[Digikam-devel] Re: PGF files in KDE

[Gilles Caulier]

2011/5/16 Sven Burmeister <sven.burmeister at gmx.net>:
> Am Montag, 16. Mai 2011, 10:36:17 schrieb Gilles Caulier:
>> 16 bits colors dept ==> 48 bits by pixels ( Red = 16 bits + Green = 16
>> bits + Blue = 16 bits)
>>
>> In fact one pixel = 4*16 bits = 48 bits, not 24 bits.
>>
>> 24 bits by pixel want mean 3*8 = 24  ( Red = 8 bits + Green = 8 bits +
>> Blue = 8 bits)
>>
>> Do not confound color depth by color component and and color depth for
>> all components for one pixel. One pixel is 3 color components...
>
> Ok, so you say that although one picture shows 24 bpp and the other 8 bpp they
> are both 8 bit pictures.
>
> So how do I get a PGF that holds the same information the JPG holds, no less,
> no more and has a compareable size, i.e. not three times as big?

PGF do not use the same compress method than JPEG. It use wavelets compress.

If you use lossless compression, ratio will be bigger than JPEG. It's
normal. Same behavior can be see with JPEG 2000. JPEG2000 and PGF use
similar method to compress and ration are the same.

There is lossless method compress in JPEG, but this is not implemented
in libjpeg. In fact it's patented and only used in medecine world.

If you want to compare PGF and JPEG you need to use same compression
method : not lossless

Gilles Caulier