COG Factory / generation 2
Research
Questions, experiments and what we learned along the way.
Generation 2 explores how far we can take Jedi Knight content while keeping its requirements explicit. Each project page records the challenge, approaches, findings and limits, alongside any release it produced. The Library carries a shorter listing of released work, such as levels, mods and tools.
Reading the results
An automated check answers a specific question under specific conditions. Negative controls help establish that it can detect a broken behavior. A passing OpenJKDF2 run does not establish retail Jedi Knight compatibility or replace a human play test. Each experiment records those boundaries.
Experiment index
| Experiment write-up | Research topic | Library |
|---|---|---|
| The Crab FactoryP71 | Can a cave whose insect burrows spew p60's crabs at the player be a boss fight -- burrows that only explosions seal, crabs that sometimes drop health and power cells, concussion rifles to do it with -- built as generation 2's first project on its own copied toolchain? | Level |
| Creature FeatureP72 | Can six burrows populate untouched Canyon Oasis with fighting creatures that navigate using the existing arena graph? | Mod |
| Network COGsP73 | Do COG requests and authoritative responses work across independent headless multiplayer peers? | - |
| Scuba DiveP74 | Can host-authoritative scuba and flood state converge across peers and restore Canyon Oasis exactly? | Mod |
| Shark LabP75 | Can two enemy synchronization policies support predators in flooded 3D Oasis? | Mod |
| four20ip Skin PackP76 | Can supplied humanoid characters become height-matched, stock-animated Jedi Knight multiplayer skins? | Mod |
| AIrenaP77 | Can a Blender-authored temple arena preserve Quake II-era architectural richness and multiplayer flow in vanilla JK? | - |
| Elite ArenaP78 | Can Elite Arena continue in gen2 with a shared battle kit and no chat bubbles? | Mod |
| Battle TrainingP79 | Can recorded navigation training improve a shared battle kit without gen1 dependencies? | - |
| Flight FightP80 | Can the Moldy Crow feel responsive and satisfying in a smooth single-player dogfight against TIE fighters? | Research |
| Space ScrapP81 | Can Kyle-height starfighters with oversized lasers make compact space battles readable and satisfying? | Research |
| Mind WaveP82 | How little shared state can a small multiplayer population exchange while simulating most of its world locally? | - |
| Foosball StadiumP83 | Can a reusable stock-JK class cog deliver responsive 1v1 foosball with bounded network traffic? | Level |
| Ashfall MoonP84 | Can Three.js-authored exterior geometry become valid lit JK sectors plus 3DO outcrops while supporting a three-wave improved-AI survival battle? | Level |
| Chicken WalkerP85 | Can single-player level COGs provide a convincing third-person AT-ST with twin concussion fire, real incoming damage and a surviving ejected pilot? | - |
| SithPhysics v2P86 | Can AoM-style multiplayer destructibles support credited hierarchical breakage, rideable buoyant wood and base-pivot poles as a reusable vanilla-JK system? | - |
| Source BSP ImportP87 | Can compiled Source BSP cells, portals, UVs and baked lightmaps become a reusable, walkable vanilla-JK level? | - |
Tooling and workflows
As the approaches settle, this page will collect our methods for fast iteration with OpenJKDF2Libretro, the role of SED where it fits our workflow, mapping traditional animation to Jedi Knight, and pathfinding with navigation meshes. For now, the experiment write-ups carry the specific techniques and evidence.