Re: [pygame] Surface.blit() BLEND_ADD vs. BLEND_RGBA_ADD

[René Dudfield <renesd@xxxxxxxxx>]

Hi,

The RGB ones just do the RGB channels, but RGBA also does the ALPHA channel too.

They are simply adding the channels.  They don't inspect the alpha channel to see if they should add that pixel, or do an alpha blend as well as an additive blend.


These defines show what they do...

#define BLEND_ADD(tmp, sR, sG, sB, sA, dR, dG, dB, dA)  \
    tmp = dR + sR; dR = (tmp <= 255 ? tmp : 255);       \
    tmp = dG + sG; dG = (tmp <= 255 ? tmp : 255);       \
    tmp = dB + sB; dB = (tmp <= 255 ? tmp : 255);

#define BLEND_RGBA_ADD(tmp, sR, sG, sB, sA, dR, dG, dB, dA)  \
    tmp = dR + sR; dR = (tmp <= 255 ? tmp : 255);       \
    tmp = dG + sG; dG = (tmp <= 255 ? tmp : 255);       \
    tmp = dB + sB; dB = (tmp <= 255 ? tmp : 255);       \
    tmp = dA + sA; dA = (tmp <= 255 ? tmp : 255);




I think what you want is probably something different... that can probably be done via some other combination of blits though.


cheers,

On Fri, Dec 16, 2011 at 8:10 AM, Florian Berger <fberger@xxxxxxxxxxxxxxxxx> wrote:

Hi!

Ian Mallett <geometrian@xxxxxxxxx>:

>
> Simply call .convert_alpha() on each surface after you make your
> window.  This clears up most problems.

I am doing that already. In fact I blit the sprite, so I can see that
alpha works; then I create a Surface.copy(), manipulate that and blit
it over the sprite using BLEND_RGBA_ADD.

The result is that the blit incorporates the RGB values of the fully
transparent pixels and adds them as well, resulting in an ugly square
around the sprite.

The solution in my case is to make sure that the transparent pixels
also have a (0, 0, 0) RGB value. But this is annoying. I think this
calls for a bug report. ;-)

Regards,
       Florian