Final Grounds NG 2lap

[arctic trench]

I really love this track an always wanted to make a TAS of it, although flurry does have a WIP, Cesar (Darkz)'s NU TAS is much more optimized, and optimized it is, but not perfect. There was just barely enough timesave to be able to get a full frame ahead of Cesar by the time you get to the QM heavy hairpin turn, which allowed me to not just "resync" the QM (I used AGC but never copy-pasted inputs), but to end up even further than a frame ahead by the time you get the the boost panel. I'm now working on the shroomless sc. If anyone want's to collab, i guess you could just dm me (i don't want more that 2 other people involved if any that is)

kinda want a lil chat but not a huge discussion @bright pilot

cound't really find a tag to fit that niche

removed the updates only tag

[bright pilot]

alright

[arctic trench]

I think this will be good agc practice

[lusty burrow]

nice

[arctic trench]

Fastest shortcut I was able to get (note the ghost is delayed by one frame)

@lusty burrow would you like to help me with the shortcut here

[lusty burrow]

sure thing

why not

[arctic trench]

I’ll send the RKG as soon as I can (not for like 30min)

[arctic trench]

Thanks @lusty burrow we were able to improve on the NU TAS by over 4 frames total

(racing the NU TAS)

[pearl shadow]

am i to blame for the testicles mii name lmao

surely my stupid meme wasnt that important to your brain

i hope not but if that's the case i'm sorry

jokes aside glad to see this being improved on and cool to see césar's also helping out here as well

[arctic trench]

when I thought about the name Tactical, I also thought about the name Testicles

I don't really think of you when that joke comes up

[arctic trench]

Was finally able to get the shortcut, used agc to carry the 4f improvement at the beginning to around where the NU cut is. Showcased a potential 1-2f timesave a little after the ss state ejection, but I didn't bother to optimize it so I lost the time. Maybe more improvements are possible but idk yet (ghost is the NU TAS delayed with agc by 4 frames)

[arctic trench]

@lusty burrow do you plan to keep working on this?

[lusty burrow]

yes

but i'll probably don't work actively

cause of world

[arctic trench]

was able to carry the 1f improvement from earlier, now we gotta figure out the sc (ghost is NU with a 4 frame delay)

[arctic trench]

New shortcut strat (idk if it's faster)

[gritty crown]

you have to cancel your wheelie so idk

[arctic trench]

you would have to do that normally I think

where is the wheelie cancel?

[gritty crown]

when you go through the one way wall

if you do it normally you can just keep a wheelie from cut

[arctic trench]

my cut still gets an incredible line so idk

and an mt before the turn

[whole quiver]

Reo drift looks entirely pointless

You need to hold the mt and rotate the opposite way

Getting a chain drift from the clip might be useful though

[arctic trench]

New Reo driftless sc strat (saves ~5f/lap), I did the spiral after to the best of my ability and I did the up and down ramps the way I thought was best. I also used agc to help compare my slow ramps to cesar's slow ramps (ghost is NU with a -605f delay)

[quasi falcon]

did you try using luke's rule to make the first drift softer

doesn't look like you used it

[arctic trench]

I'll test it rq

saved 1f

just gotta resync it rq

[quasi falcon]

and if you remove the wheelie at the chain drift do you still get the drift

[arctic trench]

I get the drift, but It wasn't faster because it didn't turn as tight

wait, which chain drift

[quasi falcon]

at 47.53

also here did you try missing the wall and just starting a regular softdrit much later

[arctic trench]

yeah I tried that too, it was slower

[arctic trench]

wasn't able to resync anything but I was able to maintain the timesave

sending video soon

[gritty crown]

you don't fully accel from the mt

[arctic trench]

ikik, I couldn't ever get fully acceled at any point, no matter what I tried, there was air pockets all over, I got the most speed that I was able to get and used that

[quasi falcon]

you could try neutral gliding off the edge of the cliff (after the 5th frame of airtime) to change QM

[arctic trench]

Doing that after 5f of airtime gives me too bad of a line, but doing it before 5f of airtime gives me a tighter line, and the ability to have a higher speed wheelie, saving 2f

I also managed to get a ridiculous spiral, saving another frame

[quasi falcon]

Nice

[quasi falcon]

Would you save time by waiting until you reach 84 IV to wheelie

After the shortcut landing

[arctic trench]

I get a unavoidable pocked of air when landing which means I would reach 84 IV much later, loosing time

[whole quiver]

