Golden Sun 1 any% clear TAS without Retreat Glitch. Used Sanctum Return to shorten the walking.
Any% Clear without Retreat Glitch
Emulator used: VBA v23.6 svn480
Clear the game without Retreat Glitch
Final Time is 777305 frames, 3:35:55 with 13841 re-records. Current RTA world record is 2:46:48+38 min intro with Retreat Glitch By Tl_Plexa. (No intro run described by him in [dead link removed])
feos: That's a tricky case. Conceptually, the "no retreat glitch" category is perfectly valid. Here it was executed at an average level (TAS wise), and it might be acceptable, if not the following:
This category isn't vaultable, so if the reception was good, it could be mooned, and then, after the "retreat glitch" version is done ultimately, it might have been obsoleted by it, if content is mostly similar (that can happen for different branches). Or they could both remain published, if different enough.
But right now, people don't have what to compare this run too, and also aren't sure they like this version. It very well can happen so that after the "retreat glitch" run is done (which doesn't need to be well received to be accepted, and which all hope to see) this branch is redone on a more professional level (if possible), and they finally like it!
Currently though, the numbers are:
1 user enjoyed the run.
2 users said it's an "okay" category, but didn't address the entertainment value.
6 showed interest towards the "retreat glitch" branch instead.
Meh votes overweight all the rest (in total, 59% support).
We can't say it's a well-received Moon run with such statistics. Rejecting for now.
1. What the heck is "Retreat Glitch"?
2. What exactly does "Retreat Glitch" do to warrent a different branch?
3. I recall another TAS that forgoes this being submitted here before, but none for "Retreat Glitch". Why is that so?
2. What exactly does "Retreat Glitch" do to warrent a different branch?
It's a wrong warp. It's the only major glitch relevant in a speedrun. Basically you can think of 'no retreat glitch' as 'no warp glitch' or 'low glitch' or whatever style of category.
I haven't watched the TAS, but it should be compared to tl_plexa's and werster's real time speedruns of golden sun, which are well routed and well optimized though for categories that allow for retreat glitch, so they're not 100% comparable.
I believe that was answered in the post above you.
@jlun2: For your answer to #3, you may be thinking about Golden Sun: The Lost Age (Part 2). The second game was submitted at least once or twice and both were rejected due to optimization issues, if memory serves.
I believe that was answered in the post above you.
@jlun2: For your answer to #3, you may be thinking about Golden Sun: The Lost Age (Part 2). The second game was submitted at least once or twice and both were rejected due to optimization issues, if memory serves.
Huh.
Anyways curious that an "no retreat glitch" would be submitted sooner than a shorter, glitched version.
I'm not going to vote since I haven't seen it yet, but 13841 rerecords for a 3 hour TAS along with a very brief submission note doesn't give me confidence that this doesn't also contain optimization issues. But I may be wrong.
^It's an RPG with relatively simple movement. Most rerecords will be in luck manipulation, with which a bot may have helped (and unless you are called MrWint bot rerecords don't count in the total). This sort of game can get away with a low rerecord density.
(and unless you are called MrWint bot rerecords don't count in the total).
What? Who says this?
That retreat glitch is pretty interesting. You’d think that if it failed, the flag or whatever wouldn’t be triggered to go to the first room…
I'm glad someone's interested in speedruning Golden Sun. However, I must say it's pointless not to use retreat glitch that would probably save a chunk of time.
Retreat glitch does involve wrong warp but it's not a kind of game-breaking glitch and far from "do this and suddenly you're in front of final boss". It can send the party in wrong place, but only within the same dungeon so its usage is pretty limited.
http://www.youtube.com/watch?v=4i6c6BfK_u0
This is a demonstration of retreat glitch used in the first dungeon by Kirkq. It skips some puzzles in the early part and forgoes dialogues, but you still have to go through the later part. Not using the glitch only prolongs the wandering and makes the run quite dull.
I didn't watch the movie yet so I didn't cast the vote, but I doubt the category is worthy. But I hope the author won't be let down by this and produce a better run in the future. Also, I hope it gets redone in English version as the run will be then more comprehensible to viewers.
Finally, I thank solarplex for making encodes for submissions and WIPs. I admire his dedication to the site.
From what I’ve seen the Retreat glitch is not fully understood and has more potential than how it’s currently used…
Edit: at 1:59:48 the movement looked bad… you went too far to the left and hit the wall while going up.
Also the submission text has a 1 too many in the rerecords count.
^It's an RPG with relatively simple movement. Most rerecords will be in luck manipulation, with which a bot may have helped (and unless you are called MrWint bot rerecords don't count in the total). This sort of game can get away with a low rerecord density.
