RGBgradient

Category:	HTML Utilities
Author/Contact Info	Jeff `japhy` Pinyan
Description:	Wrote this for work. Creates an iterator to return the next RGB code in a gradient.
# usage: # $iter = range(\@fromRGB, \@toRGB, \@weights, $size); # my $iter = range( # [0,0,0] => [255,255,255], # [ (1) x 100 ], # 255 # ); # # while (my ($rgb_ref, $width) = $iter->()) { ... } sub range { my ($from,$to,$w,$s) = @_; my $i = 0; my @steps = map +($to->[$_] - $from->[$_])/$#$w, 0..2; my $c = $from; my $f = 0; $i += $_ for @$w; # this (CRUFTILY) converts an array of floats # with a sum of X to an array of ints that still # has that same sum of X $w = [ map { my $v = $f + $s*$_/$i; $f = $v - int($v); int $v } @$w ]; $w->[-1] += $f; sub {{ my $ret = [ @$c ]; return if !@$w; $c = [ map $steps[$_] + $c->[$_], 0..2 ]; return ($ret, shift @$w) if $w->[0]; shift @$w, redo; }}; }

Comment on RGBgradient Download Code

Replies are listed 'Best First'.

Re: RGBgradient
by grinder (Bishop) on Mar 15, 2001 at 14:40 UTC

I think you want to

  sub {{
    my $ret = [ map {int} @$c ];
    ...
[download]

to stop returning floating point back to the caller. Makes for funny HTML otherwise.

Update:

  $w = [ map {
       my $v = $f + $s*$_/$i;
       $f = $v - int($v);
       $v # not int $v
     } @$w ];
     $w->[-1] += $f;
[download]

The problem is that it works fine for linear gradients, but when you use non-linear gradients the round-off adds up and your iterator poops out before it gets to the end.

Why would you want to use a non-linear gradient? Because the (in)?famous Netscape Color Cube(TM) is a terrible thing: it lends to much emphasis to highly saturated colors, and as a result there are too few colors to represent the many subtle hues the eye is capable of distinguishing.

So what I attempted to do was to create a sinusoidal curve, peaking at 1 in the middle of the gradient and dropping to 0 at the ends. That's where I uncovered the int problem. Once I had my weight array the way I wanted it, I plugged it into the iterator, but for some reason that escapes me, I'm not seeing the behaviour I expect. The numbers should climb away quickly from first value in the interval move slowly around the midpoint, and then drop quickly down to the end value.

It may be that I have completely misunderstood what the weights' purpose. Whatever, here's the code. No, I didn't use CGI, this is a quick hack, not production code.

use constant PI => 4 * atan2(1,1);

my $intervals = 40;
my $iterlin = range( [  0,255,255] => [  0,255,  0],
    [(1) x $intervals],
    $intervals );
my $itersin = range( [  0,255,255] => [  0,255,  0],
    [ map {sin(($_/$intervals)*PI) } 0..$intervals],
    $intervals );

print qq{<html><head><title>Gradient</title></head>
<body bgcolor="#ffffff"><table>\n};

while (my ($rgb_lin, $width) = $iterlin->()) {
    my ($rgb_sin, $widthsin) = $itersin->();

    my $lin = join '', '#',
        map { sprintf '%02x',$_ } @$rgb_lin;
    my $sin = join '', '#',
        map { sprintf '%02x',$_ } @$rgb_sin;

    print qq{<tr>
<td bgcolor="$lin"><tt>$$rgb_lin[0]</tt></td>
<td bgcolor="$lin"><tt>$$rgb_lin[1]</tt></td>
<td bgcolor="$lin"><tt>$$rgb_lin[2]</tt></td>
<td bgcolor="$sin"><tt>$$rgb_sin[0]</tt></td>
<td bgcolor="$sin"><tt>$$rgb_sin[1]</tt></td>
<td bgcolor="$sin"><tt>$$rgb_sin[2]</tt></td>
</tr>};
}
print "</table></body></html>\n";
[download]

[reply]
[d/l]
[select]


Problems? Is your data what you think it is?
	PerlMonks