Unwanted duplicate points ggplotly animation

Plotly R, MATLAB, Julia, Net Plotly R
1

Lets say I have four groups of numbers (groups 1-4) that have some x (1-10), y (1-10) and time (1-10 coordinates. However, groups 1 and 2 stop existing after time-point 5.

pt <- data.frame(
  xx = rep(1:10, 10),
  yy = jitter(rep(1:10, 10)),
  tt = rep(1:10, each = 40),
  gg = rep(1:4, each = 10)
) %>%
  filter(!(gg %in% c(1,2) & tt > 5)) %>%
  mutate(yy = yy + gg)

I create an animated plot, color coding the points by the group gg and animating by value tt

pp <- pt %>% ggplot(aes(x = xx, y = yy, fill = as.factor(gg), frame = tt, group = gg)) +
  geom_point(shape = 21, colour = "black", size = 2) +
  scale_fill_viridis_d()

  ppp <- ggplotly(pp)
  
  ppp

This gives me a plot that behaves as expected for time points 1-5

timepoint5
timepoint5.png870×558 25 KB

However, when the plot advances to time-point 6 two related strange things happen.
(1) The points from groups 1 and 2 move into the position into the position of the points from groups 3 and 4.
(2) The points that were originally from groups 3 and 4 remain frozen in their final positions. I am left with something that looks like this. With two points per group, only one of which moves correctly and the other of which is frozen in the position from time point 5.

timepoint6
timepoint6.png877×561 21.9 KB

If, instead of treating gg as a factor fill = as.factor(gg) when I set up the plot, I instead leave it as a number fill = gg I get the afformetntioned behavior (1) but not behavior (2). That is the points shuffle positions in a way that I do not want, but at least the old groups don’t hang around confusing the viewer.

Any suggestions about how I can resolve these two plotting issues? Thanks.

2

I discovered that I can solve problem (2), with even with the factors, if I specify animation_opts(redraw = TRUE).

So now my call is

pp <- pt %>% ggplot(aes(x = xx, y = yy, fill = as.factor(gg), frame = tt, group = gg)) +
  geom_point(shape = 21, colour = "black", size = 2) +
  scale_fill_viridis_d()

  ppp <- ggplotly(pp) %>% animation_opts(redraw = TRUE)
  
  ppp

However, problem (1) where the points from groups 1 and 2 jump into the position of points from groups 3 and 4 (and then replace those points), rather than just disappearing persists. I’d still appreciate help on that front.