Posts for FatRatKnight

Post subject: Re: guardian legend
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
RedLander wrote:
To FatRatKnight! On the first boss, how do you get the enemy on the left side to open up at frame 4926 in your abandoned run? I can only get it to open up at frame 4930. Has it something to do with that you use the new ppu? i use the old. Can it be the rom version? i use the (U) version, but i dont know what dump it is, it doesn´t say.
Honestly, I don't know. It just worked out that way, and I have made no serious investigation into it.
RedLander wrote:
I see you use (!), which is a good dump right? i´ve been looking for a good dump every where but cant find it. do you know where i can find it?
Can't answer the latter question. Rules of this site. But yes, I use the good dump. I think, anyway. When playing a movie, the checksum should match whatever ROM you're playing. As I'm using a theoretically good dump, and when looking at the movie information, if the movie file's checksum matches with your ROM, you should be good.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Twisted Eye wrote:
My plan was to bury the heavy weapons guy and just manipulate, say, grenades from one of the on-screen agents and the rapid gun from a building using the new life's three grenades. Sounds like it really would be faster in a perfect world, but that's a dang lot of manipulating.
Might be a good idea. But like you say, the amount of manipulation required to have the right drops come from the nearby agents makes it particularly difficult to do. I might want to look into it for a bit, however. Would be a great way to go if the drops work out. But in return, I will need a weapon that produces a short death animation, two extra grenade drops, and something to cure the headache of trying to even get it to work. Might not be possible given how the RNG works, but we'll see...
Twisted Eye wrote:
Why would flame/ice be better than spread? Did I miss something when I was checking their damage output? --Yeah, sometimes it does an extra one, even two damage every 8 frames, I never noticed. This makes your final battle, starting with fire switching to ice, much more impressive than I first thought.
The unique thing about flame/ice weapons (and the green cloud weapon) is that they leave a lingering hitbox. No other player-usable weapon in the game has a hitbox that lasts more than a frame after hitting something. There's a few properties involving this fact that the game has to deal with, such as keeping a short list of hitboxes to ignore. The length of this list is apparently 4. The cooldown of the weapon and the length of each "cloud" means I can only have 3 ongoing clouds at once. However, I can still trigger an overflow of this list with the regular projectiles and grenades. So the game forgets to ignore the first hitbox, allowing it to hit again. In essence, each hit is effectively improved by one damage. If it were possible to have 5 ongoing clouds at once, it breaks the hit detection. The game forgets the first cloud, lets it hit again. Oh, another hit, better forget something else: The second cloud! Oh, let that hit again. Then it forgets the third cloud... See where this is going? Repeat until the final cloud. This is all in the span of a single frame, so with 5 ongoing clouds, it hits 5 times per frame for 5 damage in that frame. But you need to break the natural cooldown of your gun to do this. This can be done by switching weapons, either by running out of ammo or picking up a new weapon. Unfortunately, to break the hit detection like this, it requires at least 3 weapon drops right by the boss in question, all ready to be obtained 2 frames apart, and the player has one of three weapons (green cloud, ice, flame) already in hand. Still, for one shining moment against the final boss, you see me pick up the ice weapon. For a short time, I have 4 ongoing clouds. Each hit from a grenade dealt an equivalent of 7 damage instead of 3 -- Grenade hit made game forget the first cloud had hit. In turn, it forgets the second cloud had hit. Then the third and forth. 3 from grenade + 4 from clouds re-hitting. So the flame/ice weapons are usually optimal versus bosses. But the cost is a worse start-up: The spread can at least be shot from a long distance and have every hit count by the time the boss shows up; The flame/ice weapons are 1/3 as effective in this sort of case. There may be other reasons like the fact the 1-1 boss stands too short to be hit by the cloud when it is moving.
