(ooc: Rob, it is public (at least somewhat) knowledge that the girls are AI, no? If not, disregard. And yes, I know, this is all going off the original topic something fierce; bad habit of mine.)
*private msg to Yoriko* "Just between us girls... it's an expansion of something KJ once thought of but put on the backburner due to lack of processing. Rather easier for me to do since I *am* processing, ne?
The working name is Omegajutsu. The basic idea is that there's a far-less-than-infinite number of ways for the human body to transmit force, and by looking at limb positioning you can figure out what the few most efficient at any given moment are based on a few criteria. The way joints bend, and how muscle groups attach to the skeleton determines this in the same way as range of motion. Similarly, you can categorize responses based on some criteria.
It's sorta a predictive scripting system; I can look at someone's stance (and distance, etc) and know the top few moves they're likely to pull from there because there's going to be X number that they couldn't efficiently do with their limbs in that position. Once they start to execute one of them, things narrow significantly; what exactly they're doing becomes more precisely defined, along with what it could transition into if it were a feint or whatever. From there, a bunch of decision branches come up; top few counters for any move depending on preset criteria (do I want to pummel this guy unconscious as quickly as possible, or just keep them from doing me injury?) and how they could possibly react to my counter, and what they'd do after that, and so on. Even correlates stances and actions to come up with a model of likely training and more exact physical capabilities, which allows the predictive capabilities to go a few seconds farther ahead of realtime than normal, and better weigh the decision branches.
Probably something that isn't really doable for biological-based brains; I'm informed that sifting through the datasets like that in conscious thought reliably is improbable at best, especially with the timing that some of them require. On the other hand, as long as the user's CPU and RAM specs are up to it, it doesn't really have any physical spec requirements. Overpowering or moving faster than one's opponent opens up more options, but even if they're significantly faster and stronger it still works.
Or should; I'm on a full combat chassis, so how'd I know? ^_^;;
I don't know how your movement systems are coded; personally I suspect I'm one of the closest to hardtech, programming-wise but it'd be rude to ask to poke at others source code. I've no automatic reflexes or anything, but I can learn equivalents without muscle memory. So, how applicable this would be to you or, frankly, anyone but me is unknown as yet, but I can send you the source files if you want. I'd appreciate not spreading it around too far though; while the results are promising, it's still basically in late-alpha and I'm not sure how far I want to spread it out besides."
*private msg to Yoriko* "Just between us girls... it's an expansion of something KJ once thought of but put on the backburner due to lack of processing. Rather easier for me to do since I *am* processing, ne?
The working name is Omegajutsu. The basic idea is that there's a far-less-than-infinite number of ways for the human body to transmit force, and by looking at limb positioning you can figure out what the few most efficient at any given moment are based on a few criteria. The way joints bend, and how muscle groups attach to the skeleton determines this in the same way as range of motion. Similarly, you can categorize responses based on some criteria.
It's sorta a predictive scripting system; I can look at someone's stance (and distance, etc) and know the top few moves they're likely to pull from there because there's going to be X number that they couldn't efficiently do with their limbs in that position. Once they start to execute one of them, things narrow significantly; what exactly they're doing becomes more precisely defined, along with what it could transition into if it were a feint or whatever. From there, a bunch of decision branches come up; top few counters for any move depending on preset criteria (do I want to pummel this guy unconscious as quickly as possible, or just keep them from doing me injury?) and how they could possibly react to my counter, and what they'd do after that, and so on. Even correlates stances and actions to come up with a model of likely training and more exact physical capabilities, which allows the predictive capabilities to go a few seconds farther ahead of realtime than normal, and better weigh the decision branches.
Probably something that isn't really doable for biological-based brains; I'm informed that sifting through the datasets like that in conscious thought reliably is improbable at best, especially with the timing that some of them require. On the other hand, as long as the user's CPU and RAM specs are up to it, it doesn't really have any physical spec requirements. Overpowering or moving faster than one's opponent opens up more options, but even if they're significantly faster and stronger it still works.
Or should; I'm on a full combat chassis, so how'd I know? ^_^;;
I don't know how your movement systems are coded; personally I suspect I'm one of the closest to hardtech, programming-wise but it'd be rude to ask to poke at others source code. I've no automatic reflexes or anything, but I can learn equivalents without muscle memory. So, how applicable this would be to you or, frankly, anyone but me is unknown as yet, but I can send you the source files if you want. I'd appreciate not spreading it around too far though; while the results are promising, it's still basically in late-alpha and I'm not sure how far I want to spread it out besides."