Skip to content

Instantly share code, notes, and snippets.

@kriegsman
Last active January 29, 2021 17:48
Show Gist options
  • Save kriegsman/5408ecd397744ba0393e to your computer and use it in GitHub Desktop.
Save kriegsman/5408ecd397744ba0393e to your computer and use it in GitHub Desktop.
Twinkling 'holiday' lights that fade in and out.
#include "FastLED.h"
#define LED_PIN 3
#define LED_TYPE WS2811
#define COLOR_ORDER RGB
#define NUM_LEDS 50
CRGB leds[NUM_LEDS];
// Twinkling 'holiday' lights that fade up and down in brightness.
// Colors are chosen from a palette; a few palettes are provided.
//
// The basic operation is that all pixels stay black until they
// are 'seeded' with a relatively dim color. The dim colors
// are repeatedly brightened until they reach full brightness, then
// are darkened repeatedly until they are fully black again.
//
// A set of 'directionFlags' is used to track whether a given
// pixel is presently brightening up or darkening down.
//
// For illustration purposes, two implementations of directionFlags
// are provided: a simple one-byte-per-pixel flag, and a more
// complicated, more compact one-BIT-per-pixel flag.
//
// Darkening colors accurately is relatively easy: scale down the
// existing color channel values. Brightening colors is a bit more
// error prone, as there's some loss of precision. If your colors
// aren't coming our 'right' at full brightness, try increasing the
// STARTING_BRIGHTNESS value.
//
// -Mark Kriegsman, December 2014
#define MASTER_BRIGHTNESS 96
#define STARTING_BRIGHTNESS 64
#define FADE_IN_SPEED 32
#define FADE_OUT_SPEED 20
#define DENSITY 255
void setup() {
delay(3000);
FastLED.addLeds<LED_TYPE,LED_PIN,COLOR_ORDER>(leds, NUM_LEDS).setCorrection(TypicalLEDStrip);
FastLED.setBrightness(MASTER_BRIGHTNESS);
}
void loop()
{
chooseColorPalette();
colortwinkles();
FastLED.show();
FastLED.delay(20);
}
CRGBPalette16 gPalette;
void chooseColorPalette()
{
uint8_t numberOfPalettes = 5;
uint8_t secondsPerPalette = 10;
uint8_t whichPalette = (millis()/(1000*secondsPerPalette)) % numberOfPalettes;
CRGB r(CRGB::Red), b(CRGB::Blue), w(85,85,85), g(CRGB::Green), W(CRGB::White), l(0xE1A024);
switch( whichPalette) {
case 0: // Red, Green, and White
gPalette = CRGBPalette16( r,r,r,r, r,r,r,r, g,g,g,g, w,w,w,w );
break;
case 1: // Blue and White
//gPalette = CRGBPalette16( b,b,b,b, b,b,b,b, w,w,w,w, w,w,w,w );
gPalette = CloudColors_p; // Blues and whites!
break;
case 2: // Rainbow of colors
gPalette = RainbowColors_p;
break;
case 3: // Incandescent "fairy lights"
gPalette = CRGBPalette16( l,l,l,l, l,l,l,l, l,l,l,l, l,l,l,l );
break;
case 4: // Snow
gPalette = CRGBPalette16( W,W,W,W, w,w,w,w, w,w,w,w, w,w,w,w );
break;
}
}
enum { GETTING_DARKER = 0, GETTING_BRIGHTER = 1 };
void colortwinkles()
{
// Make each pixel brighter or darker, depending on
// its 'direction' flag.
brightenOrDarkenEachPixel( FADE_IN_SPEED, FADE_OUT_SPEED);
// Now consider adding a new random twinkle
if( random8() < DENSITY ) {
int pos = random16(NUM_LEDS);
if( !leds[pos]) {
leds[pos] = ColorFromPalette( gPalette, random8(), STARTING_BRIGHTNESS, NOBLEND);
setPixelDirection(pos, GETTING_BRIGHTER);
}
}
}
void brightenOrDarkenEachPixel( fract8 fadeUpAmount, fract8 fadeDownAmount)
{
for( uint16_t i = 0; i < NUM_LEDS; i++) {
if( getPixelDirection(i) == GETTING_DARKER) {