Twisted Eye wrote:
I'd like to hear why 2-4's boss fight (the second spaceship stage) came out how it did. Was turning away from him better for controlling his movements or some such? I never got to the point of trying to TAS him, so I'm only going on a feeling of it seeming counter-intuitive.
I chose to use the back guns because each hit from those dealt 9 damage. As it produces two shots, that's 18 damage. The main gun only produces 1 shot, despite the graphic, and deals only 13 damage. EDIT: The boss stays enough away so that I can't stay inside the boss. There might be some tricky movement we could do, but on a time restriction, I didn't look too hard. The boss naturally wants to just hang a small distance away from you.
Twisted Eye wrote:
I can't think of anything else in particular strategy-wise that could possibly help Team 5's run, just awesome TASing on display there!
We have indeed worked pretty hard at it. But even so, there are improvements to make. I've already found a way to save a frame on our starting menu navigation, anyway. Involving the use of the start button instead of A when applicable.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
MR? ... I have a habit of renaming functions and forgetting to include the rename... MR is short for Memory Read. That was my train of thought, anyway, and I'd rather not have to retype the entire function name every time. A single line at the top of the file can fix that: MR= memory.readbyte Note: No parentheses. I'm not calling the function. I'm assigning the function itself to MR, not whatever value the function spits out. Nowadays, I've gotten a new habit of using R1s, R1u, R2s, R2u, R4s, and R4u. Any guesses here? The fact your values are "off by one" for C353 is just fine, actually. Monsters are put in one of three slots, and even if the other slots are empty, the game still rolls for those slots. If the monster was placed in slot 1, the game's first roll is given to that monster, whereas if the monster was in slot 3, the third roll is given instead. It's not really visible which one of three slots the monster is in from the battle screen, though. If nothing else, eat the extra line of text and GAZE or STONE them anyway. The BATTLE sword, in my plans, are never given to the GAZE mutant in the first place -- She uses the HAMMER versus SEI-RYU. The spare POWER mutant gets the BATTLE sword. I was using the STONE book against BYAK-KO in my Agility test -- What else would I use? Have you tried putting the Arts & Crafts mutant (STONE book wielder. Just in case you haven't figured out what I mean by art by now) in slot 1 directly? There's no avoiding placing the mutant in slot 1 eventually. I'm not sure if that extra battle in W1 is so "free". The TELEPOR route requires that you switch twice to pick up the various Mana boosts. There's also the fact that TELEPOR is given to the third member, and you most certainly must swap the spare POWER mutant at least once to get that and POWER. Three swaps are required here, with three menu transitions likely. Not eating the extra swaps, we can spare a Mana+5 for a Mana+4, or even Mana+3 if we're giving GAZE to the starter MUTANT F. Even if we are still restricted to a Mana gain, we have more possible values to use.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
I made a somewhat sloppy real-time test between 4 SHIELD and 5 SHIELD. Just walking from the SHIELD power-up in the lower-right corner of the map and then walking to each of the endgame corridor rooms, with a death-warp from Corridors 9 and 10. The difference is around 2000 frames, but given the sloppiness, I might be off by around 500 frames. As this is simply a walking test from the same starting point, this doesn't take into account collection times. Even if the walking alone isn't enough, the defensive advantage might just be enough to handle boss attacks more boldly. Still difficult to say if it's worth it, but looking at what having 1 SHIELD means for defense, I would lean towards getting some SHIELD pick-ups, unless we can avoid everything without slowing down our attacks against the boss. I took a look at the CHIP power-ups. There are 3 free pick-ups. Obviously, we want the one in AREA 0. If we're grabbing the SHIELD off the mini-boss in AREA 3, the next one is just one screen out of the way. The last one is in area 9, only a few steps out of the way. So we have 200 CHIPs to use by the end if we avoid mini-bosses. 150 CHIPs for a long part of the game. If we do go out of our way for a mini-boss, we'll have 400 by the end. Two mini-bosses, and we should have 800 by the end. So two extra pick-ups gives a generous amount of extra CHIPs for us to use, with more use from the occasional red round pick-ups. Will we need that many? I think the item boxes drop depending on how many kills you make. I took a look into the game's memory a bit. Not much, but to start with, this is what I have:
