Re: [AGG] Blending (again) [resend]

[Wojciech M. <wm...@gm...>]

This is why I said that I *might* have a patch :)
As far as I remember this patch resolved some warnings related to some
high precision types, I believe.
multiply does its caclulations on calc_type, yet returns value_type, eg:

        static AGG_INLINE value_type multiply(value_type a, value_type b)
        {
            calc_type t = a * b + base_MSB;
            return value_type(((t >> base_shift) + t) >> base_shift);
        }

It is an improvement, but a small one.

2014-10-28 23:51 GMT+01:00 Michael Droettboom <md...@st...>:
> I get the same result with the patch. I’m not sure how the patch is meant to
> work, particularly this part:
>
> --- agg_pixfmt_rgba.h    (revision 104)
> +++ agg_pixfmt_rgba.h    (working copy)
> @@ -264,10 +264,10 @@
>          {
>              if (alpha > color_type::empty_value())
>              {
> -                calc_type a = p[Order::A];
> -                calc_type r = color_type::multiply(p[Order::R], a);
> -                calc_type g = color_type::multiply(p[Order::G], a);
> -                calc_type b = color_type::multiply(p[Order::B], a);
> +                value_type a = p[Order::A];
> +                value_type r = color_type::multiply(p[Order::R], a);
> +                value_type g = color_type::multiply(p[Order::G], a);
> +                value_type b = color_type::multiply(p[Order::B], a);
>
> It looks like this actually has less precision than before, or at best is
> equivalent, since r, g, and b were being downcast in the call to lerp
> anyway.
>
> Mike
>
> On 10/28/2014 04:49 PM, Wojciech Mamrak wrote:
>
> Have you tried my patch?
>
> 2014-10-28 21:03 GMT+01:00 Michael Droettboom <md...@st...>:
>
> Any thoughts/pointers about where I could start to fix this?
>
> Cheers,
> Mike
>
>
> On 10/17/2014 02:25 PM, Wojciech Mamrak wrote:
>
> I thought it was more sophisticated, but it turns out that changes to
> revisions [99, 102) were more involving.
>
> 2014-10-17 19:56 GMT+02:00 Wojciech Mamrak <wm...@gm...>:
>
> I might have a patch ready for this, brb.
>
> 2014-10-17 19:42 GMT+02:00 Jim Barry <ji...@ch...>:
>
> On 17 October 2014 17:52, Michael Droettboom <md...@st...> wrote:
>
> In agg-2.4, if I draw a rectangle path with RGB value of (0, 255, 102,
> 64), the resulting pixel value is (0, 255, 102, 64).
>
> In agg-svn, I get (0, 255, 104, 64).
>
> it seems the difference is not so much in anything intentional
> but from the fact that everything but the final demultiply is downgraded
> to color_type, and then the demultiply happens. Whereas in the old
> version, it’s still doing calculations in calc_type (uint16) up until
> the final demultiply.
>
> Yes, I see the problem. I suppose it must have happened when refactoring for
> the floating-point support. I will try to think of a solution.
>
> - Jim
>
> ------------------------------------------------------------------------------
> Comprehensive Server Monitoring with Site24x7.
> Monitor 10 servers for $9/Month.
> Get alerted through email, SMS, voice calls or mobile push notifications.
> Take corrective actions from your mobile device.
> http://p.sf.net/sfu/Zoho
> _______________________________________________
> Vector-agg-general mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vector-agg-general
>
>
>
> ------------------------------------------------------------------------------
> Comprehensive Server Monitoring with Site24x7.
> Monitor 10 servers for $9/Month.
> Get alerted through email, SMS, voice calls or mobile push notifications.
> Take corrective actions from your mobile device.
> http://p.sf.net/sfu/Zoho
>
>
>
> _______________________________________________
> Vector-agg-general mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vector-agg-general
>
>
>
> --
> Michael Droettboom
> Science Software Branch
> Space Telescope Science Institute
>
> http://www.droettboom.com
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> Vector-agg-general mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vector-agg-general
>
> ------------------------------------------------------------------------------
> _______________________________________________
> Vector-agg-general mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vector-agg-general
>
> --
> Michael Droettboom
> Science Software Branch
> Space Telescope Science Institute
>
> http://www.droettboom.com
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> Vector-agg-general mailing list
> Vec...@li...
> https://lists.sourceforge.net/lists/listinfo/vector-agg-general
>