Hello I'm drawing tinted bitmaps and work with the blending modes.
I understand that tinting multiplies the colors and the alpha of bitmaps before drawing them.
But when I use blending mode ADD / ALPHA / INVERSE ALPHA I don't understand exactly what happens. I assume the rgba of the source bitmap are multiplied with the source's alpha and the destination bitmap's rgba with the inverse alpha of the source bitmap.
To me it seems that the multiplying with alpha/inverse alpha only lightens or darkens the colors of the bitmap, but not making them less or more solid or visible. Maybe anyone can explain it in a good way?

Okay, the problem is that when blending I don't understand why the r, g and b components of the pixels are multiplied with the alpha/inverse alpha components, too, because I think that changes their color/tone only, and not their solidity/visibility. So to me it looks like that multiplying the color with the alpha/inverse alpha would Change their intensity or color tone which I don't want, that's what tinting is for. So I wonder why it does not suffice to just tint with 1, 1, 1, 0.5 to let the color unchanged and change visibility of the pixel to 0.5 and the blending mode (ADD, ALPHA, INVERSE ALPHA) then does the change of the visibility. Why are in that blending mode the RGB multiplied by alpha/inverse alpha, too? That changes the color I think.

Tinting is done before blending, as Elias pointed out.

Tinting is done with the vertex color of the triangle(s). It's essentially free, and allows you to 'tint' the vertex. It multiplies each color in the texel by the amortized distance from the vertex and the color of the vertices after they've been weighted.

Tinting :
Weighted Vertex color 0.5,0.5,0.5,0.5 (using premultiplied alpha here)
Texel color 0.8,0.6,0.4,1.0 (source texel)
Result 0.4,0.3,0.2,1.0

Note the color changed, because when you apply alpha to a color, it can only make it lighter (less to contribute to the final color). That's why it's being pre-multiplied here.

Closely related, but slightly more involved is the blending process.

This uses an operator, and two factors to determine the blended pixel color.

Now if you're drawing a translucent image, and you want it to draw the alpha correctly, you use pre-multiplied alpha blending.

al_set_blender(ALLEGRO_ADD , ALLEGRO_INVERSE_ALPHA , ALLEGRO_ALPHA);

This will give you SRC*INVERSE_ALPHA + DEST*ALPHA for the blending mode.

Example :

Source      texel color 0.0 , 0.7 , 0.0 , 0.7 (PM alpha again)
INVERSE_ALPHA         * 0.3 , 0.3 , 0.3 , XXX
---------------------------------------------
                        0.0 , 0.21, 0.0 , XXX
+

Destination texel color 1.0 , 0.0 , 1.0 , 1.0
ALPHA                 * 0.7 , 0.7 , 0.7 , XXX
---------------------------------------------
                        0.7 , 0.0 , 0.7 , XXX

---------------------------------------------
                        0.7 , 0.21, 0.7 , XXX

So you can see it overlays a translucent green pixel over a hot pink pixel and ends up slightly desaturated due to the addition of green.