These are inspired by ideas of the Luddites going in for a lot of pomp in their warship designs to awe the faithful, as opposed to the plainness of Luddite society. This was typical of many medieval societies, where the major faiths, being support systems for the societies' castes and hierarchies, wanted to show off their power through decoration and symbology. Grand cathedrals in space, ridiculous levels of metalwork on warships? It totally happened, historically.
Today's emphasis on functional design for military objects is a relatively modern idea; one can go back to the 19th Century and see that even serious military-grade systems often came with aesthetic flourishes reflecting their cultural heritage. Even today, it's kind of funny how people think some new weapon system is "cool" or not largely based on how it looks, and that can influence what gets funded for serial production and what remains merely conceptual. One presumes the Lockheed Martins of the world have stylists thinking about how to make things
look cool after the engineers figure out to make them work.
The basic idea was that these ships, under the filigree, remain Luddite in design- heavy on armor, low on high-tech features. So they're largely based on base designs I'd already decided were "Low Tech" or "Midline". But they're both a functional device and a symbol, which felt right.
The Path, of course, eschews such frippery, as they both lack the financial resources and have decided that this is antithetical to Ludd's vision of individual simplicity and harmony with life everywhere... and is hubristic besides. This is one of the few points of their heresy that gets them much positive attention from other Factions- the Pirates, for example, agree that it's ridiculous that Ludd's armed representatives swan around in gilt while their people live archeotech lives, and even representatives of such factions as the Hegemony, while polite in public, often laugh at the waste and expense that these ships' extraordinary ornamentation require.
After a few experiments, black and gold and white and gold both conveyed the sense I was after.
On the technical side, this was a fun series to mess with. I've learned a lot about the concepts of
statistical convergence (i.e., how many inference steps typically result in the best probability of a given set of inputs being followed, or "how to get OK results without waiting so long") and
LoRAs to stabilize stylistic elements and keep stuff roughly on-theme. In this case, I'm
using this LoRA on top of the main model, prompts and img2img base designs.