[MPlayer-dev-eng] explanations needed ot libmpeg2 idct.c algorithm

[mirak]

Michael Niedermayer a écrit :
> Hi
>
> On Sat, Sep 02, 2006 at 01:18:17PM +0200, mirak wrote:
>   
>> hi,
>>
>> I am trying to implement the idct_row and idct_col in coldfire assembly 
>> using MAC instructions.
>> The algorithm used in idct.c seems different than a theoric idct.
>> I use octave/matlab to test the algorithm and don't get an expected result.
>> The algorithm used uses a lot of factorisation, and even when 
>> developping the equations, I don't get the expected equation.
>>
>> I understand the shifts used, why the coef are multiplied by sqrt(2) ans 
>> also tha matrix multiplication behind this, but the block values doesn't 
>> seems to be the right ones.
>>     
>
> questions about libmpeg2 belong to the mpeg2dec-devel mailinglist
>
> [...]
>   
just for the record I found the solution to my problem.

the rows and columns in 8x8 block are  like
0 2 4 6 1 3 5 7
and not  0 1 2 3 4 5 6 7