And my suggestion way up at the top of this thread was to use Pandoc MMD → PrinceXML → PDF for the second part of your puzzle. CSS for print is more logical than LaTeX, and links more closely to the web[1]. But I understand we all have limited time and nerdling on code distracts us from the more important things in life, like music in your case!
[1] for nerds, it would be possible to use the same output and switch the CSS for browser vs. print using the same document, this is the whole point of separating content and style… But that qould require pretty ninja-level CSS skills…
I’m finding two ways PDFs can be generated: going through the DOCX format.
One is the DOCX generated by Scrivener. It is clean, compatible, and works well. The issues are with the images that have to be included as embedded or linked files. There are serious issues with the smaller images, like icons, that can’t be smaller than 25px (too big for icons shown inline with the text).
The other one is a bit trickier, but it can work very well: exporting a MultiMarkdown file from Scrivener, and then let Quarto convert it into a DOCX file. This lets one use the same syntax used for a Website, including the wonderful list tables added as a Lua filter. No image is a problem, since what is compiled is a Markdown link. Any embedded/linked image is just a preview to be removed during compile.
Word says that this latter is a damaged file. It can however repair it, and open it well. It can then convert it to a “modern” file format.
As discussed on this forum, Scrivener’s DOCX generator, which is not from Microsoft, sometimes is not good enough for some people. The recommendation is to instead use RTF which is Scrivener’s native format thus requiring no conversation and easily importable into Word and most all other word processing programs.
These days, I’ve been working with Word as a formatter for the DOCX files generated by Scrivener.
I’m nauseated. Bugs, obscure ways of doing things, the magnum chaos of the ribbon, the obtuse way of promoting things that I don’t find more relevant than others, the user discussion pages never getting to the point, or even containing viruses…
I even had to find discussions about the lack of a Keep with Previous feature, that most users can’t understand it can’t be replaced by Keep with Next. I obviously come from a different civilization.
And the generated PDF is really unsatisfactory. It’s simply bad. I guess the best solution is to load the DOCX file into a page layout program, and do all the final formatting there.
Word, at least on Mac, is utterly terrible at PDF generation. It has an online mode that depends on microsoft servers and an offline mode that totally screws document structure and navigation. It is all sorts of terrible. LibreOffice is head and shoulders a better tool, with many detailed options for PDF production. LibreOffice is more flexible in a whole bunch of other ways too, but it is also quirky and you’ll hit bugs as well, but on balance it works better than Word…
For technical manuals, a programmatic solution which involves either/and Pandoc, LaTeX, Typst or HTML + CSS for print will always get you a more refined workflow with the caveat of a higher technical knowledge required…
That’s the route I’m pursuing. At the moment I can generate excellent websites from Scrivener, going through Ioa’s excellent multisplit script, and processing it via Quarto.
Unfortunately, I also need a PDF, both for proofreading and as the final output. Here, LaTeX is out of discussion, because how complicate it is to use fonts, languages, and the lack of accessibility (whose development is so glacial to give little hope).
Typst has not been mature in my tests of the recent months. I’m following its development, since it is the/a planned tool for formatting PDFs. It is not there, yet.
But while evaluating the alternatives, I’ve got suspicious that using a page layout program could be a viable solution. I’m currently developing a workflow to go to InDesign/Affinity Publisher though a DOCX file generated from Markdown.
A page layout file would also be an intermediate file format that I can give to my distributors who want to stay in the proven path. I can ask my partners to edit an InDesign/Affinity Publisher file, hardly a Typst file.
Going from the DOCX file to a page layout program has the inconveniences I described elsewhere about linked images. Once mangled by the DOCX format, linked images have to be manually relinked one by one, or you have to develop a workflow where you let the page layout program generate its own set of mangled images.
At the moment, I’ve not developed a working Scrivener → DOCX → page layout workflow, so I’m using Word as a temporary solution for formatting the output. I can’t wait to get rid of it.
Apple Pages would at least be a more pleasant working environment, but it also lacks things and can’t converts some finer details. In addition, it will never be a viable exchange file format for my partners.
To be fair with XeTeX or LuaLaTeX, OTF fonts with full UTF8 support and a zillion languages are really well supported (much better than almost anything else, you can do detailed OTF adjustments as required). Accesibility (in terms of ease-of-tweaking) is the major downside IMO. But LaTeX remains the king of the hill for really complex layouts, there is almost nothing that can’t be done without one package or another.
The critical remaining issues for me are float placements (especially with multi-column layouts) and line numbering (I mostly output draft research papers where line numbers for reviewers is a critical feature). These are being actively worked on and so I have great hope. There are some workarounds for floats, which may be good enough to be honest. As i do like Tufte-like margin-for-content layouts a bit more work on this would also help. But people are already doing amazing things with Typst. My hope is we are months away from a robust PDF generator for 90% of all usecases.