003F  subweapon frame counter for CHIP consumption
0043  Related to enemy kills. Used for item drops?
0044  Related to enemy kills. Increments when 0043 overflows
004C  Current CHIP count
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Bobo the King wrote:
Just curious: Why does the Mutant M need to die and why does the fodder need to die? (I know at least one of them has to.) Also, what potential advantages do you see to having a female POWER wielder? In any case, it's a cute idea.
There is one chief reason for killing MUTANT M at SEI-RYU: The first party member will only go first against a higher speed enemy if the fourth member has lower Agility than the first member. If I don't give GAZE to the starter MUTANT F, I'm stuck with a slightly slower secondary MUTANT F. The only member that would be slower than this one would be the MUTANT M. When a party member dies, they go straight to the fourth slot. If someone with too high an agility is dead and thus forever stuck in slot 4 (reviving takes too long), this kills whatever chances I have of ever going first against the SABERCATs, BYAK-KO, and ATOM ANTs. The outcome is rather bleak for the survival of our heroes. The only real advantage in having two F POWER users instead giving one to the M is so I can kill the M off in this fashion. The remaining fodder is put to good use as a humanoid body shield at Monster Trio. After Monster Trio, our party should be: 1) GAZE gal, with Agility higher than member 4. 2) Spare POWER mutant. Useless beyond Monster Trio, though. 3) Dead fodder. Tragic loss at the Monster Trio. 4) Dead since SEI-RYU. Must have less Agility than member 1. There's 3 ways to have Member 4 be slower than Member 1: - The dead member is the slow MUTANT M, thus slower than anyone else. - The intended STONE book user is the starter MUTANT F, thus faster than anyone else. - Fight an extra battle so that Member 1 ultimately has more Agility than Member 4. Equal Agility won't work. Member 1 must have higher Agility than Member 4. This is to get around the strange turn order system the game uses. The second and third slots don't really matter, from what I recall of my basic tests. EDIT:
Bobo the King wrote:
I'm going to toy around with your Lua scripts. I wasn't able to decipher them by looking at them, so I'll just see what they can do for me.
Those two are just stand-alone functions. Stick them together under one file, call the "trigger" known as ScanReset2() every frame, and gui.text the variable str somewhere.
Bobo the King wrote:
My Monster Trio strategy differed slightly from what I believe is the prescribed method. As I recall, you intended to have our POWER user (B, in my case) pump himself up and take care of one of the monsters. Unfortunately, POWER doesn't quite cut it and it still takes two turns to defeat a monster. [...]
The plan for the Monster Trio is to hand the BATTLE sword to the spare POWER mutant. The GAZE mutant isn't using it, is she? On the other hand, I can't remember if even the BATTLE sword is enough...
