Hello!
XnView is a perfect tool to organize and browse large directories of pictures.
Also the functions to crop and resize pictures would be very helpful.
There is only one problem: I compared jpegs saved with XnView and Fireworks.
For Example: JPEG 400 * 300 pixels size: 40 kbyte
JPEG Quality: Fireworks: good
XnView: middle-bad
My question: Could you work on the the jpeg-compression algorithm?
Thanks
Mika (Germany)
JPEG: Big files - Bad quality
Moderators: XnTriq, helmut, xnview
-
- Posts: 8705
- Joined: Sun Oct 12, 2003 6:47 pm
- Location: Frankfurt, Germany
Re: JPEG: Big files - Bad quality
I'm not deep in the JPEG compression, but I think that there is a standard implementation which is used by most graphic programs, and programs can set/change parameters of the compression, only.
As written in some other posts (e.g. Smaller image has same file size), JPG compression is tricky. Additionally, when compressing with specific JPG parameter, the image might compress well and look good, with other parameters it might not.
Perhaps someone knows or can find out, whether there is a significant difference between JPG compressions.
As written in some other posts (e.g. Smaller image has same file size), JPG compression is tricky. Additionally, when compressing with specific JPG parameter, the image might compress well and look good, with other parameters it might not.
Perhaps someone knows or can find out, whether there is a significant difference between JPG compressions.
Last edited by helmut on Wed Mar 28, 2007 8:45 pm, edited 1 time in total.
Re: JPEG: Big files - Bad quality
Well, lets first talk about compression. You have three types:
Huffman (normal size) common
Optimized huffman (0~2% less) usual
Arithmetic (3~10 less) rare (for lack of readers i think)
There also image matrix fitter. Jpg divides the image into tiles with 8x8 pixels. So it would reduce/increase image size to make whole tiles. (width and height of image should divided over 8 with integer number).
Another thing is modifying content to meet the DCT transformation in a way it produces a more compressable code.
This is only some hints..
yours
MAAD
Huffman (normal size) common
Optimized huffman (0~2% less) usual
Arithmetic (3~10 less) rare (for lack of readers i think)
There also image matrix fitter. Jpg divides the image into tiles with 8x8 pixels. So it would reduce/increase image size to make whole tiles. (width and height of image should divided over 8 with integer number).
Another thing is modifying content to meet the DCT transformation in a way it produces a more compressable code.
This is only some hints..
yours
MAAD