Plotting a difference curve involving the reference smooth I'm wondering if there is a way to plot a difference curve for two smooths involving the intercept.
Here is a summary of the model smooths.
s(percent)                 
  s(percent):x1x2voiced.D    
  s(percent):x1x2voiceless.f 
  s(percent):x1x2voiced.G    
  s(percent):x1x2voiceless.h 
  s(percent):x1x2voiceless.s 
  s(percent):x1x2voiceless.S 
  s(percent):x1x2voiceless.T 
  s(percent):x1x2voiceless.x 
  s(percent):x1x2voiceless.X-
  s(percent):x1x2voiced.z 

The following code (from itsadug) works fine.
plot_diff(m1, view='percent', comp=list(x1x2=c('voiced.D', 'voiceless.f'))) 

But I'm interested also in other contrasts such as s(percent) vs. s(percent):x1x2voiced.G. Is there a way to include the first smooth s(percent) in the comp call?
I'm aware of this code (from mgcViz) which works fine:
plotDiff(s1 = sm(b, 1), s2 = sm(b, 2)) + l_ciPoly() + l_fitLine() + 
    geom_hline(yintercept = 0, linetype = 2)

But I'd love something following the same approach as itsadug in order to integrate the output with other functions such as get_smooths_difference() from tidymv.
 A: I guess I figured it out following Gavin's comment.
Here is the answer for anybody else who might come later looking for the answer.
This code extracts the reference level for the ordered factors.
plot_diff(cog_ini__aa_m2, view='percent', comp=list(x1x2=c(cog_ini__aa_m2[["xlevels"]]$x1x2[1], "voiceless.X-")))

And here is how I fed it to tidymv for extra customizations.
get_smooths_difference(cog_ini__aa_m2, percent, list(x1x2=c(cog_ini__aa_m2[["xlevels"]]$x1x2[1], "voiceless.X-")))-> inter_diff
inter_diff %>% ggplot(aes(percent, difference, group = group)) + geom_hline(aes(yintercept = 0), colour = "#8f5f3f") + geom_ribbon(aes(ymin = CI_lower, ymax = CI_upper, fill = sig_diff), alpha = 0.3) + geom_line(aes(colour = sig_diff), size = 1) + scale_colour_manual(values = c("#e35760", "#6f849c")) + scale_fill_manual(values = c("#e35760", "#6f849c")) + labs(colour = "significant", fill = "significant") + theme(legend.position = "top")
This produces the plot I'm looking for.

Thanks Gavin for your time and help!
