Strange looking vertical skip in headline of outer scheme `miniframes`

[yizhaozhang11]

Issue

Consider the following simple test document using theme Dresden, which essentially uses miniframes for outer theme. When comparing the results compiled using beamer 3.36 vs 3.70, we can visually observe a difference between the headlines of the two versions. Personally speaking, there seems to be too much space for the 3.70, and the 3.36 one looks much better.

Cause

I'm not a professional in TeX, but I tried to investigate the issue myself and the following relevant code snippet:

https://github.com/josephwright/beamer/blob/c0d91f15165421646b5383546e6195187b7f97c9/base/themes/outer/beamerouterthememiniframes.sty#L103-L120

The snippet itself wasn't changed quite much throughout all the versions. However, after some experiments, I believe I'm able to say that it's the difference of \insertnavigation that causes the different appearances. More concretely:

• In 3.36, the \vskip2pt is directly on top of the text.
• In 3.70, the \vskip2pt is a \baselineskip above from the baseline of the text.

Cure

I have also tested some possible patches. I'm not completely sure how and why this works, but they look good:

• If we want to keep the appearance with the 2pt skip above, just replace the \vskip2pt with \vskip-\lineskip:

  \begin{beamercolorbox}{section in head/foot}
      % \vskip2pt              % before
      \vskip-\lineskip         % after
      \insertnavigation{\paperwidth}
      \vskip2pt
  \end{beamercolorbox}
  

  
• However, note that the subsection headline below actually uses slightly different measures (2.5ex above baseline, 1.125ex below), we can also accordingly modify the skip for the section title to match the look:

  \begin{beamercolorbox}{section in head/foot}
      % \vskip2pt              % before
      \vskip2.5ex              % after
      \vskip-\baselineskip     % after
      \insertnavigation{\paperwidth}
      % \vskip2pt              % before
      \vskip1.125ex            % after
  \end{beamercolorbox}
  

  

The difference between the two suggested changes is, indeed, really subtle. But since typesetting is an art in the details, we might still need to choose carefully between the two.

Acknowledgement

Thank you for your priceless effort in developing and maintaining this package.

[yizhaozhang11]

Friendly reminder for anyone to take a look UwU

[samcarter]

@TamotsuTokazu Sorry for the late reply.

If I remember correctly, there was some problem with overlapping elements in a derived theme and the current position of elements was a compromise to minimise problems. I need to find the problematic example again to test your suggestion.

[samcarter]

@TamotsuTokazu Thanks for opening this issue!