

A Way To Make Info Warfare Useful So Simple, It's Amazing.
#61
Posted 14 September 2015 - 03:10 PM
In the future the Hollander could be redesigned to where the cockpit would be right over the top of the main weapon allowing it to accurately fire on targets with that weapon without having to have the red dorito.
It would likely become fairly popular to mount a PPC/ERPPC/LL/ERLL on the energy hard-point immediately under the Thunderbolt's cockpit allowing for accurate long range poking without the red dorito.
The PPC/Energy nipples on the Awesome (8Q, 9M, 8V) could be moved so they're immediately below and to the sides of the cockpit potentially giving it the ability to hit other assaults (and some heavies) on the same section without having convergence.
#62
Posted 14 September 2015 - 03:13 PM
Plus it makes those target acquisition and lock duration quirks mean something.
I shall now include this idea amongst my many others in my periodic rants about how everything could be so much better.
#63
Posted 14 September 2015 - 03:22 PM
Hans Von Lohman, on 14 September 2015 - 02:48 PM, said:
I'm more inclined to make ECM disable IFF.

Hans Von Lohman, on 14 September 2015 - 02:48 PM, said:
I like this idea.
Hans Von Lohman, on 14 September 2015 - 02:55 PM, said:
If HSR becomes a problem when trying to hit the enemy without a lock-on, wouldn't that make people really, really want that lock on really, really badly?
If you want a lock on badly, then good and fast lock-ons provided by the new information warfare superiority is also something you want really badly as well.
Or is my logic flawed here?
Other than turning a bug into a feature, no.

1453 R, on 14 September 2015 - 03:09 PM, said:
Amen to that!
But, I am expecting someone to angrily chime in in 3 ... 2 ... 1
#64
Posted 14 September 2015 - 03:22 PM
Well,

#65
Posted 14 September 2015 - 03:24 PM
#66
Posted 14 September 2015 - 03:26 PM
#68
Posted 14 September 2015 - 03:28 PM
Biggest problem is the range of skill levels in a match. I already see to many people at 90% health chain firing past their range that have zero idea that the target is not an auto locking feature. It is easy to berate these folks, but I was once their and was just as truly clueless. I would like any visible enemy not under ecm to show up as a triangle, but target lock would be needed for info and convergence with a non removable proc for new players (if targetable enemy is within cross hairs it blinks until button pushed - yes lowest common denominator but they are hurting other players too)
#69
Posted 14 September 2015 - 03:35 PM
Unlocked, no convergence mode:
o >-------< o
[where the o's are the left and right arm, and the X part of the crosshair is widened to represent the mech's torso width]
And it would morph into the current crosshair when targeting something
>o<
#70
Posted 14 September 2015 - 03:35 PM
Quote
I'd like to see ECM give an effective "distance modifier" to anything under it for the purposes of sensor lock, much like BAP increases the range you can detect a 'Mech at- ECM could effectively make a 'Mech "further away" for the purposes of potential locks.
#71
Posted 14 September 2015 - 03:39 PM
wanderer, on 14 September 2015 - 03:35 PM, said:
How about this:
ECM effects: Increase time to lock by
3 seconds @ <300 m
6 seconds @ 300-600m
12 seconds @ 600m+
Still gives some solid stealth, but not a magic box in a brawl.
edit: and it would be super awesome if all of these target locks came back on a cockpit screen as "Acquiring target data: XX% complete" and the percentage increased accordingly. But that's a wish list thing.
Edited by Big Tin Man, 14 September 2015 - 03:42 PM.
#72
Posted 14 September 2015 - 03:43 PM
#73
Posted 14 September 2015 - 03:46 PM
wanderer, on 14 September 2015 - 03:43 PM, said:
Doesn't a distance modifier do the exact same thing, with a variable time to lock dependent on distance as it stands now? And linking targeting with convergence fights hill humping/poptarting.
Either way, the 90m ecm bubble nerf should help a lot, if Paul pulls that trigger.
Edited by Big Tin Man, 14 September 2015 - 03:47 PM.
#74
Posted 14 September 2015 - 03:49 PM
Now OP can go and jump down a cliff he just reached the apex of his life, the internet agrees with him!!! Oh dear!!!!!!
#75
Posted 14 September 2015 - 04:00 PM
For example: a GRF-1S like this that wouldn't have to worry too much about convergence.
#76
Posted 14 September 2015 - 04:02 PM
Quote
This is why I prefer the default point be "maximum sensor range" or a similar fixed point. PGI's coding struggles with having weapons attempting to fire at multiple points of convergence, and I can imagine servers doing the equivalent of soiling themselves trying to deal with that guy firing 5 different lasers at once.
Quote
Either way, the 90m ecm bubble nerf should help a lot, if Paul pulls that trigger.
What adding time-to-lock does is means that the ECM-protected target has a window of safety from a lock-on convergence shot every time he pops out from cover, one he can reset simply by getting back into cover again. Even 3 extra seconds is enough to poke, fire, and retreat before the enemy sensors would be able to legitimately lock-on...whereupon terrain would break the lock for them. Rinse and repeat.
Adding "distance" means that closing on the target will eventually get them close enough to detect the ECM-protected target, at which point the parties exchange fire normally. With added time to lock, a good poker could defeat locks almost indefinitely.
Quote
I'd treat anything with a TAG designator on-target and within range as effectively at "zero" range as long as it's within TAG's 750m reach. Of course, the second you lose TAG, it reverts to "true" distance and breaks lock. But yes, TAG SHOULD be useful like this. (Heck,some Firebrands use them as targeting lasers as-is.)
Edited by wanderer, 14 September 2015 - 04:06 PM.
#77
Posted 14 September 2015 - 04:10 PM
wanderer, on 14 September 2015 - 04:02 PM, said:
Adding "distance" means that closing on the target will eventually get them close enough to detect the ECM-protected target, at which point the parties exchange fire normally. With added time to lock, a good poker could defeat locks almost indefinitely.
How would this be different than a non-ecm mech poking under your current idea? The mechs shooting the poker still require some amount of time to lock when they re-establish LOS.
#78
Posted 14 September 2015 - 04:10 PM
EDIT: Lets hope he sees it and forwards it to the right departement.
https://twitter.com/...577931143454720
Edited by Iqfish, 14 September 2015 - 04:13 PM.
#79
Posted 14 September 2015 - 04:15 PM
Quote
Unless someone else is providing steady targeting data, the poker is going to lose lock as well and have to re-establish it when they come out of cover. Especially with things like radar dep in the mix.
#80
Posted 14 September 2015 - 04:18 PM
wanderer, on 14 September 2015 - 04:02 PM, said:
I don't mean as a hard mechanic: I mean you can tell where the other lasers are going because they'll hit anything the TAG beam will hit due to being mounted so close together.
2 user(s) are reading this topic
0 members, 2 guests, 0 anonymous users