Author Topic: PD Strategy Discussion Topic  (Read 16074 times)

Funky_Buddha

  • Posts: 157
  • ...with hook and scourge...
    • GE
    • PD
    • twitch
Re: PD Strategy Discussion Topic
« Reply #150 on: November 16, 2018, 07:16:38 pm »
That's some cool and useful info Henrik, cheers. If it wasn't for the whole Elvis being slow to activate the portals thing I think it'd be ~5 seconds faster than the WR run (ie. the equivalent of 53-54 Elvises should be easy). But yeah, Elvis is apparently slow so it seems touch and go for usefulness - I suspect it'd make the level a lot better for anyone not going for the WR though. 

I'll play around with it in a few years when I get a nintendo again :kappa: Thanks again Henrik!

Whiteted

  • Posts: 126
  • AF1 51 or bust
    • GE
    • PD
Re: PD Strategy Discussion Topic
« Reply #151 on: November 18, 2018, 04:00:28 pm »
Maian SOS : single guard lure

First up (and something that can be used straight away once the timing is worked out) is pause buffering the far door as described in my facts thread post, and shown kinda poorly here https://www.youtube.com/watch?v=hum5o4MfluU (read the description)

But the big save will be if we can pull pace like this off:
https://youtu.be/rlvvbYjJaEI?t=11
I came up with the idea midway through testing my psy-strat out (get psychosis gun at a time loss of just over 1s, get a guard to warp and turn them back around to open the near door), so ignore the start.

The full idea is:
  • Start like Karl's 152.
  • Take a line closer to parallel with the glass, to fire the 8th shot by the wall as late as we possibly can, so as to lure only the 2nd guard.
  • Pace to the door, to get it open and alert the guards, generating lag.
  • The lured guard loads, takes a long 'jump' close to the other guard, who then gets in the way of their new target (the other side of the lab door). This delays them enough to unload before they open the door.
  • The guard keeps it's higher speed from when it was loaded, and goes like the clappers.

But the big issue is getting the lured guard stuck on the other. We need lag otherwise the guard just turns and misses him. If he opens the door then all is really lost, because rather than unloading with high speed he unloads with extremely low speed. He can also do a little triangular loop inside the lab sometimes, which usually leads to him opening the door and going extremely slowly again. Even reproducing it in emu isn't easy. Examples:
https://youtu.be/jBw-iAqJjWc (the very slow guard luretm)
https://www.youtube.com/watch?v=KtPqVnYIqRY (probably the fastest current strat, showing the triangle)
https://youtu.be/H9DKP6jdXuc (doing the triangle but still being fast)

Alternatively there may be some wizardry in the way I alert the other guard afterwards in that video, since I've not tried that again yet. It certainly looks like the first-lured guard teleports back to his pad, but that may be a hiccup in the PD map (it is brilliant but it is also a bit buggy).

Whiteted

  • Posts: 126
  • AF1 51 or bust
    • GE
    • PD
Re: PD Strategy Discussion Topic
« Reply #152 on: November 20, 2018, 05:40:25 pm »
Turns out we can get the guard to open the door, start moving and only then unload. This is achieved by pause buffering to update his position more often. The strat will be quite tricky but should be viable on console.
I'd expect this to make a sub 28 door-open possible on console, but there's a lot of details to work out. It's also nice because the lured guard will be in exactly the same place each run when we cross the yellow & black line. So if we find a good pause-unpausing pattern it'll work the same regularly.

This TAS & vid are very rough but show the idea.


Whiteted

  • Posts: 126
  • AF1 51 or bust
    • GE
    • PD
Re: PD Strategy Discussion Topic
« Reply #153 on: December 02, 2018, 06:04:29 pm »
Rescue agent investigation & Binary speed guards

I've been looking into the "lab-clothes guy" (LCG) on rescue agent & particularly how to make him more consistent of course.

The weird thing I noticed initially is just how far LCG has to travel. He starts in the showers where you find him on SA, and doesn't start moving until the cinema ends. The secret to his (rare) speed is that he bumps into the patrolling guard in the showers, and loads. Similar to the guard I was looking at on maian, this makes him significantly faster.




There are a few things that need to go well for this though.
  • Patrolling guard speed : seems truely random, 1/2 chance.
        Some guards are assigned a random speed out of a choice of 2, high/low, when the level loads. I'm pretty sure this is in the action scripts so I should be able to track down the instruction. If the patrolling shower guard (PSG) is assigned a high speed, then he ends up heading away from LCG, who can't overtake him, so LCG's speed is capped by him. There is no way to save this, even in TAS.
     
  • Minimal delay from bumping into the 2 guards : can pause buffer to help
        When the LCG bumps into the 2 guards in his way, he loads and moves towards where they are standing. He will often repeatedly try to get at the spot where the other guard is standing, wasting a lot of time, rather than continuing with his path. Pause-buffering these sections helps a lot. This coincides with when Joanna is in the lifts, so there's a lot of potential :)
     
  • LCG super-speed after bumping into the first guard : seems random?
        When LCG unloads again (green in image), his speed is greatly increased. The problem is it seems to vary, and we need a speed close to the high end (TAS below hits the fastest speed I've seen, though not too uncommon, which makes up for a couple of seconds that it loses near the start).

