Van Gogh’s Life through his Favorite Colors

I set out to understand the life of a man through the colors he chose to use. Vincent van Gogh was an amazing painter who changed the art world more then he will ever know.

What would life be if we had no courage to attempt anything? – Vincent van Gogh

As Vincent declared I wanted to have the courage to keep exploring data and discover the meaning behind it. This analysis was mostly used to hone my skills and better understand the capabilities of R. As I started to explore the packages that can interoperate and manipulate color, I was curious about the colors used by famous painters. I started with Van Gogh. Luckily, I was able to find a large collection of his painting on wikipedia. From there I scraped the images and then ran them through a package that pulled the most prominent color from each painting and then I plotted it over a timeline.

Van Gogh painted 900+ painting, each line represents the most prominent color of that painting!

The code that I used to scrape the images and extract the color is below:

#loading up all of the required library
library("rvest")
library("magrittr")
## install Jo-Fai Chow's package
require("devtools")
install_github("ramnathv/rblocks")
install_github("woobe/rPlotter")
library("rPlotter")
library("stringr")

#pulling tables from wikipedia that contains the images from Van Gogh's paintings
van_table <- read_html("https://en.wikipedia.org/wiki/List_of_works_by_Vincent_van_Gogh")
img_link <- html_nodes(van_table, css = "table") %>% html_nodes("td:nth-child(2) a") %>% html_attr("href")
van_files <- img_link[1:912] 
van_text_split <- gsub(pattern = "/wiki/", replacement = "", van_files) %>% data.frame()
van_data <- html_nodes(van_table, xpath = '//*[@id="mw-content-text"]/table[1]') %>% html_table() %>% data.frame()
van_data <- van_data[-c(370), ]
van_data <- cbind(van_data, van_text_split)
colnames(van_data) <- c("num", "img", "title", "year", "location", "img_url")

#does a second call to wikipedia to pull the links to the images
pic_list <- data.frame()
for (i in 1:length(as.character(van_data$img_url))) {
  tryCatch({
    img_container <- read_html(paste("https://commons.wikimedia.org/wiki/", as.character(van_data$img_url[[i]]), sep = ""))
    link <- html_nodes(img_container, "#file > a") %>% html_attr("href")
    pic_list <- rbind(pic_list, data.frame(i, as.character(van_data$img_url[[i]]), link))
    print(i)
  }, error=function(e){
    print("error")
    })
}

#deletes images that are corrupt and then combines that column from the pictures
van_data <- van_data[-c(592, 806, 808), ]
van_data <- cbind(van_data, pic_list)

#writes a files that contains the data
write.csv(van_data, "color_extract/van_davs.csv")

#downloads all the photos to local machine in order to speed up the color extraction
sapply(1:length(as.character(van_data$link)), 
   function(x) download.file(url=as.character(van_data$link[x]), 
   destfile=paste("color_extract/van_paintings/", 
   sprintf("%06d", x),sep="", ".jpg"), mode ="wb", quiet = T))

#re-reads the data about the images and also converts the factors to character
van_data <- read.csv("color_extract/van_davs.csv", header = TRUE, stringsAsFactors = TRUE)

#extracts the most prominent color from each image and puts into a data.frame
van_colors <- data.frame()
for (i in 1:length(as.character(van_data$img_url))) {
  van_color <- extract_colours(paste("color_extract/van_paintings/", sprintf("%06d", i), ".jpg", sep=""), 2)
  van_colors <- rbind(van_colors, t(data.frame(van_color)))
  print(i)
}

#combines the orginal van gogh data with the extracted color data
van_con_color <- cbind(van_data, van_colors)
write.csv(van_con_color, "color_extract/van_con_color.csv")

#extracts the the color and years of the painting
simple_van <- van_con_color[ ,c(4, 7,10)]

#transforms the year column into useable data
grep("\n\n\n\n", as.character(simple_van$year[2]))
year <- lapply(as.character(simple_van$year), function(x) str_extract(x, "18[8-9][0-9]"))

#appears the modified year to a new data.frame
simple_van_mod <- simple_van
simple_van_mod["year"] <- data.frame(t(data.frame(year)))

#creates the barplots of color
barplot(rep(1, 909), 
        col=as.character(simple_van_mod$V1), 
        bty="n", 
        xaxt="n", 
        yaxt="n", 
        xlab="year", 
        ylab="",
        border= NA,
        space = c(0,1)
        )

#attches the the years to the bottom of the plot and addes a title
axis(1, at=(1:length(1:1000))[1:1200 %in% seq(1,1000,91)], labels=c(1881:1891), las=2, lwd=0, lwd.tick = 1)
title("Vincent van Gogh's Life through Color")