Export RGB image from google earth engine's landsat8 collection using GEE's script editor

export_rgb.js

var year_string = '2019'

// Bounding box for imagery. The smaller it is, the faster to process
var bounding_box = ee.Geometry.Rectangle([-16.6147, 15.8444, -16.4147, 16.2444]);

var use_landsat_7 = true;

// Landsat7 imagery comes with "striping" that's an artifact of a hardware failure
// Compositing helps a little, but is not a fix. 
// More here: https://blog.mapbox.com/debanding-the-world-94439af16e7f 

if (use_landsat_7) {
	// LANDSAT7: 1999-2021
	var landsat_mission_string = 'LANDSAT/LE07/C02/T1'
} else {
	//LANDSAT8: 2013 forwards
	var landsat_mission_string = 'LANDSAT/LC08/C02/T1_RT'
}

// Many more choices here: https://developers.google.com/earth-engine/datasets/catalog

// Use dry-ish season as compositing window
var start_date = year_string + '-01-01'
var end_date = year_string + '-06-01'

var filtered_shp = ee.ImageCollection(landsat_mission_string).filterDate(start_date, end_date)
  .filterBounds(bounding_box)

// https://developers.google.com/earth-engine/apidocs/ee-algorithms-landsat-simplecomposite
var composite = ee.Algorithms.Landsat.simpleComposite(filtered_shp).float();

var rgb = composite.select('B4', 'B3', 'B2').unitScale(0,255)
var gray = composite.select('B8').unitScale(0,155)
var huesat = rgb.rgbToHsv().select('hue','saturation')
var satellite_imagery = ee.Image.cat(huesat, gray).hsvToRgb()


// The band color names are text at this point
// The maxes are a human-generated heuristic
// Other prior art uses slighly lower values
// Sensor drift may mean this should change slightly year to year
// but it is uncorrected here. 
var viz_params = {
  bands: ['red', 'green', 'blue'],
  min: [.02, .04, .06],
  max: [.4, .4, .4],
  gamma: 1.2
}

Map.addLayer(satellite_imagery, viz_params)
Map.setCenter(-16.5147, 15.8444, 13)

var landsat_shorthand = 'ls8';
if (use_landsat_7) {
	landsat_shorthand = 'ls7';
}

Export.image.toDrive({
  image: satellite_imagery.visualize(viz_params),
  description: 'stlouis_' + landsat_shorthand + '_' + year_string,
  scale: 30,
  region: bounding_box.bounds(),
  maxPixels:1e11,
  skipEmptyTiles: true
});

This requires a login to Google Earth Engine and a variety of google cloud permissions. Imagery is actually saved to google drive.