About Public Domain Pulp

Public Domain Pulp is a free, online collection of visual novel transformations of public domain books.

Free!

Books in the public domain are free, so their visual novel counterparts should be too. You can read them all here: no payment, no account, no nothing required.

Open Source!

Each visual novel on this site has its own corresponding public Github repo. You can clone them and build them and read them offline. (They're all just static html pages!)

The website's code is all MIT-0 licensed, and its text and image contents are all CC0 licensed.

Entertaining!

Regular books are boring. Visual novels are fun.

The Best Prose!

On this website, I steal from the best. (And also Steinbeck.)

All visual novels here keep the original text of their respective books. There's no modifying, simplifying, or bowdlerizing of prose — only the unimpeachable greatness of the classics.

Heavy Editorializing!

Okay, so maybe sometimes even the classics have their questionable moments. (Looking at you, Tolstoy.)

As a solution, Public Domain Pulp's visual novels contain highly opinionated editor's notes: ostensibly there for adding context, but mostly there for pointing-and-laughing at the authors' writing choices.

Blog Posts.

There's also blog posts (yay…). Many of them also involve pointing-and-laughing.

Actually, the rest of this about page is basically just a blog post. Read on if boring details interest you:

---

The Background

The Public Domain Pulp site began with a dream to provide free and open-source visual novels to everyone.

Actually it was less of a dream and more of a business idea. And less about providing free visual novels altruistically and more about undercutting the VN market for myself. In any case:

The Concept

The situation is that all existing visual novels have terrible prose. Therefore, a visual novel with good or even average prose would be a standout.

At the same time, regular novels are boring — books being inherently the least interesting medium. They make up for this by being the best-written medium, but that advantage has only slowed their inevitable and ongoing decline.

However, if we could combine the goodness of books with non-boringness of visual novels, we could have a pareto improvement on both. Since unlike your typical movie or TV adaptation, visual novels can retain all the original good prose.

The Prerequisites

Getting the good prose is an easily solved problem: I steal it from Standard Ebooks. Not only do they have all the public domain classics — their em dashes are nicely formatted as well.

The "visual" part of the "visual novel" is harder. In fact, it was strictly uneconomical prior to 2025. See, while AI models existed before then, they simply weren't good enough for what visual novels require:

• Hundreds of images — sprites and backgrounds — all drawn in the same style.
• Self-consistent variations on the same image (e.g., different sprites for the same character).
• Images to all be period- and setting- and description-accurate to both the book as a whole and the particular contexts around individual scenes.

But the AI progress was (and still is) continuous, and 2025 was the tipping point. The models are now sufficiently smart and accurate that VN-adaptations at scale are practical.

The Implementation

Once you've got the writing and the images, you need to smush them together somehow. For a complete visual novel, you need metadata specifying all of:

• Line break positions.
• What background to have on-screen.
• What characters to have on-screen.
• What sprite to show for each on-screen character, taking into account all of:
  • Character's current age.
  • Character's current clothing.
  • Character's current expression.
  • Character's current position (e.g., whether they're holding something).
• Where to position each character on-screen.
• How to size each character (both relative to background and to other characters).
• Who the current speaker is.
• What name to display for each speaker (which might change over the course of the book).

The solution for containing all this structured metadata is… a text file! (Since if you aren't using plain text in <current-year>, what are you doing?)

The metadata-enhanced pulp.txt file controls all of the above, living alongside the book.txt file (the original text) in all the VN Github repos. Both files are read in by the Pulpifier as part of converting text to static html.

And that output, along with all the images in the image/ directory, is all it takes to "run" the visual novels.

(TODO: In the future, create and link to a blog post explaining the pulp.txt metadata format in more detail.)

The Goal

Ideally, all public domain texts could be converted automatically into VNs. And in theory, a sufficiently advanced (and well-funded) LLM+T2I pipeline would be able to do so.

For now though, full automation isn't yet practical, for a variety of reasons. Aside from the "drawing" of the individual images, the VN conversion process is still largely manual.

But this doesn't change the goal: the adaptation of all the classics of the public domain into VNs. It just means that it will take time to get there, and that it'll be dependent on:

The Monetization

Ultimately the site does need a source of revenue, since the building of the catalog is costly: there's the server ($32/month), and there's also the image-generation API credits (~$100/book, but with big error bars based on book size).

But the biggest cost is opportunity cost, what with the VNs taking significant time to create. They currently take about a full week's worth of work (again with big error bars for book size) in the reading, editing, re-reading, generating, re-editing, re-re-reading, and re-re-editing.

(TODO: In the future, create and link to a blog post explaining the VN-conversion process in more detail.)

In order for the site to be a successful full-time endeavor, it needs a revenue strategy. The current plan is that I'll incorporate Amazon affiliate links to books. I reference books ad nauseam anyway, so such links should slot unobtrusively in. (At least, in comparison to regular ads.) Plus, who wouldn't benefit from owning more books?

Course, this is all tentative. If this plan doesn't work out, I may switch to another. (Soliciting donations is always an option, but for now, I don't have anything set up. So just donate to Standard Ebooks instead.)