[XviD-devel] iDCT cooefficients different...
Winnie Weber
xvid-devel@xvid.org
Thu, 25 Jul 2002 12:42:13 +0200 (MEST)
>The standard allows (I)DCT coefficients to slighty differ from
>reference implementation, as far as I know, +-1 is the threshhold, so it's
>rounding effects are covered, but "calculation errors" are not.
Hm, very interesting to hear.....
What about a sequence where the 'correct IDCT value should be '0, but the
MMX version yields '1' for a duration of 180 p-frames?
In fact, we have this effect here: The image is getting worse, because
the ('wrong') MMX values of 1 add 'something' to the image which is
NOT correct. And 180 frames is a LONG time.....
At next I-frame, we notice a 'hard' picture pop, if we get back to the
'correct' picture.
In a word, getting 0 or something different from 0 in IDCT is very
important, isn't it?
Winnie
--
GMX - Die Kommunikationsplattform im Internet.
http://www.gmx.net