Why Plots.jl produce figures looks not as good as other packages even with pyplot

My question is actually why the defaults of Plots.jl is not pretty. Despite personal preferences, who can say the defaults of Plots.jl is better than matplotlib ?

Comparing your two plots, is say about half of the differences are advantage Plots, yes. But nobody is forcing you to use it if you don’t like it.

What are you talking about? You say that the defaults of Plots.jl is not pretty is the advantage?

As said multiple times in this thread, some of those differences are subjective. I personally prefer the grid and plot edges on Plots.jl plot.

5 Likes

I really can’t see how can one answer to this. If it is not a matter of personal preference, how to tell if something looks better than something else …
As @nilshg said you can set your own defaults, so what is the point of arguing over other defaults ?

Personally, even though I use Plots with different backends for the day to day work, when it comes to publication, I will always go to pgfplots either directly or through the various julia backends.

4 Likes

I do prefer the gray box around the legend, rather than black, and it’s nice that the legend doesn’t overlap with the lines. Other than that, I prefer the Julia version: I prefer the larger font, the default grid lines, fewer tick marks, and the default non-boxed plot. It’s so easy to theme your plots, I really don’t know why you’re trying to push your tastes as a universal standard.

3 Likes

I just wondering why people just cannot understand that the point is why Plots.jl doens’t provide better default.

Saying it one last time: “better” is largely subjective, so it’s questionable whether there actually is an objective problem with the Plots default to begin with (apart from the concrete points mentioned above).

(Also what would that “better new default” be? What if someone doesn’t like it? What would we do if he/she opened a thread like you saying that the ggplot defaults are much “better”? That’s why people argue for customizability because it’s the only way to making people happy. But apparently not everyone.)

3 Likes

And, BTW, personally I’ve always used matplotlib style sheets to specify my own “matplotlib default” because I’m not a huge fan of their default either. Again, customizability is much more important in particular when talking about publication quality plots (which often take a long time to prepare anyways).

3 Likes

Alright, this is going in circles and doesn’t seem to be productive or positive at this point. I’ve split out one of the actionable improvements into a new topic.

Topics about difficulties and potential improvements are always welcome, but for something as subjective as this, everyone needs to be participating in the discussion with a willingness to see the subjectivity of their own preferences.

10 Likes