 
 
As per usual, I haven't worked out many of the details, but I would advise pause-buffering the start of the first lift, as in my TAS below. The collision with the 2nd guard depends on the 2nd guard's speed, and will be roughly at the end of the 2nd lift, so for now atleast I'd say don't do it.

TAS    https://www.youtube.com/watch?v=64w9cGIiOaM

Fundamentally though this WR pace asks a lot from LCG. I think even in TAS he can't get to the end of his path in much less than 34 seconds.

PS: There is a clean setup ... scratch that it didn't work.
PPS: wouter pointed out to me that we have x-ray vision! May be very useful for manipulating guards and/or seeing if we succeeded.

Edit: X-ray vision strats would have been so cooked but unfortunately it can only hinder really on WR pace. The only place I can see it being useful for WR pace is if we can make Jo faster (by lucky lifts or box-boost shenanigans) so that we leave the lift before 29s. Alternatively it might be pretty good for consistency on WR+2 or slower runs. Idea is:
  • Equip the x-ray glasses in the bottom of the second lift.
  • Left strafe around the catwalk, looking slightly above horizontal. This loads the main corridor, even if you are pressed against the bannister
  • Throw and shoot dragon (easier with x-ray vision?)


Also a small note, there's no need to equip the disguise until just before the door guard's 2nd line of speech, if we can cope with the long room undisguised? Isn't keeping full speed even more key on PD?
« Last Edit: December 03, 2018, 03:50:54 pm by Whiteted »

Whiteted

  • Posts: 126
  • AF1 51 or bust
    • GE
    • PD
Re: PD Strategy Discussion Topic
« Reply #154 on: December 07, 2018, 05:47:08 pm »
Attack ship ending - 5x luck or more - stay "out" of the door

Alka sums it up well:
Quote
ALL of these runs have been affected by whether or not we did a thing we thought to be arbitrary

I while back I modelled & crunch some numbers for the attack ship ending. It gets complicated quickly, but good runs rely heavily on the spawn of the 2nd skedar that we kill (the clone). The final skedar's script rolls to spawn this skedar each time that it runs.
But as mentioned in my facts topic post, how often a 'guard's script runs depends on where you are and which way you are looking.

If you go too far forward when blocking the door, the frequency that the script runs plummets from ~15hz to 2hz. This ruins my model and really ruins your odds of a good ending. The solution is to only just block the door from closing, as shown below. It's dead time so the strat is really easy.


For each of these vids, the bar on the right is the time between the last 2 times the skedar's script ran.

Just-in the door strat.

Going too far in
Vid to compare to (mimicing most of the high point runs)


Back when I did the maths I had no understanding of the game so I'm going to investigate & crunch the numbers again and present them in an approachable way. This find should explain most of the discrepancy between my model and reality - i.e. Boss' 2:08 is probably a lot luckier than 1 in 12.

Wyst3r

  • Posts: 4164
  • Train Strat Master
    • Henrik
    • GE
    • PD
    • twitch
Re: PD Strategy Discussion Topic
« Reply #155 on: December 08, 2018, 12:22:20 am »
Lol how the hell is the difference that large. If anything, you’d expect being inside the room to be less laggy. Any clue what’s taking up so much processing power?

KevinDDR

  • Posts: 31
    • PD
    • twitch
Re: PD Strategy Discussion Topic
« Reply #156 on: December 08, 2018, 12:53:46 am »
I believe if I understand correctly that the difference is that large not because of the lag but because going "into" the first Skedar room unloads the portal to the other Skedar room, causing the Skedar that matter (the last one) to go into the "kind of unloaded and not running the script often" mode. Going into the room probably does help with the frame rate, but because it causes the Skedar to switch "modes" it still hurts in the end.

Wyst3r

  • Posts: 4164
  • Train Strat Master
    • Henrik
    • GE
    • PD
    • twitch
Re: PD Strategy Discussion Topic
« Reply #157 on: December 08, 2018, 04:15:15 am »
Oh I see, for some reason I didn't realize the lag meter script had been modified. That makes alot more sense.

Great find.

Whiteted

  • Posts: 126
  • AF1 51 or bust
    • GE
    • PD
Re: PD Strategy Discussion Topic
« Reply #158 on: December 08, 2018, 10:21:07 am »
Oh sorry for the confusion, I forgot to say that I've stolen most of the code from your lag-o-meter Henrik :angel:

Here's the script (adapted from a script for the AF1 president), which looks for changes to the timer. It's atleast 90% your code:
https://github.com/whiteted-strats/Attack_ship_strat_dev/blob/master/right_skedar_heart-o-meter.lua

In my experience (i.e. AF1) I've found that positions are always updated each time the timer changes, but scripts aren't always run. I've just checked on PAL by duplicating the key loop in the skedar's script, and the script is running every time the timer changes, so the strat is looking good.