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
The RNG is affected by what the CPU is doing internally. Generally, the music itself has a significant impact on the RNG. So does many different actions like exiting the menu. Part of the reason why I was considering using the Sound Test, of all things, to manipulate luck. As for why I didn't bring this up earlier, it's simply that I don't think to say these details. It's quite important to know, and I forget. ... I have a few things I want to look at, though. I think I should look now... One test: Party of mutants, F F M F and see how the random battles go from there. At SEI-RYU, the M must die (two Fs will have POWER and would steal his HAMMER away from him). At monster trio, the fodder F must die. After BYAK-KO, anything goes after we get our manly man. Another test: Party of mutants, F M F F. No swap. Fight an extra battle to give extra Agility to the fourth F. Even a point or two will make the run feasible. The third test: Party of mutants, F M F F. Swap first with fourth so the GAZE user is the one with the extra Agility. What you've been doing, with several mistakes. Here's what I use to scan the RNG:
Language: lua

local str= "???.??" --***************************************************************************** local function RefreshRNG() --***************************************************************************** local addr= 0xC300 local Rn= MR(addr) str= string.format("%3d",Rn) .. "." local J= Rn while J == Rn do addr= addr+1 J= MR(addr) end if addr == 0xC308 then str= str .. " 0" return end local val= 64 - 9*(addr - 0xC300) Rn= J local count= 0 while Rn == ((J-count)%256) do count= count+1 J= MR(addr + count*7) end val= val + 9 - count str= str .. string.format("%2d",val) end
And the trigger that will detect resets for me:
Language: lua

local RsState= "Idle" --***************************************************************************** local function ScanReset2() --***************************************************************************** --The "read memory for joypad instead of the broken joypad.get" alternative. if RsState == "Idle" then local a= memory.readbyte(0xC3A0) if bit.band(a,0x0F) == 0x0F then RsState= "Rs?" end elseif RsState == "Rs?" then --Yes, elseif. local a= memory.readbyte(0xC3A1) if a == 0 then RsState= "Oh" else RsState= "Idle" end elseif RsState == "Oh" then local a= memory.readbyte(0xC300) if a == 0 then a= memory.readbyte(0xC320) end if a ~= 0 then RefreshRNG(); RsState= "Idle" end else --Oh, snap. An error. I typoed? RsState= "Idle" end end
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
The only way I can see us reach the ending without first completing Corridors 1 through 10 is by passwords. Either by having them completed as per password stats or by exploring some off-map area to find Corridor 22 or some such off-map glitching, again reachable by password. But passwords are out of the question. If we allow them, we may as well pick one that starts us in the Corridor 21 room with optimal stats. Kind of like this April Fools' submission. Probably would be viewed as cheating in some form. I can't think of any way to enter Corridor 21 early from a usual start. Hard to say what's the best way to go about things. At a minimum, we will have 2 ATTACK and Lv2 saber. Lv2 double saber is easy to have with one free pick-up a few steps out of our way. Lv3 double saber is very much out of the way, Lv3 saber requires an optional corridor, and 3 ATTACK needs another mini-boss for low gains. Is there anywhere we would want a Lv1 double saber before AREA 4? This makes me want to have some sort of testing going on. I can produce passwords that skip ahead so we can check a specific part. For one thing, I want to see how much walking there is from the point where we get the 5th SHIELD. For another, having a specific combination of power-ups means a different password instead of hunting it ourselves. These passwords will be for testing purposes, of course.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