Is it not faster to get an mt upon landing

[arctic trench]

it's faster by a few frames

i don't feel like recording rn tho

[arctic trench]

some nice slow ramps I got

[quasi falcon]

I assumed you already tested that so I didn't mention it

[arctic trench]

I did test it, assumed it would be slower when testing since it because I would have to hold the mt for much longer than the clipdrift (seeing the immediate breakaway the clipdrift had for a bit made me stop testing), but that meant I would have 96.61 IV for the entirety of the straight, making it 2-3f faster

[arctic trench]

with cesar's help, we were able to improve the cut and spiral by ~1.3f. When recreating the slow ramps, I was able to get the best second slow ramp I've ever seen

(racing one of cesar's ghosts)

[gritty crown]

I feel like any more angle to the right would save a lot of time

[quasi falcon]

or getting a left bounce upon landing and doing a reo drift to the right

idk I'm just also unconvinced that the turn before the cut is the correct approach

[small creek]

i would have never thought to do a full slow charge drift there

[whole quiver]

^

Taking the drift into the sc earlier too

[arctic trench]

Everyone is saying so much things, what should I work on next

I’m gonna go insane if I have to redo the sc and the spiral again

[thin crow]

you should work on whatever you want to. you've already improved over the old run and if you're satisfied with how it looks (and don't want to redo it) then that's alright

[quasi falcon]

i don't recommend spending hours tasing sections that are like 10 seconds past the part that you're up to

[thin crow]

that's my personal favorite thing to do ..

[quasi falcon]

until you're absolutely certain that you've tested everything

[quasi falcon]

bad habit and waste of time unfortunately

i used to do it until i snapped myself out of it

now my tasing is far more efficient

resist the urge to carry your improvement as far as possible, switch the player to the ghost, and keep trying to save more time

[thin crow]

[quasi falcon]

unless of course the next turns actually heavily depend on the one you're working on

but that's not really the case here since you have a long straight

[thin crow]

but also if it's a CT just do enough to make it look ok 👍

[quasi falcon]

when you get more experience tasing you'll naturally think of more ways to approach turns

[arctic trench]

I keep hearing “do this”, “do that”, and that messes with me and it makes me think that my run isn’t optimized enough, I think it’s because of that kc run but idk

[quasi falcon]

nobody really cares about optimisation on CTs as ellu said

but i think people should at least strive to have the best strats for every turn

it's a weird, complicated turn - people are going to have a bunch of ideas on how to take it. the suggestions shouldn't be taken personally

ultimately you can either test the ideas or not test them, nobody's forcing you to do anything

you can make the executive decision to move on from the section if you want

it's your tas

[arctic trench]

Ok

ig I’ll move on

[quasi falcon]

strat testing is fun

i enjoy trying different ideas personally

plus, the more ideas you listen to, the more methods you have in your mental toolkit that can come in handy when making future TASes
e.g.
"hmm.. this turn is quite similar to the one I did in X track. the best strategy for that one ended up being Y strat/inputs, so maybe I'll test those here"

[arctic trench]

Ig the bc spiral (on a IDB slightly lower than flame runner) is similar to the spiral in final grounds. I did make that connection on how I approached the turn.

For the sc though

I had never done something like that

So I had to rely on suggestions until I went insane

[quasi falcon]

I'll give you an example of thought process

[quasi falcon]

take this suggestion

you might be thinking, from the moment of the below image onwards, I'm already going as far left as I can

you've done a full softdrift with luke's rule, then a hard chain drift left, and also neutral gliding

yet, we want a wheelie off the ledge that's more like the blue and not the white

so how can we do this?

a potential solution could be to shift the entire sequence of the previous section to begin earlier

which could involve hopping much earlier around here

and that might also even involve modifying the previous line slightly

to get more left rotation (yaw) at the hop, so that you don't turn as tight during the first drift

[arctic trench]

That goes back very far for me, since unlike my mii name, I have a life, and won’t dedicate it completely to TASing, so unless you or any other people suggesting ideas were contributing, I won’t backtrack that far

[quasi falcon]

No need to get hostile with me, I was just providing an example of what mental strat development might look like

I don't dedicate my life to TASing and yet I don't mind backtracking to test ideas. Patience is worthy when making runs

And I will also point out that you shouldn't complain people aren't contributing when the thread has the "invite only" tag

[arctic trench]

Oh, sorry If I was hostile, and the contribute thing I was talking about an example project with multiple contributors already

