Experienced Forum User, Published Author, Player
(8)
Joined: 1/27/2019
Posts: 11
Location: The Netherlands
DemonStrate wrote:
Did you look into fixing the extra log push? You push the left log down at 1:11:54 instead of just pushing the right log to the right, going under the left log, pushing the left log up, and then pushing the bottom log down. Meaning, the left log only needs to go up, instead of down and then up.
Fx wrote:
At 1:11:53, this movie makes a mistake that we have already fixed in the submission. We've saved around 2.7 seconds due to this. Other suggested alterations were either not timesaves for TAS, or impractical as they would cause large portions of the game to change with them, which simply isn't worth the effort as it would take weeks to fix and save a couple frames at best.
Experienced Forum User, Published Author, Player
(8)
Joined: 1/27/2019
Posts: 11
Location: The Netherlands
deuxhero wrote:
Seems fine so far, but a bit on the boring side like most jRPG speedruns.
I know the TASvideos community isn't that big on JRPG, which is fine. We basically went into this expecting it's going to be Vault no matter how good the run is. For anyone else considering watching, I'd recommend you skip ahead to 44:00 to not sit through the intro. There's nothing to see there other than minor optimizations and cutscene mashing.
deuxhero wrote:
How much would an "All Djinn" or "Perfect Password" (All Djinn plus the three triggers for the link events) differ?
Route-wise it'd be unrecognizable. First of all we'd need to collect Jupiter Djinn, but some dungeons we skip entirely would have to be played through at least partially. There would be a couple unique-looking glitches that have a basis in the ones already used. Assassin Blade would be picked up again because it'll oneshot every djinni and every bonus boss.
All Bronze Transferrables (what you call perfect password) wouldn't only do the event triggers, but also requires you collect all psynergy items (Orb of Force, Lifting Gem, Halt Gem, Cloak Ball would be the extras). This means you have to play through certain dungeons in full.
From a viewer perspective, it wouldn't add much to the run. Isaac goes on a big fetch-quest for 1 extra hour, then obliterates S&M+FD because 28 djinn KEKW
Most interesting part would be Deadbeard. I say interesting but it's really going to be some 4-turn boss battle at best where his actions are manipulated to be some trash like Impact or Atk. Late-game encounter manip strats wouldn't be necessary anymore as we'd be much higher level and therefor can just run from everything on the first turn like we do up to Suhalla Gate in the Any% run.
Experienced Forum User, Published Author, Player
(8)
Joined: 1/27/2019
Posts: 11
Location: The Netherlands
File that includes the post-credits cutscene:
http://tasvideos.org/userfiles/download/69611605441251371Link to video
Commentary on the TAS by me, and two of the main speedrunners of the game.
At 1:11:53, this movie makes a mistake that we have already fixed in the submission. We've saved around 2.7 seconds due to this. Other suggested alterations were either not timesaves for TAS, or impractical as they would cause large portions of the game to change with them, which simply isn't worth the effort as it would take weeks to fix and save a couple frames at best.
On the topic of timesave:
Is there time to be saved? Short answer: yes.
Long answer: it's complicated. The way lag frames present in this game is something we do not fully understand, but they impact many parts of the run and it's not unthinkable that more optimal placement of the lag frames would produce a faster movie.
Other than that, there are some theoretical timesaves we could get but couldn't make work in this movie. This amounts to a couple seconds to save.
Even further optimization of RNG manip, most likely through writing a script or bot that will brute-force the optimal path forward, could yield more timesave, in the range of 10-30 seconds (unless something major is found).
Experienced Forum User, Published Author, Player
(8)
Joined: 1/27/2019
Posts: 11
Location: The Netherlands
arflech wrote:
At about the 0:01:50 mark in your commentary, you said, in essence, that your movement down the stairs was un-optimized; is it reasonable to go back, re-optimize that movement, and continue improving the TAS from there?
Finding out about sub-optimal movement halfway through the game means you can know about it and not want to go back to fix it.
Early on in the run there are small timesaves like that because of inexperience. There is another timesave of 1 frame per random encounter for a decent portion of the game as well. Things add up to about 15-20 seconds total over the course of the run if I had to hazard a guess.
That's not counting improved menus and re-routes now that we know what a good TAS of this game is supposed to look like. The previous TAS just did things "sort of quickly I guess" without really trying to make use of what a TAS can do in this game, while my TAS ends up really pushing the boundaries of RNG manipulation.
With this route complete, you can instantly start looking at potential re-routes and timesaves with optimizing menus further in advance. When creating a new route you just kind of do menus as you go because you need new things each fight, while a complete route lets you know in advance what you're working towards and so you can do menus differently (faster) earlier on to set up for later.
So yeah, there is timesave. No, I'm not going back to fix it anytime soon. This run is miles ahead of the previous time listed on this website, and I'm happy with it. If there was an easy way to fix execution errors like that without desyncing the entire run, I would've probably gone and fixed it, but it's not that simple.
Experienced Forum User, Published Author, Player
(8)
Joined: 1/27/2019
Posts: 11
Location: The Netherlands
Spikestuff wrote:
Please avoid using preformatted text on your next edit.
I've cleaned it up from having it and formatted to what looked to how it was intended by you.
Edit: I wanted to include a 2nd movie file that performs these couple of inputs as well, but could only upload one file.
You can upload your input here and post it back here: http://tasvideos.org/userfiles/my#uploadfile
There's a bit of an issue because of this now. We're currently comparing early input end to the full movie with all input.
I am curious on how much time is saved when the full input is provided to the party, for reference here's the previous:
[2893] GBA Golden Sun by kien_ in 3:12:47.59
But I believe you clock in about 10 minutes faster if I'm not mistaken compared to the previous movie.
---
The author being a runner of Golden Sun does have a YouTube and Twitch account so an encode might be provided from them down the line so just wait on the author for their encode and even the possibility of commentary due to being a talkative runner.
http://tasvideos.org/userfiles/info/52841319963352038
Thanks for the clean-up, I was wondering what happened when I first submitted everything. Looked pretty ugly.
As for the timing thing, that's just where timing ends for the run in recent times.
I wanted to add that the previous TAS does not use a BIOS file, but Hard Resets EVERYWHERE, ALL THE TIME. Not using a BIOS saves him tons of time illegitimately.
That's not why I chose this though. The reason I went with this timing instead of the previous TAS'ers timing is that the rules from this website suggested that when the end has clearly been reached, but afterwards requires minimal, simplistic inputs, those inputs don't have to be counted.
I thought it looks nicer, is less tedious to watch, is consistent with RTA, and also has a certain appeal (sub 3:00 run)
What I'm referring to:
"Some games acknowledge the game has been completed, and then require additional input to show the rest of the ending sequence. For the purposes of speed records, you may stop the movie when the initial ending sequence is guaranteed to begin, without including extra input to process the entire ending. This is conditioned on all the following being met (full movies are still preferred):
Extra input is trivial to execute for anyone replaying the movie
Extra input does not change the resulting ending "
If that's not the case for this game then the above file can be used. It pressed A a couple times. And I think 1 ''Left'' input. Time should be about 10 minutes ahead of previous TAS in that case, as you mentioned.
Either way, I've added this explanation to the original submission file as well.