EDIT2: Filling out the rest of the stuff. I hope it helps out in planning when to reset for encounters. _9 Tower encounter rate _5 Tower2 encounter rate 21 Tower side room (when we get the SAW, those non-stair tiles have this rate) 64 Particular steps in the near-final climb has this increased encounter rate _5 W1 Open land _6 W1 Forest 13 W1 Swamp _5 W1 Bandit cave _5 W1 Castle Sword 1F _6 W1 Castle Sword 2F 13 W1 Castle Sword 3F _5 W2 Land _5 W2 Network of caves _6 W2 Sailing Island _5 W2 Underwater _5 W2 Underwater caves _5 W2 Castle 1F 21 W2 BLUEKEY hall _5 W2 Castle 2F _5 W2 Castle 3F 13 W2 Castle 4F _5 W3 Cloudy terrain _6 W3 Forest _3 W3 Glider 64 W4: "Invincible" SU-ZAKU (no bike) _5 W4: Subways (no bike) 13 W4: "Invincible" SU-ZAKU (have bike) _6 W4: Hidden town (have bike) _6 W4: Subway to ERASE99 (have bike) _6 W4: ERASE99 place _5 W4: ERASE99 place, top floor _5 W4: Aftermath, open land (no bike) _9 W4: Aftermath, craters _5 W4: SU-ZAKU tower _9 W4: Elevator 5, the last elevator _5 W4: Train station & trains _3: 7C 96 FF _4: 7C 96 A1 FF _5: 2D 7C 96 A1 FF _6: 00 2D 7C 96 A1 FF _9: 00 13 2B 2D 7C 96 A1 B2 FF 13: 00 13 2B 2D 38 55 7C 96 A1 B2 BA F9 FF 21: 00 05 0E 13 1E 2B 2D 34 38 49 55 67 7C 96 97 A1 B2 BA C4 F9 FF 64 ("invincible" SU-ZAKU): 00 01 05 0C 0E 13 1E 2B 2D 2E 2F 34 37 38 43 49 51 54 55 58 5B 67 70 73 76 7C 7D 86 90 96 97 99 9A 9C 9F A1 A2 A4 A5 A6 A8 A9 AA B2 B5 BA C2 C4 C7 C8 CA CB CD CF D2 DD E0 E3 EC F1 F9 FB FE FF Step counts. Counting only the ones that increment the appropriate RNG. 172: 5,6 Tower to P-FROG Bandit _77: 5,6 Tower to Castle Armor 114: 5,6 Tower to bridge prior to swamp _14: 13 Steps into the swamp _80: 5,6 Forest, 1F, 2F of Castle Sword (w/ perfect NPC manipulation) __8: 13 3F of Castle Sword _48: 5,6 Tower to Castle Armor __5: 5,6 Castle Armor to Town of Hero 150: 9 Through the Tower to World 2 431: 5,6 World 2 errands to BLUEKEY hall _23: 21 BLUEKEY hall 120: 5 BLUEKEY hall to Underwater Castle 4F _30: 13 Through 4F to SEI-RYU _58: 6 SEI-RYU to Old Man Ryu-O 234: 9 Through the Tower to World 3 _14: 5,6 Tower to town _51: 3 Solve Jeanne's problem and return to floaty fort _48: 9 Tower TELEPOR! _15: 5,6 Tower to town and get Glider 104: 3 Glider to STONE Book and then BYAK-KO 118: 9 Through Tower to SAW room _11: 21 Get SAW through high encounter rate 118: 9 From SAW room to World 4 __4: 64 SU-ZAKU zone _65: 5 Subway __3: 64 SU-ZAKU zone _31: 13 To hidden town _45: 6 ROM get __9: 13 To northeast town __2: 64 SU-ZAKU zone _33: 13 Back from northeast town 148: 5,6 To EVIL EYE (using no-encounter zones) _57: 5,6 To Plutonium/ERASE99 _43: 5 Stroll to tower __6: 9 The craters need to be traversed 309: 5 Through most of the tower _21: 9 Final elevator 214: 5 Through the trains to SU-ZAKU 279: 9 Oh no, ASHURA battle! SAW, your might, SAW, go! __4: 9 Steps after battle __7: 5 To GEN-BU2! 110: 5 To SEI-RYU2! 185: 5 To BYAK-KO2! 220: 5 To SU-ZAKU2! 174: 5 To near-final zone. __3: 64 High encounter zone. _49: 5 Last elevator Also, suppose some path takes 210 steps to finally get through and meet the boss. Is it wiser to manipulate closer to the 93 step RNG per reset, or don't bother and go for approximately 70 each time? The 93 steps is still the way to go. Just because you finish some leg of the journey doesn't mean that, suddenly, the remaining 69 steps are completely worthless. You were at C33B:5E when you made the P-FROG bandit drop dead. The next encounter was at 7C. The difference between the two numbers, 0x5E and 0x7C is 30 steps. By resetting when you did, you effectively wasted 29 steps. Ouch. Keep that in mind. Just because you beat a boss doesn't mean the leftover RNG is suddenly bad.