Eliminate floating-point logic

Thanks to @noseglasses for the guidance!

Signed-off-by: Bart Nagel <bart@tremby.net>

src/kaleidoscope/plugin/LED-Stalker.cpp

@@ -111,14 +111,18 @@ uint8_t hue_end = 0;
 cRGB BlazingTrail::compute(uint8_t *step) {
   cRGB color;
 
-  // Get a float from 0 to 1 for our position in the animation
-  float pos = (0xff - *step) / (float) 0xff;
+  // Get the position in the animation, where 0 is start and 0xff is end
+  uint8_t pos255 = 0xff - *step;
 
   // Fade hue linearly from orange to red
-  uint8_t hue = hue_start + pos * (hue_end - hue_start);
-
-  // Fade value from full following a -x^4+1 curve
-  uint8_t val = 0xff * (1 - pos * pos * pos * pos);
+  uint8_t hue = hue_start + (pos255 * (hue_end - hue_start) >> 8);
+
+  // Fade value from full following a 1-x^4 curve
+  uint8_t val =
+    0xff // Maximum brightness
+    - ((uint32_t) pos255 * pos255 * pos255 * pos255 // Animation position to 4th power
+       >> 24) // ...pulled down to 8-bit range (but this has a maximum of 0xfc rather than 0xff)
+    - pos255 / (0x100 / 4); // Correction to bring the end result into a full 0 to 0xff range
 
   color = hsvToRgb(hue, 0xff, val);