As far as I know Golden Sun's RNG doesn't seem to care about menu movement or input times (just your actions, what gets hit) and a human on actual hardware can manipulate it to (at the very least) get rare drops consistently so brute forcing the RNG with a bot is unneeded. There may be another RNG at work though.
With one exception - some rooms have environmental elements (like a swaying berry on a tree) that advance the RNG every frame. In these rooms, realtime runners are basically screwed, they can't manipulate the RNG.
In battle, I'm pretty sure you can do things like select an action to use then cancel it to advance the RNG.
From what I’ve seen the Retreat glitch is not fully understood and has more potential than how it’s currently used…
This is not really true. Pretty much every normal door in the game has been checked for where it leads to with a "normal" retreat glitch. "Save" retreat glitching involves doing a similar thing and saving to translate your coordinates to the first floor of the dungeon to go off map. Once off map, after underflowing your coordinates by walking off the top or left there are various glitched doors that lead to various rooms in the dungeon. I made a brute forcing script to record numerous instances of these glitched doors (literally thousands), where they are, and where they lead. Each dungeon typically has like 5-10 options. All of my notes were available to werster (he also helped run the script in various dungeons) a couple years ago, so I can only assume they've for the most part been implemented into the real time run where applicable. In short the glitch may not be well understood, but the effects have been much more rigorously tested than your post speculates.
ALAKTORN wrote:
Edit: at 1:59:48 the movement looked bad… you went too far to the left and hit the wall while going up.
That's probably like 4 frames bad. If that's the worst thing you found I'm inclined to believe the movement is pretty okay overall.
-
No retreat glitch is an okay category in my opinion. It's not spectacular, but it's a pretty sensible way to play through the game. It's probably really roughly around 20-30 minutes longer than a normal run, but it does not skip key dungeons.
There are a few aspects of the RNG that are more random than others. I know that who gets attacked is much more random than which attack the enemy chooses. Real time runners are pretty familiar with the intricacies. It's also complicated to manipulate the normal RNG without starting from reset.
The movement in this game is pretty straightforward. 1 frame can be saved around most corners due to precise TAS level movement optimization, but other than that, just hitting directions, reloading, and frame advancing should get you within 1-2 frames of optimal.
As far as the optimality of this run, I would look at the battles, leveling, djinn, and equipment more than anything.
I glanced at the Saturos fight. Crystal powder wasn't manipulated from a random encounter in mercury lighthouse. (I can't guarantee it would be faster overall, but it could possibly save a turn.) The player can't manipulate eruption entirely without resetting. Some things could be done in previous battles to try and set up a minimal number of eruptions, but it would not save that much time.
He ends with 5/6 of each djinn type, which I really don't feel is optimal, but it's very hard to calculate the time gains/losses. Bosses are either weak to fire or water, and most major bosses are weak to water, but a summon from 4 earth and 4 jupiter isn't bad at all. Djinni also make your characters stronger when not in summon mode. My best guess is probably 4/4/4/4 is optimal, but that's mostly speculation. Adding more would save time in battles and lose time acquiring them. I'm not sure if it's faster or not.
I can't really say for certain how good this run is, but for what it is, it appears to be an average level execution (or maybe slightly below average, I'd have to look closer). Things seem to be done for sensible reasons, the characters are probably a little stronger than they minimally need to be.
I did not watch the whole run, I just glanced at notable parts.
EDIT: I'm wondering if sacred feathers would've been a good investment. They double distance (EDIT: maybe my memory is bad?) to next encounter for a while. Most encounters are run from pretty quickly though, so I'm not sure. They take some time to buy and some time to use as well.
As this run isn't vaultable it'd need to be well received, and I see little to no enthusiasm about this goal. Add average level of execution (the author would probably need to do a couple more TASes to get the idea), and see this run rejected, with hopes the one with the retreat glitch is done some day.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
As this run isn't vaultable it'd need to be well received, and I see little to no enthusiasm about this goal. Add average level of execution (the author would probably need to do a couple more TASes to get the idea), and see this run rejected, with hopes the one with the retreat glitch is done some day.
Having been really been into golden sun once, i really liked this run (especially how all the bosses were dominated), but it would have been impenetrable and boring if i had not played the game, Partly because it was on the Japanese version. Was there a reason for this given that time saved due to faster Japanese text scrolling is not an improvement? There's no rule against using the Japanese version, but it cuts down on entertainment(unless the japanese version has additional glitches, which is not true here). Also, i was confused about things like the the rational for specific djin being taken, or what was equipped at any given time. A longer submission text would have helped this. It's too bad that this isn't a vaultable catagory, as this run would have a sizable niche audience.
