ive been interested in non-shooting FPSs for a while, and watching our german shepherd play at the park is a big inspiration.

FPSs are interesting to me for a variety of reasons, one of which is the capacity for fluid motion in 3D space constrained to a surface. the world record half-life speed run really drives this home https://youtu.be/VtI5HM7GVGY?t=746

so the idea is a game where you can run and bark to shepherd sheep where they need to go. the sheep are driven by a boids-like flocking algorithm

mechanically its a 2D game played in 3D, like early FPSs eg Doom and Wolfenstein. so my plan is to prototype it in 2D and then reproject that into 3D with a basic terrain, trees as obstacles, fences marking out boundaries and goals. i think i can skip any skinned mesh animation too and just model one sheep and make them bobble procedurally as they move around, and that's what we call a game dev production hack kids 😎

mechanically im thinking that your position will repel the sheep near you, and barking will increase the velocity of the sheep within earshot. so you're running around and barking at them to get them where they need to go. that's the jumping off point at least.

sheep

sheeps

forming "neighborhoods"

brb walking the actual dog

ok im back, now we have neighborhood centroids

alignment + cohesion

hmm why are they all going to the corner...

hm my implementation is probably off. im going to refactor following https://processing.org/examples/flocking.html which seems to get it right

flocks of sheep!

basic boundary avoidance. they kind of get caught between avoiding the boundary and following the flock, might clean that up later

boundary avoidance works against arbitrary line segments, so i have the tools to design levels

so I'm reworking boundary detection to not be based on a ray cast forward from the sheep but rather from a circle around the sheep which is more like how the other flocking rules work. I expect that to be a little bit more fluid, I don't like how janky the current thing is. I was hoping to get boundary detection and the dog in by tonight but that might have to wait until tomorrow!

ugh linear algebra is so beautiful when it clicks. i have a working implementation of projecting a point onto a line segment, giving you the shortest distance from that point to the line segment. simple enough to compare that length to a circle's radius to see if the circle is touching the line at all.