View Single Post
  #4  
Old October 5th 07, 01:50 PM posted to rec.photo.digital
mike
external usenet poster
 
Posts: 10
Default Compress different parts of a pix by different amounts?

Paul Bartram wrote:
"mike" wrote

I don't know the proper terms, but here's the idea.
Take a big pix at high resolution (eg spectators at a football game).
Use a combination of pixel scaling and compression to
get a highly compressed picture, small file size, picture 1.
Print picture 1.
Print the original picture.
Take scissors and cut out small parts of the high resolution
(eg a face)
pix and glue them on Picture 1 in the right places.

Wanna do that digitally.


Don't know if this is exactly what you mean, but XAT.COM Image Optimizer has
a 'magic compression' setting that only selects areas of unchanging colour
and density and hammers those while leaving the intricate areas alone. I use
it for compressing files to below a certain size for uploading to a website,
and it seems to work well. I think the older versions are now freeware, or
the current ones are available for trial.

Paul


Thanks for the tip.
I tried it. In an uncontrolled experiment on one picture, I found it
to be ineffective.
Started with an 800KB jpg at 98% quality level.
Loaded that into the xat program. With the quality slider at 50% and
"magicness slider at 50% the
output file size was about what I got with Irfanview.
Maximum magicness reduced the file size a few percent.
Minimum magicness increased the file size 20%.
If you just move the magicness slider, you do see a difference in file
size, mostly on the bigger end.
But when compared to Irfanview, it's not very impressive.
You get more benefit from reducing the quality level a point.

For the pictures I tried, it's not very magic.

As I described in the original question, the crowd picture
is gonna be very busy. I don't expect anything automatic to
work.
I think I need to compress MANUALLY SELECTED areas differently.
I remember tripping over such a program once, but have no idea the
context or location. And my freeware constraint is likely to be
a show-stopper.
mike

--
Return address is VALID!