Jupyter seems to be working towards support for converting notebooks to Reveal.js slideshows, but it’s not working quite yet, at least for me. Luckily, there’s a pretty easy way to turn your Jupyter notebooks into slideshows not only with Reveal, but also S5, DZSlides, Slidy, Slideous, or the latex beamer package: Pandoc.

Pandoc is a document converter that can read a good number of formats and turn them into an even greater number. The idea is to use nbconvert to convert your notebook to markdown—which, if anything, is Pandoc’s native language—and use Pandoc to convert to the slide format of your choice.

Here’s the one liner:

nbconvert --to markdown --stdout path/to/notebook.ipynb | pandoc -Ss -t revealjs -o slides.html

The -Ss handles smart quotes and produces a standalone file. You can change the -t format to whatever slide engine you want. If you would like your lists to appear incrementally, add the -i flag to the pandoc command, like so:

nbconvert --to markdown --stdout path/to/notebook.ipynb | pandoc -Ssi -t revealjs -o slides.html

This works well for me with images and code blocks. You do have to keep in mind the way Pandoc parses markdown for slides, but it’s pretty simple. The thing to keep in mind is that the highest-level header that is followed by non-header content is the slide-level title, the level above that becomes the section-level title, and levels below that are in-slide headers. You can actually get a fair amount of customization with Pandoc options, like the location of the slide libraries or slide themes, but I’ll leave that as an exercise for the reader.