ggplot geom_tile is distorted in ggplotly-CodePudding

The

With this, one possible workaround (outside the plotly package) would be to manually ensure there are (NA) data available across the whole x/y range, so expand.grid generates a sufficiently dense grid when the plot is translated to plotly.

set.seed(1) X1 <- data.frame(xcoord = c(sample(1:10, n, replace = TRUE), 1:10), ycoord = c(sample(1:10, n, replace = TRUE), 1:10), value = c(runif(n), rep(NA, 10))) gg1 <- ggplot(X1) geom_tile(aes(x = xcoord, y = ycoord, fill = value)) ggplotly(gg1)

Update

While the example above demonstrates it is sufficient to have a single value for any x and y in the dataset, I'll also add an arguably cleaner solution as suggested by Waldi in the comments. By (automatically) generating the full grid in advance, there is less reliance on quirks from the plotly translation. For grid spacing different than 1, the sequence of course needs to be adjusted.

# X: original dataframe as defined in the question X2 <- tidyr::expand_grid( xcoord = seq(min(X$xcoord), max(X$xcoord)), ycoord = seq(min(X$ycoord),max(X$ycoord)) ) %>% dplyr::left_join(X, by=c('xcoord','ycoord')) gg2 <- ggplot(X2) geom_tile(aes(x = xcoord, y = ycoord, fill = value)) ggplotly(gg2)