NaNoGenMo / 2019

National Novel Generation Month, 2019 edition.
97 stars 5 forks source link

"Starship Trader Monthly" Magazine #7

Open greg-kennedy opened 4 years ago

greg-kennedy commented 4 years ago

Last year I wanted to generate a catalog of "Strange Items" - procedurally generated objects, rendered in POVRay, along with wacky descriptions, laid out in magazine fashion and rendered to PDF.

I didn't get it done.1 Part of the problem was that I took a vacation mid-month that ate a lot of time, the other issue was that "items" was too broad. The output was boring because there was nothing to optimize towards.

But it seems a shame to let all that PDF + POVRay code go to waste AND I still think the idea of a generative catalog has merit. It just needs more focus! So this year I'm taking another run at it, but with the goal of generating a Catalog of Starships. Or maybe more like a Classified Ads. All manner of used and new space vehicles, satellites, escape pods, and maybe even space stations, at low low prices!

greg-kennedy commented 4 years ago

Starship Trader Monthly

Read the novel here: https://archive.org/details/nanogenmo_2019_starshiptradermonthly

(126MB PDF file, 54 pages)

View the code here: https://github.com/greg-kennedy/StarAndDriver

Preview: image

greg-kennedy commented 4 years ago

Some of you may notice this entry contains far less than 50,000 words of text. This is intentional. As the GitHub repo explains:

...since the book is mainly a frame to show pretty pictures, and the common exchange rate is "One Picture == One Thousand Words", there is another way to meet the NaNo requirement: the book contains 50 pages of generated pictures (48 pages, plus front and back cover), thus the equivalent of 50,000 words.

Script / module language is all Perl. Most text data was taken from Corpora. Pretty pictures rendered by POV-Ray, a raytracer which IMO is especially well-suited to procgen art. Collation and layout by PDF::API2 library. Rendering took approx. 8 hours on a Pentium 4 (top-of-the-line for 2006!)

The very high-level of how the ship generation works is:

I went hard on the "art book" this year, but I probably won't do it again: collections of generated objects, or compilations of little vignettes, are probably better suited to Twitter bots than a "book".