// This pixel is getting darker
leds[i] = makeDarker( leds[i], fadeDownAmount);
} else {
// This pixel is getting brighter
leds[i] = makeBrighter( leds[i], fadeUpAmount);
// now check to see if we've maxxed out the brightness
if( leds[i].r == 255 || leds[i].g == 255 || leds[i].b == 255) {
// if so, turn around and start getting darker
setPixelDirection(i, GETTING_DARKER);
}
}
}
}
CRGB makeBrighter( const CRGB& color, fract8 howMuchBrighter)
{
CRGB incrementalColor = color;
incrementalColor.nscale8( howMuchBrighter);
return color + incrementalColor;
}
CRGB makeDarker( const CRGB& color, fract8 howMuchDarker)
{
CRGB newcolor = color;
newcolor.nscale8( 255 - howMuchDarker);
return newcolor;
}
// For illustration purposes, there are two separate implementations
// provided here for the array of 'directionFlags':
// - a simple one, which uses one byte (8 bits) of RAM for each pixel, and
// - a compact one, which uses just one BIT of RAM for each pixel.
// Set this to 1 or 8 to select which implementation
// of directionFlags is used. 1=more compact, 8=simpler.
#define BITS_PER_DIRECTION_FLAG 1
#if BITS_PER_DIRECTION_FLAG == 8
// Simple implementation of the directionFlags array,
// which takes up one byte (eight bits) per pixel.
uint8_t directionFlags[NUM_LEDS];
bool getPixelDirection( uint16_t i) {
return directionFlags[i];
}
void setPixelDirection( uint16_t i, bool dir) {
directionFlags[i] = dir;
}
#endif
#if BITS_PER_DIRECTION_FLAG == 1
// Compact (but more complicated) implementation of
// the directionFlags array, using just one BIT of RAM
// per pixel. This requires a bunch of bit wrangling,
// but conserves precious RAM. The cost is a few
// cycles and about 100 bytes of flash program memory.
uint8_t directionFlags[ (NUM_LEDS+7) / 8];
bool getPixelDirection( uint16_t i) {
uint16_t index = i / 8;
uint8_t bitNum = i & 0x07;
// using Arduino 'bitRead' function; expanded code below
return bitRead( directionFlags[index], bitNum);
// uint8_t andMask = 1 << bitNum;
// return (directionFlags[index] & andMask) != 0;
}
void setPixelDirection( uint16_t i, bool dir) {
uint16_t index = i / 8;
uint8_t bitNum = i & 0x07;
// using Arduino 'bitWrite' function; expanded code below
bitWrite( directionFlags[index], bitNum, dir);
// uint8_t orMask = 1 << bitNum;
// uint8_t andMask = 255 - orMask;
// uint8_t value = directionFlags[index] & andMask;
// if( dir ) {
// value += orMask;
// }
// directionFlags[index] = value;
}
#endif
@embedded-creations
Copy link

Hi Mark, there's a bug in this sketch if used with certain palettes, pixels will get "stuck" with state GETTING_BRIGHTER, until all pixels are stuck. Forked sketch with a palette showing the bug, and a potential fix here: https://gist.github.com/embedded-creations/00dfedaad7b94cb603ce5aabe1a90ee9

Credit for the fix goes to WLED (or WS2812FX) developers, found here (mode_colortwinkle()): https://github.com/Aircoookie/WLED/blob/master/wled00/FX.cpp#L2012

I'm not sure this fix is valid for all colors, consider the case where there's a color with e.g. a dim red channel, and even dimmer green channel, and the fix could double color until nscale8 is able to scale the red channel, but the green channel will still be stuck below the threshold where nscale8() would return a non-zero value.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment