Beamer vs. TikZ: uncover nodes step-by-step

Question

I'm trying to uncover a TikZ image piece-by-piece like so:

\documentclass{beamer}

\usepackage{tikz}
\usetikzlibrary{positioning} 

\begin{document}
\begin{frame}{Linearity}
\begin{center}
\begin{tikzpicture}
  \node (Ctx) {\visible<2->{$\Gamma$}};
  \node[below right=0.5cm and 0.1cm of Ctx] (E) {$E$};
  \node[below right=0.5cm and 0.1cm of E] (psi1) {\visible<3->{$\Psi_1$}};
\end{tikzpicture}
\end{center}
\end{frame}
\end{document}

However, this gives the following error message:

! Package tikz Error: Giving up on this path. Did you forget a semicolon?.

See the tikz package documentation for explanation.
Type  H <return>  for immediate help.
 ...                                              

l.16 \end{frame}

If I use \only instead of \visible, pdflatex runs OK but the output jumps around when going from slide to slide, since the invisible nodes don't take up space and thus the E node's position changes. So \visible is what I need.

Answer 1

In the one case I overlayed TikZ I was using \uncover<n-m>{ELEMENT} in order to show ELEMENT in slides n to m.

Answer 2

It works if you place the \visible command and its arguments into a { } group:

\documentclass{beamer}
\usepackage{tikz}
\usetikzlibrary{positioning} 
\begin{document}
\begin{frame}{Linearity}
\begin{center}
\begin{tikzpicture}
  \node (Ctx) {{\visible<2->{$\Gamma$}}};
  \node[below right=0.5cm and 0.1cm of Ctx] (E) {$E$};
  \node[below right=0.5cm and 0.1cm of E] (psi1) {{\visible<3->{$\Psi_1$}}};
\end{tikzpicture}
\end{center}
\end{frame}
\end{document}

It seems that the scanning code from \visible removes the ; or otherwise interferes with the \node code. Note that \node doesn't read its content as argument but as box content to allow verbatim inside. The \visible macro might do the same and therefore such errors can happen.

Answer 3

Here is another solution using \setbeamercovered{invisible} and \pause

\documentclass{beamer}
\usepackage{tikz}
\usetikzlibrary{positioning} 
\begin{document}
\begin{frame}{Linearity}
    \setbeamercovered{invisible}
\begin{center}
\begin{tikzpicture}
    \node(Ctx){$E$};\pause
    \node[above left=0.5cm and 0.1cm of Ctx] (gamma) {$\Gamma$};\pause  
    \node[below right=0.5cm and 0.1cm of Ctx] (psi1){$\Psi_1$};\pause
\end{tikzpicture}
\end{center}
\end{frame}
\end{document}

This can also be contained in \columns where one can have tables, minipages or other nodes to explain what is going on in a diagram using \onslide<n->. I use this frequently to simultaneously expose a geometric diagram and a two-column proof. Hope it is of interest.