Warning

 

Close
Confirm Action

Are you sure you wish to do this?

Cancel Confirm
AR15.COM
4/24/2006 9:29:16 AM EDT
Does anyone have a link to a simpleton discussion on how the JPEG picture format works? I'm curious as to how it works, but whenever I do a search on it, all I can find is a ridiculously complictaed technical discussion on Discrete Cosine Transform functions and DC Coefficients.

Any ideas?
4/24/2006 9:29:54 AM EDT
[#1]
tag
4/24/2006 9:30:16 AM EDT
[#2]
PFM
4/24/2006 9:31:50 AM EDT
[#3]
My understanding of it was that it specifies the information for a pixel, and then if the next pixel is the same it just says "same as before" in order to save space.  I could be totally wrong though.
4/24/2006 9:32:11 AM EDT
[#4]

Quoted:
PFM



Pretty much.
4/24/2006 9:32:37 AM EDT
[#5]
jpeg explained
4/24/2006 9:33:40 AM EDT
[#6]

Quoted:

Quoted:
PFM



Pretty much.



Pure F*/King Magic?
4/24/2006 9:34:23 AM EDT
[#7]

Quoted:
My understanding of it was that it specifies the information for a pixel, and then if the next pixel is the same it just says "same as before" in order to save space.  I could be totally wrong though.



Pretty much my understanding of it too.

As you dial up the compression, the more the software will say "That's close enough" and just make anything relatively close to a certain color, that color.

Extreme examples will show banding where the color jumps from one shade to another.
4/24/2006 9:34:48 AM EDT
[#8]

Quoted:

Quoted:

Quoted:
PFM



Pretty much.



Pure F*/King Magic?



Yup.
4/24/2006 9:34:54 AM EDT
[#9]
Some info, not sure if it answers what you are asking, though:

Link

More (better)Link2...

4/24/2006 9:36:28 AM EDT
[#10]

Quoted:
jpeg explained



Seems good enough for me. Thanks.
4/24/2006 9:38:06 AM EDT
[#11]
JPG is an image compression technique of taking a line in an image and compressing it to its like color pixels.

Think about it like an old type writer and taking all the abcdefgijklmnopqrstuvwxyz and taking the same letters in a line and compressing it to one letter then having a table of where in the line the other like letters go.

4/24/2006 9:48:12 AM EDT
[#12]
Thanks for the links y'all. Here's what I was able to come up with for my research paper (copyright roboman, 2006)


In order to achieve its impressive file compression, the JPEG file format knows the limitations of the human eye and exploits them in its algorithm. In doing so, it uses a mathematical technique known as DCT, or discrete cosine transform. As explained by Honeyman and Provos, the DCT essentially takes 8 x 8 pixel sections of an image and transforms them into 64 DCT coefficients. In layman’s terms, the DCT function and its coefficients examine clusters of pixels and “simplify” the information in which a human eye cannot normally notice. If the bit values for a pair of adjacent pixels are small enough that your eye will not discern the difference, the JPEG algorithm will assign both pixels the same value, making for much less information to be stored (“JPEG Compression”, 2006). Please note this is a vastly simplified description of how JPEG works. A technical discussion of the DCT function and quantization would yield a separate research paper unto itself!
4/24/2006 9:50:20 AM EDT
[#13]

"In doing so, it uses a mathematical technique known as DCT, or discrete cosine transform."


In doing so, it uses a mathematical technique known as discrete cosine transform, or DCT.

4/24/2006 9:51:36 AM EDT
[#14]

Quoted:

"In doing so, it uses a mathematical technique known as DCT, or discrete cosine transform."


In doing so, it uses a mathematical technique known as discrete cosine transform, or DCT.




Thanks. Rough drafts are just that......rough
4/24/2006 10:33:09 AM EDT
[#15]

Quoted:
My understanding of it was that it specifies the information for a pixel, and then if the next pixel is the same it just says "same as before" in order to save space.  I could be totally wrong though.



That's what LOSSLESS compression schemes, like GIFs, TIFFs, and PNGs do.  JPG is LOSSY compression, meaning that the compression algorithm actually changes pixel colors compared to the original image, in order to reduce the file size.  And the amount of data lost is adjustable (as is the color space), so that you can adjust how much data is lost/changed to improve the compression rate and reduce file size.

This is important, for example, in web graphics, which need to be as small as reasonably possible so that they transfer quickly (use less bandwidth) and don't tie up the web server too long.

MPEG video and MP3 audio are other popular examples of lossy compression, where quality is reduced somewhat in exchange for much smaller file sizes.

-Troy