Computers in IRL are not 100% accurate, I've posted a response in a Sensors topic about turret/AI "getting better info" about target position, speed, movement direction. this would make the AI system feel more realistic, the standard system thats here now can stay when there isn't any sensor system in place, but if a ship/turret has sensor systems the AI gets a boost to it's accuracy.
A new Turret accuracy mechanic.
- Thread starter Keptick
- Start date
- Joined
- Jul 6, 2013
- Messages
- 451
- Reaction score
- 108
I had an idea but its based on some assumptions about the current AI targeting mechanic.
I am assuming that the AI is aiming at a circle around the core and firing at a random direction from the core on that circle. The radius of that circle is scaling with the distance and accuracy setting. So the closer you are the smaller the circle is etc. It is also aiming at that circle around where the core should be when the ammunition gets there ie. putting lead on it. The AI is also firing focus firing so the enter array of weapons is focused on that target point.
Where it breaks down is when the target isn't changing its vector much its impossible for weapons to hit the core as their aiming around the core. Unless the target movement shifts the core into that circle before the ammunition gets there. (impossible with beams)
Assuming all the above assumptions about AI targeting. It is possible to change the targeting mechanic to give it a chance to hit stationary or small targets without causing core drilling and hopefully simple enough to not create to much lag. Change the targeting to a sphere around the target with a random 3D direction on the sphere around the core. That way while the majority of the shots should be in that circle area, but some shots would end up behind or infront of the core and closer. There is small chance of aiming directly infront or behind the core. You might have to make the default accuracy a little worse to take this into account.
I am assuming that the AI is aiming at a circle around the core and firing at a random direction from the core on that circle. The radius of that circle is scaling with the distance and accuracy setting. So the closer you are the smaller the circle is etc. It is also aiming at that circle around where the core should be when the ammunition gets there ie. putting lead on it. The AI is also firing focus firing so the enter array of weapons is focused on that target point.
Where it breaks down is when the target isn't changing its vector much its impossible for weapons to hit the core as their aiming around the core. Unless the target movement shifts the core into that circle before the ammunition gets there. (impossible with beams)
Assuming all the above assumptions about AI targeting. It is possible to change the targeting mechanic to give it a chance to hit stationary or small targets without causing core drilling and hopefully simple enough to not create to much lag. Change the targeting to a sphere around the target with a random 3D direction on the sphere around the core. That way while the majority of the shots should be in that circle area, but some shots would end up behind or infront of the core and closer. There is small chance of aiming directly infront or behind the core. You might have to make the default accuracy a little worse to take this into account.
NeonSturm
StormMaker
- Joined
- Dec 31, 2013
- Messages
- 5,110
- Reaction score
- 617
dva, dha = vertical/horizontal angle difference (current - last)
or dx, dy = vertical/horizontal position difference (now - previously)
dd = distance difference (now - last)
Just multiply these as in "value *= random(0..1)*0.2" == 20% inaccuracy.
And add it to the enemy core position / aiming orientation.
or dx, dy = vertical/horizontal position difference (now - previously)
dd = distance difference (now - last)
Just multiply these as in "value *= random(0..1)*0.2" == 20% inaccuracy.
And add it to the enemy core position / aiming orientation.
