a minimal example of an RGB colorcoded ternary diagram

rgb-ternary.R

# a function from: Schöley, J., & Willekens, F. (2017). Visualizing compositional data on the Lexis surface. Demographic Research, 36(21), 627–658. https://doi.org/10.4054/DemRes.2017.36.21
TernaryCentroidCoord <- function (k) {
        centroids <- matrix(nrow = k^2, ncol = 5,
                            dimnames = list(NULL, c("j", "i", "p1", "p2", "p3")))
        for (j in 1:k) {
                for (i in 1:(2*k - 2*j + 1)) {
                        p1 = (6*k - 6*j - 3*i + 4 + i%%2) / (6*k)
                        p2 = (6*j - 2 - 2*i%%2) / (6*k)
                        p3 = (3*i - 2 + i%%2) / (6*k)
                        centroids[i+(j-1)*(2*k-j+1),] = c(j, i, p1, p2, p3)
                }
        }
        return(centroids)
}


df_leg <- TernaryCentroidCoord(100) %>% 
        as_tibble() %>% 
        mutate(color = rgb(p1, p2, p3))


df_leg %>% 
        ggplot() +
        geom_point(aes(p1, p2, p3,
                       group = 1,
                       color = color),
                   size = 3) +
        # scale
        scale_color_identity()+
        scale_L_continuous(breaks = seq(0, 1, .2)) +
        scale_T_continuous(breaks = seq(0, 1, .2)) +
        scale_R_continuous(breaks = seq(0, 1, .2)) +
        # coord
        coord_tern() +
        # theme
        theme()+
        theme_classic() %+replace%
        theme(legend.position = "none",
              plot.background = element_rect(fill = "transparent", colour = NA)
        )

gg_tern_legend <- last_plot()