 'TIFF' form of my image, with detailed pictures
>> of all one hundred of my coins. But if I recreated my image using
>> your 'JPEG' version then those won't really be one hundred different
>> pictures - all the 'heads' will be exactly identical, as will all
>> the 'tails'. Once information has been lost, it can't be recovered.
>>
>
>
