Denver, CO, USAI was watching Niss this morning and noticed a peculiarity. During transitions/kid switches between dark areas, the game seemed to load faster. This is probably due to not having to actually render all the obstacles. Most room transitions turn into 40-60 frame savings which should add up quite a bit. Of course the drawback is having to play in the dark haha
Here's a not at all optimized WIP for discussion or suggestions.
To add to Ness' awesome research, the #3 option which is shown in the video (which may be desirable in some situations that I may or may not be routing out currently) is around a 2 frame window, and it's got a bit of a delay to it so just going ham is probably not gonna hit #3, but luckily #4 is a huge window and is probably more RTA friendly than what I'm working on.
I broke stuff again.
Basically using enough new kid walks delays Ed's movements (and timers and stuff) so you can encounter him before the security room. If he's "in" the GT room when you go in he can't actually spawn there though he can move in from the red hallway. For some reason this bugs out GT (but Ed will finish his walking to the door to answer the bell.)
I didn't do it in the video but you can new kid walk through the GT room to reduce lag.
Trying to figure out how to best make this useful since he answers the door and that costs some of the time you saved.
This is on topic for once. I found an even faster Fred Skip.
Basically you throw the switch and run back out the door. Do a new kid storage as the lab screen is loading (select then A on new kid, then Start at the right time.) Fred's cutscene starts faster than doing the new kid to someone else and then back to Dave. This saves around 2s.
I poked around with the prototype today. Outside of the stuff that Nintendo made them remove, the game appears identical even down to the bugs and glitches. It's basically the same exact game but with more visible boobs.
Since I've been turning this thread into a vod archive. Here's a TASed 5:35 without a lot of optimization using Edna juke and opening the valve as early as possible. I actually have to do that new kid walk toward the gargoyle or else I can't close the valve in time. It's that tight. This could easily creep towards the 5:30 mark but I want to share the ideas as maybe someone else can build off it.
I was messing around with stuff today and found that you can pre-store a New Kid command and activate it in certain situations. Could be useful for setting up dungeon skips ahead of time, plus it can be used during cutscenes.
Basically during a screen transition (this can also be between most doors/rooms) or a regular new kid switch you would hit select, then press A on New Kid while the screen is fading out. If done right, your next A press (or start during cutscenes) will immediately bring up the New Kid menu interrupting whatever is happening. I believe this works by queueing the "GET" command but overwriting it with New Kid, which the game just lets you do in most situations for some reason.
I'm not as familiar with meteor police % but could this be used to regain control in Fred's lab as the little police dude is running through it? Edit - in this video there was already a capture cutscene involving purple tentacle to bypass the new kid lock out in the lab.
While we're at it here's a 5:39 RTA equivalent TAS with a new route that is probably not RTA viable. Don't try this at home kids!
Threw together a theory TAS (some inputs were borrowed from Arc's published TAS on TASvideos) and it clocks in around 6:09 by RTA timing. Not sure if the route is optimal but I'll try a couple things today.
The reduced lag is significant, and I feel like moving the capture to the beginning of the Lab sequence lets us multitask a little bit more beforehand. I'm working on TASing stuff out to see how it all shakes out in the end but I have a feeling this is going to lead to a ridiculous time. I'm glad I got to see it live!
I don't value my free time whatsoever so I did a human theory TAS of Cyberball. Clocked in at 8 minutes, but 8.5-9 minutes is a more realistic goal. There may be optimizations that I missed as well but this should provide a good blueprint for the overall route.
Advanced mode against the Pro team lets you burn more time off the clock per play, so the objective is to use every down. Defuse the ball first time it's critical, then intentionally blow up on the 1-2 yard line on the second time the ball goes critical. This pins the AI team into safety position which seems to work more often than not when they're in that spot. Call the Prevent defense to cover them downfield and the QB just kinda stands there. Once you get the safety, run out the clock as you've already won.
However, be careful that your QB doesn't get hit too many times, else you'll turn it over when he goes kaboom. Incomplete passes or running out of bounds don't stop the clock in this game. Just watch out for interceptions.
Have fun!
I was screwing around and made a TAS for this while looking for extreme early ending input runs. Turns out the advance mode is better than the rookie mode for being able to drain much more time per play when you're on offense and should drive down the RTA time quite a bit.
Hello. Someone pointed out on my TAS of this that the best ending is actually considerably higher than we thought.
It's nearly 6 million.
I recently knocked out a TAS for this game ( ) and I may have solved the chopper dilemma.
Memory value 0x0019 displays the chopper's health starting at 18. (If anyone has video evidence of them getting 17 or 16 kills to beat the boss, please let me know.) I tried numerous things and it always started at 18. It decrements by 1 each time an enemy has killed, but not all kills count.
It seems like if you hit the enemy twice (either from a standalone double hit or like, kick them in mid air then hit them again on the ground) that it doesn't decrement the counter for that kill. This was consistent across all my tests.
Janky game is janky.
From what I understand about this game, the AI has a pool of moves to choose from per move. There's a rolling timer byte that dictates the move that is chosen. The timer for this does not advance when you're holding A (among other situations) even if it's not your move.
For whatever reason Ke7 is in the pool of moves after e4 but not after e3 so I don't think a faster move route will be possible
