Section 5.1 - Selection Masks

We have always applied the effects to the whole bitmap. It will sometimes be useful to use an intermediary project to "filter" this effect:

Example: You have two projects 0 and 1, a mask in project 2 and want to draw the result in project 3:

0:AA ; 1:BB ; and  2:01 will result in 3:AB

In this example, you could have obtained project one applying some effect to project zero.

This works a bit like an alpha channel, project two holding that alpha value (of course projects one and two may be two completely different loaded pictures).

Let's say that a value of zero in project 2 results in outputing the value of project 0 and a value of 255 outputs project 1.

To read from zero, you put r(0,x,y)*(255-r(2,x,y))/255, and reading from one is

Note that if you only use 0 or 255 (i.e. none or full alpha), you may make your mask have only values zero and one, so that you aren't required to divide by 255 everytime.
Note well that r(r(2,x,y),x,y) is not allowed, as the project number must be a constant value! (Because of freezing problems)

Well, now add the two values, to obtain: (r(0,x,y)*(255-r(2,x,y))+r(1,x,y)*r(2,x,y))/255
In all components, you obtain:
You may also read the different components from the mask to have different transparencies for the different components, if you want, instead of having one mask for the whole picture.

To obtain these masks, you can use colour extraction (section 4.1)
Some mathematic formula (e.g. implicit graphes, section 3.2, but anything else is ok :-))
You can combine these two ways. If for example you did some colour extraction and got some superfluous areas, you can try to find a function that covers everything but these areas and multiply both.

If you want to add masks (a bit like a logic OR), use the max() function between them. Another way to do it is to multiply their inverses, something like r(2,x,y)=255-(255-r(0,x,y))*(255-r(1,x,y)).

And to apply a mask to a transformation, you can 'multiply the changes' by c(k,x,y)/255, replacing c and k accordingly.
This is mostly useful for instance with blur and other convolution effects, and mayve with colour effects.


Chapter four: Inter-referencing

4.1: Having fun with colours
4.2: Convolutions
4.3: Linear transformations

Chapter five: Using masks

5.1: Selection masks
5.2: Translation masks
5.3: Absolute masks