How can I achieve grid typesetting with a figure?

Question

One of the requirements of our publisher for the book to be published is grid typesetting. Of course I already had a look into how to use \raggedbottom and adjust all vertical lengths to achieve this. But there is one last thing where we need manual fiddling and I don't know where my error in thinking is.

So we have an image which originally is W267.0mm x H100.5mm and because we have 20mm margins on A4 paper, we scale the image to \textwidth and now have W170.0mm x H63.99mm or converted to Adobe Points W481.89p x H 181.39bp. My thinking is, that to achieve typesetting according to the grid. I find the next mutliple of 12bp which is 192bp and the difference is 10.61bp. If I apply this space, the text below the image should follow the grid again. Everything put in code it looks like this:

\documentclass[a4paper]{scrartcl}
\usepackage[left=20mm,right=20mm]{geometry}

\usepackage{lipsum,graphicx}

\setkomafont{caption}{\fontsize{10bp}{12bp}\selectfont}
\setkomafont{captionlabel}{\fontsize{10bp}{12bp}\selectfont}

\setlength{\textfloatsep}{0bp}
\setlength{\intextsep}{0bp}
\setlength{\abovecaptionskip}{10.61bp}
\setlength{\belowcaptionskip}{0bp}

\raggedbottom
\begin{document}
    \fontsize{10bp}{12bp}\selectfont
    \lipsum[1-3]
    \begin{figure}[h]
        \centering
        \includegraphics[width=170mm,height=63.99mm]{example-image-A}
        \caption{Just a test}
    \end{figure}
    \lipsum
    \lipsum
\end{document}

However I prepared a visual check of this page and used LaTeX for this as well, so I use the output of above and use it as Registerhaltigkeit.pdf for the following code:

\documentclass[a4paper]{scrartcl}

\usepackage[absolute]{textpos}
\usepackage{graphicx,pgffor}

\pagestyle{empty}
\begin{document}
\foreach \x [evaluate=\x as \y using int(\x+1)] in {1}{
\newpage
\begin{textblock*}{105mm}[0,0](0mm,0mm)%
    \noindent%
    \includegraphics[page=\x,trim=0mm 0mm 105mm 0mm,width=105mm,clip]{Registerhaltigkeit.pdf}%
\end{textblock*}%
\begin{textblock*}{105mm}[0,0](105mm,0mm)%
    \noindent%
    \includegraphics[page=\y,trim=105mm 0mm 0mm 0mm,width=105mm,clip]{Registerhaltigkeit.pdf}%
\end{textblock*}%
~
}
\end{document}

And the final result looks reveals that up to the figure everything is in register, thereafter there is an offset between the two pages:

What am I doing wrong, which numbers am I'm messing up or which internals skips do I miss? Thanks for your support as always.

---

After comment of @barbara beeton

The comment of @barbara beeton brought me to the point that I wanted to answer just use article instead of scrartcl, it doesn't change a thing, but this is not true. Using scrbook instead of scrartcl is not really changing something, but using book instead of scrbook does.

So here is my new code, which uses the standard book class and the caption package:

%\documentclass[a4paper]{scrartcl}
\documentclass{book}
\usepackage[left=20mm,right=20mm,a4paper]{geometry}

\usepackage{lipsum,graphicx}

%\setkomafont{caption}{\fontsize{10bp}{12bp}\selectfont}
%\setkomafont{captionlabel}{\fontsize{10bp}{12bp}\selectfont}
\usepackage{caption}
\DeclareCaptionFont{MyFont}{\fontsize{10bp}{12bp}\selectfont}
\captionsetup[figure]{font=MyFont,labelfont=MyFont}

\setlength{\textfloatsep}{0bp}
\setlength{\intextsep}{0bp}
\setlength{\abovecaptionskip}{10.61bp}
\setlength{\belowcaptionskip}{0bp}

\raggedbottom
\begin{document}
    \fontsize{10bp}{12bp}\selectfont
    \lipsum[1-3]
    \begin{figure}[h]
        \centering
        \includegraphics[width=170mm,height=63.99mm]{example-image-A}
        \caption{Just a test}
    \end{figure}
    \lipsum
    \lipsum
\end{document}

And now applying the second script, my result looks like this:

So you can see that after the figure now everything is set on grid, which means my calculation was right. However now I have some more questions:

1. What extra spacing is added between the figure caption and the following text?
2. Why does it match the vertical distance or where is this space taken from so the lines are matching afterwards again?
3. What extra space is added by KOMA-Script that obviously messes up the vertical spacing?

Maybe you have some hints for me?

---

OK, I found a partial answer and here is an intermediate result, I will also post an answer in the next days:

Answer 1

Basically you need to make sure that your figure and the space surrounding it will stay within the grid. As you used a figure environment but with "h" I presume you want it within the text (but I hope you are aware that this can float away as "h" only means "h if it fits". The problem is that floating figures get spae only on one side while intext figures get space above and below (and possibly \lineskip because the figure is just a big box).

Your best bet I guess is to define your own environment that starts a figure env but then typesets the figure content inclduding caption into a named box you measure that end alter its dimensions to fit your grid requirements by adding extra space above and below and then \usebox{myadjustedbox} and then end the figure env. In addition you need to adjust the floating pararameters to be rigid and fitting to whatever you use as height steps.

Alternatively if you always explicitly place the figures inside the text you can just avoid using the figure env and instead use \captionof (from the float package) to get a caption. The you are free play with the spaces around your fake figure as you like.

I know that this isn't a finished recipe, but this is roughly what is needed.

A good helper is \showoutput to see what spaces get added where. The extra space added by caption depends on the class (or package) you use but it is often \abovecaptionskip. The float parameters to adjust are \intextsep, \floatsep, \textfloatsep, etc.