Posts for Hetfield90


1 2
5 6 7
12 13
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Alright I understand how it works now. In areas where lag isn't involved I can reliably see what is going to drop, but as you say the areas with lag(which is practically everywhere) the drops do not correspond with what is being shown in the script. Do you know how much it can get off by, or will it still be able to give me an general idea of around where upcoming drops are going to be when heavy lag is involved?
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Actually, it turns out that I was TASing new progress in read-only when it wasn't getting saved(the reason it always coincided with accidentally loading a state in read+write was because I thought I was switching from read+write to read-only when the inverse was actually happening). The teal frame counter looks like it is simply in read+write once it goes beyond the end of the movie in read-only, whereas most emulators continue to display it as "101/100 (movie finished)" or such. Currently it appears that there is no way of distinguishing whether you're in read-only mode after a completed movie or read+write.
Post subject: Bug with loading old savestates in read+write
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Using Dolphin 5.0, I noticed that sometimes when I accidentally load an older savestate in read+write instead of read-only, I lose all progress after that savestate without any way of getting it back. Loading a newer state in read+write and then loading any state in read-only after the one that caused the bug will simply load it in read+write, and loading any state from before the bug in read-only will not retain the progress made in newer savestates. Even loading a newer savestate in read+write and saving the dtm will not save any of the progress.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
So how exactly do I read it? Does each color correspond to a certain item drop, and the value that corresponds to what is being polled on the current frame is the value at the top of the right column? That's how I assumed it would work but I'm not noticing any sort of correlations like that.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Thank you very much for the help, FatRatKnight. Basically, the most important info we need is when ammo will drop from enemies, since with super arrow out there is little option for when you can kill enemies, so much of the manipulation has to be done far in advance. Which value should I look at value in which column should I be looking at to determine what type of drop would occur on the current frame? Also, I am getting this error when I try to run the script in FCEUX, which is what we're making the TAS on(right now I'm looking at the script in BizHawk). \MM5_RNG.lua:4: attempt to call field 'usememorydomain' (a nil value)
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Alright, thanks for the info. Using HLE now and no sync issues from switching over.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
18 seconds saved on Wanyudo. Link to video Do you fully understand how random encounters in this game work, Bernka? I am beginning to get the feeling that they use the kind of pseudo random distribution where their likelihood of occurring starts low and gradually approaches 100% each time you dodge an encounter, and then resets when one does occur. And I think it seems as though some encounters are manipulable on the screen they occur, whereas other have to be manipulated during the previous screen? Also, did you happen to find any pertinent ram addresses pertaining to the dynamic difficulty timer or random encounters? Additionally, I recall you saying something about blades/long blades being different between Momohime and Kisuke. By this did you mean that they use the different types of blades differently, or that they just have different blades available to them that merit using one over the other? For Momohime, at least, I have blades barely beating long blades of equal base damage in DPS.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
I'm about 13 minutes into a Wii Muramasa run I started making with DPS LLE because of outdated information in the other thread. Is there a way to alter the .dtm so it doesn't automatically change to LLE upon playing it to see if it at least still syncs on HLE? If it doesn't, should I continue the TAS or start over with HLE? (Bernka recently submitted a TAS of the same game using LLE).
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Thanks for the reminder, Angerfist. Luckily everything so far was still able to sync after switching from LLE to HLE. I anticipate both runs to be done at least before the end of the year.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Do you know what causes the glitch where one of Balore's hands stops working during the first phase? Or if there's any way to make it happen earlier in the fight?
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
After learning about this game from Bernka's Kisuke submission, I decided to look into improving the Momohime run. It turns out that the air dash slash causes the least hit lag of all the slashes, and even though it deals slightly less damage than the standard combo slashes, it still has a much lower ratio of hit lag caused per damage dealt. So the first 3 phases of each boss fight should use air dash exclusively in order to delay the dynamic difficulty timer as little as possible, unless other slashes are needed to make certain transition thresholds in time. During the final phase, it seems that aerial DPS beats ground DPS by a significant margin. Even though the sliding slash is much more powerful than any of the air slashes, it causes far too much hit lag and has far too long of an animation to be useful in situations where you can hit bosses in the air. Link to video This fight improves upon the published run's Blue Monk fight by 16-17 seconds. By canceling an air dash with a normal slash while close enough to the ground, landing will cancel the slash and allow you to start a new series of air dashes immediately after. This doesn't work on the 3rd air dash, so you can only do maximum of 2 dashes at a time this way. It might be worth using Misty Slash to interrupt his desperation attack in the final phase if it can prevent his final jump, but I'll have to look into that later. Everything leading up to and including Wanyudo should be pretty straight forward. There's a different route with different shrine teleports that RTAers use leading up to Yukinojyo and the mountain god fights. I'm not yet sure if the route is actually faster or if it's just used to stock up on more smoke bombs and grab the pheasant hot pot recipe(neither of which would be worth taking a considerable detour for in a TAS I think).
