Ok, I have no typographical knowledge, but this:

just looks wrong to me, intuitively. The spacing between the 't' and the 'u' in "Einleitung" seems clearly too wide, imho.

That's the code (LuaLateX):

\documentclass[a4paper,twoside,12pt]{scrreprt}
\usepackage{fontspec}
\usepackage{lipsum}
\setkomafont{sectioning}{\normalfont\normalcolor\bfseries}
\usepackage[ngerman]{babel}
\usepackage[babel]{csquotes}
\begin{document}
\chapter{Einleitung} \lipsum
\end{document}


Have I done anything wrong? Forgotten anything important?

-
I don't think this has anything to do with either Koma or LuaTeX. It is a result of the font in use. (Compare the report class and run with pdflatex using [T1]{fontenc} and lmodern. – cfr Aug 1 '14 at 23:56
Have you tried to print it and see if it looks as bad printed? The screens know to be misleading. – yo' Aug 2 '14 at 10:00
Indeed. Latin Modern definitely has issues when viewed on screen. For example, small caps can look uneven on screen even though they will look fine printed. – cfr Aug 5 '14 at 10:34

Choose another font, eg Libertine:

or load a feature file which modifies the kerning between t and u. Save the following as lm.fea or any other file name with extension .fea:

languagesystem DFLT dflt;
languagesystem latn dflt;
feature kern { pos t u -60; } kern;


and load it inside the font definition and, of course, run it with lualatex:

\documentclass[a4paper,twoside,12pt]{scrreprt}
\usepackage{fontspec}
%\usepackage{libertine}
\setmainfont[FeatureFile=lm.fea]{Latin Modern Roman}%%%%% Feature file
\usepackage{lipsum}
\setkomafont{sectioning}{\normalfont\normalcolor\bfseries}
\usepackage[ngerman]{babel}
\usepackage[babel]{csquotes}
\begin{document}
\chapter{Einleitung} \lipsum
\end{document}


-
Unrelated: Where can I learn about those .fea files? – Henri Menke Aug 2 '14 at 8:36
However, I wouldn't do that. While the kerning seems wrong at the first sight, I think you won't notice it in print. And Latin Modern Roman is a print font, isn't it? – yo' Aug 2 '14 at 9:59
It is also seen when printed! – Herbert Aug 2 '14 at 11:14
Yes, but it looks different printed! Definitely true of small caps for Latin Modern. Those can look really uneven on screen but perfect on paper. @tohecz is right, I think, except that it may not be so much a question of your not noticing flaws as the non-existence of those flaws. (I haven't tried this with the tu combination to check, though, so it also may not.) If you adjust the kerning, you could end up doing more harm than good if you aren't careful. Also, it will matter whether different optical sizes behave differently. – cfr Aug 5 '14 at 10:38

This is a screenshot at very high resolution, obtained from

\font\x=cmbx12 scaled 3000
\x Einleitung
\bye


and using the maximum zoom on my previewer

As it can be clearly seen, the distance from the bar of the ‘t’ to the serif of the ‘u’ is just the same as the distance from the right stroke of the ‘u’ to the upper serif of the ‘n’.

This is the whole word, at a big zoom:

The problem you see at your small resolution seems just due to pixel rounding in the PDF viewer, which shifts the ‘t’ a bit to the left.

Look at the pictures at full resolution.

-
Equalising the distances you noticed to be equal does not necessarily result in a good kerning: For un there is a strong stem on the left side of that gap, while for tu there are only thin curves. Also, the size of the introductory scerrenshot was far too high for this to be an effect of (bad) hinting. – Wrzlprmft Aug 2 '14 at 14:00
@Wrzlprmft It explains the choice, at least; but I continue to think that the choice is good. If you notice, the base curve of the ‘t’ in your image ends more to the right than in the high resolution printout. – egreg Aug 2 '14 at 15:12
If at all, the only by fractions of a pixel – here are the two overlapped. I wonder, if we can ask the guys on Graphic design whether this is good kerning without it being a too opinion-based question. – Wrzlprmft Aug 2 '14 at 17:34
On the screen the word falls apart in both resolutions in my opinion, the reason being the the "rectangular" white space between t and u is bigger than, say, the white space between the two stems of an u. So I don't think it is just a screen rounding issue (though there it gets easily even worse). So in my opinion one could argue for a small negative kern. (by the way, the distance between bar and serif seems to be slightly smaller then between u and n (42 to 48pt on my screen) but I think it should be even smaller to reduce the bulky white space. – Frank Mittelbach Aug 5 '14 at 19:25