I was just wondering if anyone is familiar with JPEG compression.

I have written a compression algorithm in Delphi which seems to lose a little more of the crispness than the JPEG output from a program like PSP8.

Not that its overly important.. I was just wondering if anyone might know the cause of the problem, and how i might go about fixing it.
Posted on 2004-04-21 07:03:08 by Sentient
It might be that the DCT (discrete cosine transform, iirc - anyway, the "nice and special coding" used in JPEG) has been implemented lossily. I believe intel released an okay fast MMX iDCT... which had some quality problems. Unfortunately, this code crept into a whole bunch of projects.

I could of course be totally wrong, too - don't have any references handy, sorry.

Might also be something as simple as the 'quality' parameter being set low for the delphi code?
Posted on 2004-04-21 09:21:55 by f0dder
Im assuming that it is something in the DCT - most probably the fact that I truncate from floating-point to integer quite early in the piece.

All my code is from scratch, and slow as hell to ensure that its correct - which it seems to be. Might try slowing it more by using all floating point code to see if it maintains better detail.

Thanks Fodder
Posted on 2004-04-21 15:05:38 by Sentient