[arctic trench]

cooked up a prelim run to calculate a time estimate, L1 (with 1 shroom) was a little more than 1 second faster than flurry's lap 2 (with 1 shroom), it was also faster than flurry's 2 shroom lap 1 by ~0.3 seconds, I think the final time will be around 2:43.500

not feeling like recording it tho

[arctic trench]

slightly more optimized lap 1

[brave bronze]

congrats

i assume now the job is to sync lap 2 unless there more optimzations to be made

[arctic trench]

there are more optimizations I think

[arctic trench]

I found a 2ms optimization, but it can't recreate the qm on the last downwards slope

I'll just call the lap 1 completed

on to lap 2

[arctic trench]

yo @lusty burrow do you mind helping me resync this

[lusty burrow]

i am not an expert of resyncing

but i can try

[arctic trench]

me neither

that's why i need help

I'll send you my closest attempt

[bright pilot]

have either of you read the agc tutorial doc

that was a huge help for me learning to resync

you basically just want to try and match your positions and angles as closely to the ghost as you can, and then guide it along the same path it took on the previous lap

the closer you match the position and angle, the easier it is to guide it, since the inputs you used on the previous lap will sync for longer

you should have a fairly easy time doing this since there's nothing particularly precise going on - i had it tough trying to resync a supergrind and ejection on my first day using agc lmao

[lusty burrow]

i did yes

don't you remember me syncing volcanic

[thin crow]

i personally use agc as a reference for my ghost on subsequent laps

i dont usually go all out on lap 1 optimization so i tend to find slightly faster strats/driving on l2/l3 etc.

this is fine for ct tases and i think its nice to have some variation across laps. so personally i just use agc to sync short sections but i will also detach and race and improve more (like using tas code normally)

@arctic trench if for instance you wanted to try the cut again or a diff approach to the driving on l2 this might be a good strategy there

[arctic trench]

I read the doc a while ago, it is a lot harder than it sounds

[arctic trench]

I feel like I went all out other than the sc, so I would at least want to resync before the sc

[thin crow]

you can also just like drive and try to keep pace with the ghost if its too hard to make the inputs exactly the same 🤷

but yeah you only really have to match position/angle once somewhere

[bright pilot]

yeah the only time i would actually recommend full agc resyncing on a tas is if there's a really precise strat you don't want to have to re-tas or if your optimisation standard is super high

e.g. i did it for rmr because i didnt want to have to spend days or weeks matching the really well optimised driving and precise strats

[bright pilot]

wdym

[thin crow]

he used agc for volcanic skyway

[bright pilot]

brain comfused

[arctic trench]

Yeah, there is multiple strats I don’t wanna re tas

[arctic trench]

Finally managed to resync everything, I got a 2:43.148 (optimized the end to submilliseconds)

that is also 0.352 seconds faster than my estimate

[brave bronze]

[arctic trench]

found a 1f timesave (suggested by Cesar) on the final shroom

[arctic trench]

final time reveal

[brave bronze]

how the hell did u save so much u were stuggling to get 147 just last night

[arctic trench]

^

[bright pilot]

What was the previous BKT time

[arctic trench]

2:45.816 by flurry

[whole quiver]

Nice timecut

[arctic trench]

encode test

Actually the outline on the name display is better with an 8px outline

[bright pilot]

You know you can just encode a single frame, right?

[steep leaf]

this isnt aligned btw, put some more < after the IV text

[arctic trench]

ik, just dont feel like changing the config file rn

[bright pilot]

That option is literally made so that you can make quick tests and not have to wait a couple minutes to encode

[steep leaf]

you can also cancel the script w/ ctrl c, I do that most of the time if I want to see it in motion

[arctic trench]

also rq before i record

Use testicles mii
react ✅ for yes
react ❎ for no

if you say no I'll use the ihavnolife one

ig im using the testicles mii then

framedumping now, get ready for this to take hours lol

[arctic trench]

finished dumping, time for encoding

[whole quiver]

“One like and I’ll do it”
“Say no more”

[arctic trench]

just finished encoding

it looks beautiful

[brave bronze]

ev speedometer on a ng tas :ICANT:

[whole quiver]

Need to know how much gravity there is when falling

[arctic trench]

why not

thumbnail

[gritty crown]

gcn rr

[arctic trench]

ct tas of the year

[arctic trench]

https://youtu.be/XNvOt-X9njg done

[bright pilot]

@arctic trench just btw you dont have to list authors in #tas-showcase

thats only a rule for #bkt-feed

[arctic trench]

Why not

[bright pilot]

.