05-25-2005, 02:35 PM
This is to announce the release in a couple of weeks of a FB jpeg loader for gfxlib.
You have at the moment the possibility of loading jpegs in a portable way using SDL or Allegro (Allegro's loader is by lillo !!)
I'm trying to make my old (1999) QB code into a lib with these functions:
Functions generating an image:
Load Image to Gfxlib buffer (uses gfxlib API)
Load Image to general buffer (without any gfxlib call)
Functions not generating an image:
GetImagesize (just reads image size h X w from file)
Getparameters (prints to a file the jpeg file structure)
PrintError (prints the error code returned by the previous functions)
It will work in true color, not in paletted modes.
You have at the moment the possibility of loading jpegs in portable form using SDL or Allegro (Allegro's loader is by lillo !!)
At the moment I'm able to output a 1Mb Jpeg to screen in 0,44 seconds in a P4 2,4, it compares with what IE does...I'm using lillo's profiler to try to make it faster, but the bit manipulation slows everything down. A 70% of the time is spent in the Huffman decoder.
Ideas, suggestions?
You have at the moment the possibility of loading jpegs in a portable way using SDL or Allegro (Allegro's loader is by lillo !!)
I'm trying to make my old (1999) QB code into a lib with these functions:
Functions generating an image:
Load Image to Gfxlib buffer (uses gfxlib API)
Load Image to general buffer (without any gfxlib call)
Functions not generating an image:
GetImagesize (just reads image size h X w from file)
Getparameters (prints to a file the jpeg file structure)
PrintError (prints the error code returned by the previous functions)
It will work in true color, not in paletted modes.
You have at the moment the possibility of loading jpegs in portable form using SDL or Allegro (Allegro's loader is by lillo !!)
At the moment I'm able to output a 1Mb Jpeg to screen in 0,44 seconds in a P4 2,4, it compares with what IE does...I'm using lillo's profiler to try to make it faster, but the bit manipulation slows everything down. A 70% of the time is spent in the Huffman decoder.
Ideas, suggestions?
Antoni