Difference between revisions of "Talk:Targeting"
m (→Interpolation) |
(Some comments) |
||
| (3 intermediate revisions by 3 users not shown) | |||
| Line 6: | Line 6: | ||
Thank you. Just finishing checking and [[Logic]] (funny that currently he is at #14 due all duplicated entries) and [[YersiniaPestis]] does linear interpolation, as is [[Coriantumr]]. Wonder how [[Glacial]], [[Portia]] and [[Shadow]] does. Hmm... I can't see [[Gladiator]] does any kind of interpolation. Weird... » [[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] » 14:52, 22 September 2009 (UTC) | Thank you. Just finishing checking and [[Logic]] (funny that currently he is at #14 due all duplicated entries) and [[YersiniaPestis]] does linear interpolation, as is [[Coriantumr]]. Wonder how [[Glacial]], [[Portia]] and [[Shadow]] does. Hmm... I can't see [[Gladiator]] does any kind of interpolation. Weird... » [[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] » 14:52, 22 September 2009 (UTC) | ||
| + | |||
| + | [[Glacier]] is using linear, at least for now. I'm considering improving that but am not sure it's worth it.--[[User:Rednaxela|Rednaxela]] 17:31, 22 September 2009 (UTC) | ||
| + | |||
| + | Portia does it linear as well. I think calculating it precisely is impossible if the gap is greater than 1 turn and the opponent didn't do maximum rotation, speed or acceleration. --[[User:Positive|Positive]] 18:15, 22 September 2009 (UTC) | ||
| + | |||
| + | : It isn't really precise, but it is a lot more accurate than normal linear, especially when enemy reverse its direction, because it does simulated the velocity too. [[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 16:56, 23 September 2009 (UTC) | ||
| + | |||
| + | I think that the velocity transition is by far the most important part, and that if any of this can make a score difference, it would be simulating the velocity transition. Thus, I first plan to first stick with straight-line interpolation, but simulate a hypothetical velocity transition. One thing to importantly keep in mind is that bot/wall collisions can throw off one's estimation by allowing otherwise impossible velocity transitions. You need to keep an eye out for those 'impossible' situations and have a sane behaviour. --[[User:Rednaxela|Rednaxela]] 17:17, 23 September 2009 (UTC) | ||
Latest revision as of 18:17, 23 September 2009
Interpolation
In almost every melee gun, you need to do the interpolation to get approximate enemy location. I wonder how are you doing this in your melee gun? Just normal linear interpolation on fancy interpolation? » Nat Pavasant » 14:27, 22 September 2009 (UTC)
I just do linear interpolation in Diamond. Once I detect that a gun wave has passed an enemy, I calculate the angle and distance the enemy has moved since the previous scan, assume he moved an equal distance each tick at that same angle, then trace back his movement and my waves until I find the right spot. --Voidious 14:35, 22 September 2009 (UTC)
Thank you. Just finishing checking and Logic (funny that currently he is at #14 due all duplicated entries) and YersiniaPestis does linear interpolation, as is Coriantumr. Wonder how Glacial, Portia and Shadow does. Hmm... I can't see Gladiator does any kind of interpolation. Weird... » Nat Pavasant » 14:52, 22 September 2009 (UTC)
Glacier is using linear, at least for now. I'm considering improving that but am not sure it's worth it.--Rednaxela 17:31, 22 September 2009 (UTC)
Portia does it linear as well. I think calculating it precisely is impossible if the gap is greater than 1 turn and the opponent didn't do maximum rotation, speed or acceleration. --Positive 18:15, 22 September 2009 (UTC)
- It isn't really precise, but it is a lot more accurate than normal linear, especially when enemy reverse its direction, because it does simulated the velocity too. Nat Pavasant 16:56, 23 September 2009 (UTC)
I think that the velocity transition is by far the most important part, and that if any of this can make a score difference, it would be simulating the velocity transition. Thus, I first plan to first stick with straight-line interpolation, but simulate a hypothetical velocity transition. One thing to importantly keep in mind is that bot/wall collisions can throw off one's estimation by allowing otherwise impossible velocity transitions. You need to keep an eye out for those 'impossible' situations and have a sane behaviour. --Rednaxela 17:17, 23 September 2009 (UTC)