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Wow, what an unbelievable improvement. I didn't expect this to be finished nearly so soon. I always figured Claimh Solais was just the RTA substitute for Final Sword since you can't reliably manipulate a Final Sword drop in real-time, so it was nice to see it have its use in a TAS setting as well. The route this time around was completely unexpected. Everything from going to the Chaotic Realm so early, to casually strolling through Great Armor's and Julius' rooms without actually fighting them, to getting Kicker Skeleton and Malphas so much earlier on, to incorporating the Blocking Mail and underwater Catoblepas dashing made the run much more entertaining to watch. How did you start that Giant Bat zip from mid-air in the Undine room? I had no idea that was even possible.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Zakatos wrote:
Too bad the password algorithm can't hold the info of the Zero Saber independently of the 8 Mavericks defeated, since the passwords generated after obtaining it seem to be invalid. It would be great to start a run from a cracked password which gives you the Saber at beforehand, so it could be used in all 8 stages.
If you use the same glitched password as in paosidufygth's any% submission, you could go get the saber from the Doppler stages first without any Mavericks being defeated.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
I liked the part where you went so fast that your car exploded. Yes vote from me.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Oh, I just noticed your signature. I just started working on it actually. Would you perhaps be interested in collaborating? Maybe we can get someone with disassembly knowledge to help with RNG. Go ahead and send me a PM if you're interested. I'm sure you'll be very busy with AoS, but I'd be happy to collaborate on it in any sort of capacity.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Well that was an easy fix. Thanks for the help, Alyosha!
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Is there anyone familiar with this game or emulator differences that can help me to understand something? First of all, is the RNG address 00E4? I'm able to get the same drops every time by freezing it, but the sequence doesn't seem very random. Most of the time, the current value is just the previous value divided in half. But comparing the run I've started on BizHawk 1.11.6 to Glitchman's run on FCEUX 2.1.2(and Pike's Star Man WIP as well), the bytes of the values in this address are reversed. For example, on FCEUX it goes C891, 6448, 3224, 9912, CC89, whereas on BizHawk it goes 91C8, 4864, 2432, 1299, 89CC, etc..., and I can't seem to get anywhere close to the same asteroid patterns in the first room(and am getting a lot more lag as a result). Does anyone have any idea what's going on here?
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
I see. Is there anyway to rebind this? I find myself constantly fighting muscle memory because this was how the pause key worked in previous versions.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
What's the "Recording mode" you're talking about where that issue doesn't happen, Samsara? I'm having the same issue happen even when unpausing. The only way I can get the TAS to stop on the green arrow is by letting it play past it to erase it, pausing, and then clicking to make a new one. Also, I can't see what frame the TAS is on in TAStudio if there is a green arrow up. The blue arrow just stays paused at where I started it until it reaches the green arrow, then resumes as normal. I have 2 older builds with working branches, but one of them has the first issue I mentioned(but not the second), and the other doesn't have a working ram watch.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Actually, now that I think about it, this glitch might not be faster in 100% since you neon jump to the left wall before calling Zero and are subsequently nowhere near a green charge upon Mosquitus' room, which allows you to bait Mosquitus much further to the right while waiting for your charge. Here's some disassembly of the Mosquitus descent behavior that FractalFusion did, since it may be useful for the 100% run Glitchman is working on. The way it works is, every 4 frames, Mosquitus will poll 3 RNG values. Presumably, 2 of them are for the smoke cloud he creates from being damaged, and one is to determine his direction. Depending on when you kill Mosquitus relative to the global timer, these two events may or may not coincide on the same frames. If they do, the values relevant to his choice of direction are the current value the frame after 4 values are polled, and if they don't, the current value the frame after 2 values are polled. Since the RNG also increments once every frame, this means 7 values will be polled every 4 frames. As such, this text document has the RNG sequence divided up into 7 columns, so any Mosquitus descent pattern would start on a given value then go down that column for around 30 values(depending on how optimal the pattern is). RNG values with nothing next to them cause Mosquitus to fly towards Zero, and values with letters next to them represent one of eight 45° angles. This table doesn't account for whether or not Mosquitus will attempt to grab Zero at the very start of the fight, which also needs to happen, so some of these patterns that look good may not be viable. But since you dump the second preliminary buster shot off screen while falling back to the ground, you can potentially drag it off screen quicker or slower to advance RNG at a slightly different rate in between the getting grabbed and the descent(or delay your saber slash).
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
The fastest stage that could be attained with further lag reduction on the bee crashes was a net time loss of 1 frame. Even if it was a net 0 like you seem to be assuming(and perhaps my wording in the submission text may have suggested), I fail to see how displacing 1 or 2 lag frames during heavy screen shaking while waiting for a camera lock to expire to a load time could be in any way perceptible to the viewer or worth anyone's time or effort to pursue.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
You just have to kill Mosquitus when he's a certain distance from the wall. Normally he doesn't go that close to it, but he will if you let him grab you close enough to the wall, and then stay to the left of him before he chooses his next direction to move. This glitch is very convenient for categories which acquire the saber since it removes the necessity to manipulate Mosquitus' descent pattern after the kill. Good patterns for that are so few and far between that it usually took multiple stages worth of manipulation to set up. Also, it works on both the PSX and SNES versions.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
I was wondering if there are other factors that play into heart drop rates for enemies(besides the Heart Pendant). Like for candles, ones that normally drop hearts will always drop money if you have full mana. Sometimes hearts are dropping from enemies for me when the script says they should, but the overwhelming majority of the time not.
Experienced Forum User, Published Author, Skilled player (1221)
Joined: 8/29/2014
Posts: 301
Don't you need to be below a certain amount of mana for hearts to drop? Or does that only apply to candles and not enemies? Good luck on the new TAS. I can't wait to see what the route looks like.
1 2
5 6 7
12 13