Do not discount HTML + CSS for print. This can be generated from Quarto, gets you PDF in a generally flexible manner and as it uses print CSS it is quite easy to tweak and covers much of what is needed for print:
In this workflow Scrivener > Quarto as a base and then Quarto to generate EPUB / web site AND PDF all in one go… Of course this doesn’t solve the problem if you need to collaborate with Indesign / AD users, I don’t think there is much of a solution there. DOCX is also not really a solution for that as DOCX > indesign is one-way, how to collaborate back and forth?
With it, I mean the ability of reading the PDF structure with assisting technologies for visually-impaired people. It is something that has been under development forever in the LaTeX world.
Typst doesn’t have it, too, but the developers have expressed their willingness to make it a priority.
I know that PDF shouldn’t be considered the preferred file format for this type of reading, but again – I have to produce PDFs, that will also be the only available output in some countries.
InDesign/AfPublisher would only be a one-way format. I’ll have to give it to those distributors who don’t want to work on a website and only want a PDF for their version. With over 100 distributors, there will be many of them preferring this solution.
As for my collaboration with the translators I directly manage, it will likely happen on the DOCX files I will export from the Scrivener project. I would like to keep those translations assembled into separate Scrivener projects, to be constantly updated and be used as masters for the relevant websites.
My proofreaders will continue to receive PDF drafts. How they will be generated is the subject of this discussion.
I’ve recently completed a single-sourced project made with Scrivener, and ending up as PDF and Web Site.
Scrivener was as usual a bless for structuring and mass editing.
I really missed the ability of linking images from drives using the Markdown syntax. VS Code does it, and it does work. I really hope Scrivener can replace its older method with this more up-to-date one.
Thanks to the multi-file output script by Amber, and the kind power of Quarto, generating a Web Site was extremely easy, after getting the basics (and some less basic stuff…).
The big minus was generating a PDF. Templates for LaTeX were completely extraneous to the graphic style required for this project. Typst is still at the very early stages.
I ended up deciding to output to DOCX, and do the PDF in Word. It was a hell, and the result was at best mediocre.
I would have liked to go from DOCX to InDesign, but this would have resulted in an incredible amount of additional work. DOCX scrambles the purity of the original project, forcing one to redo things like image linking, lists, styles in a page layout program. It simply doesn’t think as a page layout program.
I really hope InDesign and/or Affinity Publisher will accept Markdown as a direct input, sooner or later. This would remove the horror of DOCX as an intermediate files, and allow for dealing with the PDF side of the process with a dedicated page layout program.
I think this would still be the best way to generate a PDF. There are small, fundamental things that you can easily to there – adjusting a text line length to avoid orphan syllables, adjusting the vertical space to make things fit in one page or take more space and air, do minimal edits to the position of a caption around an image.
I want page layout programs at the very end of the process, but I want them. Automatic output is not making me feel good.
You can visually alter the PDF in any way imaginable by customising templates for Quarto/Pandoc; the big caveat is that TeX is not the easiest language to learn for layout, but it is possible. In TeX you can also tweak the fine layout by adding or reducing spacing per page, structural items as much as you want. The same thing will apply for Typst and does already for CSS-for-print paths, as text formats can be easily tweaked, and in VSCode you can see a live preview of the PDF as you work…
As you mention, if the big visual DTP developers would support what is now a pretty ubiquitous format (markdown), that would be a text-to-visual layout alternative workflow.
By the way, markdown to ODT output from Pandoc (and soon Quarto) now supports image linking to external image files, and it does work better IMO for PDF generation than Word. People seem to treat LibreOffice as a poor Word, but in many ways its feature set is better…
That’s the problem: you can do everything with TeX, but you need a way to access it in reasonable time and efficiency. I was hoping in Overleaf, but in the end, all the templates around smell LaTeX and are a variation on the basic templates. Not surprising, since this system has found success in the academic world, but not in the industrial/commercial world.
I’m hoping in Typst, whose syntax is much more accessible. And, probably, it will be a more coherent system, including everything that one has to assemble from several pieces in the TeX world. In any case, at the moment I don’t see building templates in Typst and fine-tuning projects in the source code something that can be really done with an acceptable balance of cost/benefit ratio.
Unfortunately, since more than a decade InDesign has stopped developing in a way that technical communicators can really use. And in these latest times the target audience has strongly shifted away from technical communication. To be honest, I suspect that the inclusion of several FrameMaker features in CS5.5 and CS6 was just an effort of simplifying the product line, without too much conviction.
Affinity, at the moment, is running behind InDesign, and doesn’t seem too interested to support anything going further to design and architecture. Technical communication seems something that is not in their ethos.
That’s great news, and I’m happy they were successful in doing it (I might even be the one that triggered this change…). Unfortunately, here we are again in the incommunicability between words: nor InDesign, nor Publisher, can read ODT files. They pretend the destructive intermediation of Word.
