arrowsgraphicsmarginsspacingtikz-pgf
I want to create a simple figure where images are connected with arrows.
Here is what I did:
\begin{tikzpicture}
\node[](A) {\includegraphics[width=0.2\textwidth]{a}};
\node[right=1mm of A](B) {\includegraphics[width=0.2\textwidth]{b}};
\draw[->] (A) to (B);
\end{tikzpicture}
But I want the arrow to start and end right at the images with no space between. How can I do that?
If I understand correctly what you want to do then you will need to define new arrow shapes to get the outlines. The fact that these arrows are filled is hard-wired in to their definitions. Fortunately, that's not too difficult to do. The code for these arrows is in the file pgflibraryarrows.code.tex and is:
\pgfarrowsdeclare{triangle 90 cap}{triangle 90 cap}
{\pgfarrowsleftextend{+-.1\pgflinewidth}\pgfarrowsrightextend{+\pgflinewidth}}
{
\pgfpathmoveto{\pgfqpoint{-.1\pgflinewidth}{0.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{.5\pgflinewidth}{.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{\pgflinewidth}{0pt}}
\pgfpathlineto{\pgfqpoint{.5\pgflinewidth}{-.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{-.1\pgflinewidth}{-0.5\pgflinewidth}}
\pgfusepathqfill
}
There we can see the final command is \pgfusepathqfill which fills the path just constructed. To get it drawn rather than filled, we change that to \pgfusepathqstroke. Of course, to avoid name clashes, we change the name of our arrow as well (and put this in the preamble rather than editing pgflibraryarrows.code.tex.
\pgfarrowsdeclare{triangle 90 cap outline}{triangle 90 cap outline}
{\pgfarrowsleftextend{+-.1\pgflinewidth}\pgfarrowsrightextend{+\pgflinewidth}}
{
\pgfpathmoveto{\pgfqpoint{-.1\pgflinewidth}{0.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{.5\pgflinewidth}{.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{\pgflinewidth}{0pt}}
\pgfpathlineto{\pgfqpoint{.5\pgflinewidth}{-.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{-.1\pgflinewidth}{-0.5\pgflinewidth}}
\pgfusepathqstroke
}
But even this doesn't work as you want since you presumably want only some of the edges drawn and although one edge is left off (since a fill implicitly closes the path) it isn't the right one. Moreover, the line width is inherited from the main path which definitely isn't right as you want it to be the width of the "left over" piece from the doubled stroke. So we have to mess around a bit with the order in which pieces of the arrow are drawn. Moreover, to be fully flexible we should ensure that the line width is set correctly (incidentally, your two declarations of line width are unnecessary; the second trumps the first).
Summing all that up, the following adapts the arrow heads to provide an outline. It's in a \makeatletter ... \makeatother chunk because I use a couple of temporary storage slots that have @s in their names.
Here's what I got (your original on the left and my variant on the right):
Here's the code (the arrows library is only needed for your example):
\documentclass{standalone}
\usepackage{tikz}
\usetikzlibrary{arrows}
\makeatletter
\pgfarrowsdeclare{triangle 90 cap outline}{triangle 90 cap outline}
{\pgfarrowsleftextend{+-.1\pgflinewidth}\pgfarrowsrightextend{+\pgflinewidth}}
{
\pgfmathsetlength{\pgfutil@tempdima}{.5 * \pgflinewidth + .5 * \pgfinnerlinewidth}
\pgfsetlinewidth{\pgfutil@tempdima}
\pgfpathmoveto{\pgfqpoint{-.1\pgflinewidth}{.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{.5\pgflinewidth}{0.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{1\pgflinewidth}{0\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{.5\pgflinewidth}{-.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{-.1\pgflinewidth}{-.5\pgflinewidth}}
\pgfmathsetlength{\pgfutil@tempdimb}{\pgflinewidth - \pgfinnerlinewidth}
\pgfsetlinewidth{\pgfutil@tempdimb}
\pgfusepathqstroke
}
\pgfarrowsdeclare{triangle 90 cap reversed outline}{triangle 90 cap reversed outline}
{\pgfarrowsleftextend{+-.1\pgflinewidth}\pgfarrowsrightextend{+\pgflinewidth}}
{
\pgfmathsetlength{\pgfutil@tempdima}{.5 * \pgflinewidth + .5 * \pgfinnerlinewidth}
\pgfsetlinewidth{\pgfutil@tempdima}
\pgfpathmoveto{\pgfqpoint{-.1\pgflinewidth}{.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{1\pgflinewidth}{0.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{0.5\pgflinewidth}{0\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{1\pgflinewidth}{-.5\pgflinewidth}}
\pgfpathlineto{\pgfqpoint{-.1\pgflinewidth}{-.5\pgflinewidth}}
\pgfmathsetlength{\pgfutil@tempdimb}{\pgflinewidth - \pgfinnerlinewidth}
\pgfsetlinewidth{\pgfutil@tempdimb}
\pgfusepathqstroke
}
\makeatother
\begin{document}
\begin{tikzpicture}
\draw[
rounded corners=2ex,
double distance between line centers=2ex,
double,
line width=1pt,
triangle 90 cap reversed-triangle 90 cap
] (0,0) -| (1,1);
\draw[
rounded corners=2ex,
double distance between line centers=2ex,
double,
line width=1pt,
triangle 90 cap reversed outline-triangle 90 cap outline
] (2,0) -| ++(1,1);
\end{tikzpicture}
\end{document}
Problems 1) and 3): you can control the alignment using the mid anchor of nodes. Problem 2): you can use the optional argument of the line changing command \\; here's a modified version of your code, feel free to change the lengths according to your needs:
\documentclass{article}
\usepackage{tikz}
\usetikzlibrary{shapes.multipart,positioning,fit}
\begin{document}
\begin{tikzpicture}[
drs/.style={draw, rectangle split, rectangle split parts=2}
]
\node[drs,align=left]
{ foo
\nodepart{two}
foobar \\[.5em]
\tikz{
\node (operator) {$\exists$};
\node[drs,right=2mm of operator.mid]{
a%
\nodepart{two}{
b%
}
};
} \\[.5em]
\tikz{
\node (operator) {$\rightarrow$};
\node[drs,left=4mm of operator.mid] (left side)
{ 1%
\nodepart{two}
2%
};
\node[drs,right=4mm of operator.mid] (right side)
{ 3%
\nodepart{two}
4 \\
5 \\
6%
};
}
};
\end{tikzpicture}
\end{document}
Best Answer
The extra space is added according to the
inner sepandouter sepamounts. If you zero them out the image is tightly packed inside the node thanks toinner sep=0and moreover the arrows start right from the border thanks toouter sep=0