2012/2/16 Maksym Veremeyenko <ve...@m1stereo.tv>:
> Hi,
>
> during working with line compositing function i faced a strange value for
> *weight* parameter - 65534
>
> that value is a rounding problem that flag fast-math used.
>
> i prepared small code:
>
> #include <stdio.h>
> int main()
> {
> float mix = 100.0f;
> int r1 = ((1 << 16) - 1) * mix / 100;
> int r2 = (((1 << 16) - 1) * mix + 50 ) / 100;
> int r3 = 65535.0f * mix / 100.0f;
> printf("r1=%d, r2=%d, r3=%d\n", r1, r2, r3);
> };
>
> compiling running examples:
>
> [root@dev-5 core]# gcc -o q -ffast-math q.c
> [root@dev-5 core]# ./q
> r1=65534, r2=65535, r3=65534
>
> [root@dev-5 core]# ./q
> r1=65535, r2=65535, r3=65535
>
> may be it has a sense to fix such rounding issues:
>
> - int weight = ( ( 1 << 16 ) - 1 ) * geometry.item.mix / 100;
> - uint32_t luma_step = ( ( 1 << 16 ) - 1 ) * geometry.item.mix / 100 *
> ( 1.0 + softness );
> + int weight = ( ( ( 1 << 16 ) - 1 ) * geometry.item.mix + 50)/ 100;
> + uint32_t luma_step = ( ( ( 1 << 16 ) - 1 ) * geometry.item.mix + 50
> ) / 100 * ( 1.0 + softness );
>
OK, I applied this change in my branch holding your sse2 commit. I
created the commit it in your name. Also, in that branch, I
refactored the dissolve function in transition_luma.c to use the
composite_line_yuv() from transition_composite.c to benefit from that
improvement.
--
+-DRD-+
------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
Mlt-devel mailing list
Mlt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlt-devel
