Skip to content

Notes

  • Inspect: we can still read original values for resources outside our knowledge range when just pointing on something. This should give a "unknown" back. Bonuspoints for having the knowledge gradually distort the reading, while giving back a clear "unsure, unknown" etc - so we know we cant rely on this information
  • Hunting has still huge performance issues, i guess because we sample trail or agent texture each step - this needs improvement
  • general pathfinding / movement (in gathering also) has still performance issues - one idea would be to avoid updating full textures, but only parts of them that changed.

  • check warum die plant / water overlay maps (eventually all that uses the same idea) - with hard very visible low rez pattern dissapear (with the knowledge map) - because the map itself is rather smooth. What is the render logic here - just to make sure that we dont loose visual detail for nothing.

  • Trails / Tracks. Currently i would see the map we have more like a scent map, But we could make a seperate real trails map out of the same data. each step we write the current agent position into a trail texture (stamping it on) then we substract a full value (the decay) NO diffuse or anything. We basically just record exact position over time. As for the visualisation - lets say we have a "trail texture" a sprite that shows cutout trails / decal . Now we decide if we render this at a certain position (and a certain dithering applied, like the rest of the world) based on the trails texture - so longer ago trails would disappear. The resulting map would look visually different and it would show actual agent position clearly over time.

  • check if the current render / time-advancement setup could be done independendly / not so tightly coupled. The idea is that we run the simulation - and the render just consumes what is currently there - this should fix rendering lag, it wouldnt fix the buildup in the cpu.

  • When running the Pathfinding activity, clicking outside of the range should cancel the activity while in the first path selection modus.

  • idea: make local pathfinding range tied to the knowledgemap this would represent that we can identify faster and further away with more knowledge about the land. It would be most likely good to have a curve with points i could define how this exactly shall scale - but essentially we say the knowledge map has x times y points. times 255 = maximum amount this can have. and then we either linear scale (maybe with an offset, so we have a base range initially) or have a defined curve. And we need to define a max range (most likely 300 as this is the size of dijsktra we tested with) but within this range we scale. Upside is - this is now an ever moving part. Got slacky and didnt wandered around much lately, reduced range. Made a big journey through half the world - huge range increase. Essentially this is our first skill - and its entirely based on our actions and live.