Bitmap

class Pixel { has UInt ($.R, $.G, $.B) }
class Bitmap {
    has UInt ($.width, $.height);
    has Pixel @!data;

    method fill(Pixel $p) {
        @!data = $p.clone xx ($!width*$!height)
    }
    method pixel(
    $i where ^$!width,
    $j where ^$!height
    --> Pixel
    ) is rw { @!data[$i + $j * $!width] }

    method set-pixel ($i, $j, Pixel $p) {
    self.pixel($i, $j) = $p.clone;
    }
    method get-pixel ($i, $j) returns Pixel {
    self.pixel($i, $j);
    }
}

my Bitmap $b = Bitmap.new( width => 10, height => 10);

$b.fill( Pixel.new( R => 0, G => 0, B => 200) );

$b.set-pixel( 7, 5, Pixel.new( R => 100, G => 200, B => 0) );

say $b.perl;

Thanks to the rw trait on the pixel method, we don't actually need to define two separate methods, set-pixel and get-pixel, but that is an explicit requirement of the task. (Beware your presuppositions! In Perl 6, accessors only determine identity, not use. In particular, identity is considered orthogonal to lvalue/rvalue context.)