gganimate动画标签补间值异常及显示控制技术咨询

使用gganimate制作动态条形图时，条形标签的补间值显示为乱码（十六进制），原始代码如下：

library(ggplot2)
library(dplyr)
library(gganimate)

set.seed(123)

df <- data.frame(
  period = rep(1:3, each = 5),
  x = rep(LETTERS[1:5], 3),
  p = rbeta(15, 3, 2)
)

data <- df %>% 
  group_by(period) %>% 
  mutate(rank = min_rank(-p)) %>% 
  ungroup()

ggplot(data, aes(x = rank)) + 
  geom_tile(aes(y = p / 2, 
                height = p,
                width = 0.75)) + 
  geom_text(aes(y = p,
                label = scales::percent(p, 1)),
            hjust = 0,
            nudge_y = 0.005) +
  coord_flip(clip = "off", expand = F) + 
  scale_x_reverse() + 
  labs(title = 'Period {closest_state}',
       x = "",
       y = "") + 
  theme(axis.text = element_blank(),
        axis.ticks = element_blank(),
        plot.margin = margin(r = 5, unit = "cm")) + 
  transition_states(period) + 
  ease_aes('linear')

已尝试的解决方法：

• 使用label = round(p * 100)；
• 在df中提前创建字符型标签列df$labs <- paste0(round(df$p * 100), "%")并应用于label参数。

问题1：修正补间值，正确显示百分比格式 #

问题根源：当在aes()内直接生成字符串标签时，gganimate会将label视为需要补间的美学属性，对字符串进行插值运算，导致出现乱码。正确做法是基于补间后的数值动态计算标签，而非提前生成静态字符串。

方案1：使用after_stat()实时计算标签 #

通过after_stat()包裹百分比格式化逻辑，确保标签基于补间后的p值实时生成：

ggplot(data, aes(x = rank)) + 
  geom_tile(aes(y = p / 2, 
                height = p,
                width = 0.75)) + 
  # 关键修改：用after_stat根据补间后的y值（即p）生成百分比标签
  geom_text(aes(y = p),
            label = after_stat(scales::percent(y, 1)),
            hjust = 0,
            nudge_y = 0.005) +
  coord_flip(clip = "off", expand = F) + 
  scale_x_reverse() + 
  labs(title = 'Period {closest_state}',
       x = "",
       y = "") + 
  theme(axis.text = element_blank(),
        axis.ticks = element_blank(),
        plot.margin = margin(r = 5, unit = "cm")) + 
  transition_states(period) + 
  ease_aes('linear')

方案2：使用stage()分离静态与动态逻辑 #

用stage()指定label仅在补间完成后计算，避免对字符串插值：

ggplot(data, aes(x = rank)) + 
  geom_tile(aes(y = p / 2, 
                height = p,
                width = 0.75)) + 
  geom_text(aes(y = p,
                label = stage(p, after = scales::percent(.x, 1))),
            hjust = 0,
            nudge_y = 0.005) +
  coord_flip(clip = "off", expand = F) + 
  scale_x_reverse() + 
  labs(title = 'Period {closest_state}',
       x = "",
       y = "") + 
  theme(axis.text = element_blank(),
        axis.ticks = element_blank(),
        plot.margin = margin(r = 5, unit = "cm")) + 
  transition_states(period) + 
  ease_aes('linear')

问题2：仅在目标状态显示标签，隐藏补间过程的值 #

可以通过两种方式实现，核心是控制标签在过渡阶段的显隐状态：

方案1：基于closest_state判断显隐 #

在geom_text的alpha参数中，通过closest_state判断当前是否处于目标状态，仅在状态帧显示标签：

ggplot(data, aes(x = rank)) + 
  geom_tile(aes(y = p / 2, 
                height = p,
                width = 0.75)) + 
  geom_text(aes(y = p,
                label = after_stat(scales::percent(y, 1)),
                # 仅当当前帧对应目标状态时显示标签
                alpha = if_else(closest_state == period, 1, 0)),
            hjust = 0,
            nudge_y = 0.005) +
  coord_flip(clip = "off", expand = F) + 
  scale_x_reverse() + 
  scale_alpha_identity() + # 保持alpha值不被自动缩放
  labs(title = 'Period {closest_state}',
       x = "",
       y = "") + 
  theme(axis.text = element_blank(),
        axis.ticks = element_blank(),
        plot.margin = margin(r = 5, unit = "cm")) + 
  transition_states(period, 
                    transition_length = 1, # 过渡动画时长
                    state_length = 1) + # 目标状态停留时长
  ease_aes('linear')

方案2：使用enter_fade()与exit_fade()控制标签淡入淡出 #

让标签仅在进入目标状态时淡入，离开时淡出，过渡过程中隐藏：

ggplot(data, aes(x = rank)) + 
  geom_tile(aes(y = p / 2, 
                height = p,
                width = 0.75)) + 
  geom_text(aes(y = p,
                label = after_stat(scales::percent(y, 1))),
            hjust = 0,
            nudge_y = 0.005) +
  coord_flip(clip = "off", expand = F) + 
  scale_x_reverse() + 
  labs(title = 'Period {closest_state}',
       x = "",
       y = "") + 
  theme(axis.text = element_blank(),
        axis.ticks = element_blank(),
        plot.margin = margin(r = 5, unit = "cm")) + 
  transition_states(period, 
                    transition_length = 1,
                    state_length = 1) + 
  ease_aes('linear') +
  enter_fade() + # 进入目标状态时淡入标签
  exit_fade() # 离开目标状态时淡出标签

内容的提问来源于stack exchange，提问作者LMc