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Any particular reason for the Red Landers that I don't foresee? Sure, faster and wider shots at specific CHIP levels, and more for use of our sabers, but I don't intuitively see how these can help greatly. Let me know of any trouble spots you did come across in an earlier run. As for fighting normal enemies, 2 ATTACK and level 2 sabers means 8 damage per frame. Picking up another ATTACK lets us deal 12 against normal enemies and hittable projectiles. Lv3 (double) saber with 2 ATTACK deals 16 damage per frame against normal enemies and all projectiles, where 3 ATTACK with Lv3 (double) saber gives 24 per frame against the same wider set of targets. So, besides the bosses, anything in the corridors that can give trouble to someone with pitiful damage? Enemies' HP maxes out at 255, so against these, lv2 saber with 2 ATTACK would take around 32 frames to kill, where Lv3 double saber with 3 ATTACK takes 11 frames. Part of the reason the safety zone even works is probably due to enough damage or the ability to destroy more projectiles with Lv3 saber or double saber, against what the boss keeps shooting at you even in there. It does hit twice as hard as Lv2 saber, so even against regular hittable projectiles, the more durable ones are destroyed twice as fast. As for health, I think having the SHIELDs will be enough, no need to specifically spend time finding out of the way Blue Landers. With 1 SHIELD, we take full damage; With 2 SHIELD, we take half damage; With 3, one third; And so forth. 1 SHIELD would be a disaster to have, as a maximum 32 life can be completely eradicated from a single attack from a few different late-game enemies. Although, while our health is still scrolling from the last attack we took or from a recently picked up item, we are invincible. ... As I recall, anyway. If you can get an item box to drop, try timing the moment you pick up its contents so we can just ignore an attack. Should help with any lacking defense. In theory, even a blue round pickup, as little health as it gives, can let you ignore an immediate dangerous attack. Let me know if these affect any plans. Also let me know a few reasons you can think of why we need some things.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Bobo the King wrote:
Up to this point, we have believed that soft resetting resets the battle RNGs to a fixed state and the outcome of a battle cannot be changed by exploring other RNG seeds. I can tell you now that is false.
You mean this, at the bottom of the post with an edit as an afterthought? As far as I know, the only thing that can be manipulated by reset is enemy targeting, at C35A. The chosen action might change depending on earlier stuff throughout the battle, but I haven't been able to reset to change that. I would recommend putting TELEPOR on the GAZE user. Failing that, the secondary POWER user is the next best one to have it on. Speaking of, why did you set things such that you have three different characters that each must survive at SEI-RYU? One with GAZE and POWER, another with POWER, and the last with TELEPOR. It's quite possible to fit TELEPOR in one of the other two instead, letting you scoot by with two valid targets that can just drop dead instead of only one, therefore making it easier to handle.
Bobo the King wrote:
[...] We could also give TELEPOR to C, but I'd rather not if we can help it, since I've learned my lesson regarding giving TELEPOR to cannon fodder. It's sticky, to say the least.
Although, it seems you already thought of this. An optimal case, as I like to see it, is that we lose a "useless" party member at SEI-RYU and just move on without a second thought to our lost friend. Less menu time in future battles thanks to only 3 members instead of 4. Also, if you're going to have a dead character after the battle (which will happen. Why did I use an if?), remember that this mutant will no longer be counted as a roll for the RNG. So if the Mana+5 is going to be given to the third mutant, put the one that will get the Mana boost in slot 4. Once someone dies, that mutant will slide into slot 3, and get the boost as though the mutant was in slot 3 to begin with. Incidentally, the boost that goes to character 4 is useless in such a case, as the RNG simply won't roll enough times to hand it to any mutant. Somehow, ESP blocks the physical attack of SEI-RYU on the first turn. I've had success with that in my tests against SEI-RYU, if you want to keep that in mind.
Bobo the King wrote:
I discovered it takes 1544 frames to buy the sword from Port Town but just 1121 frames to get it from East Island Village, saving 423 frames. That includes the time it takes to leave and return to the main path. Much thanks to Shiner for pointing that out!
That is certainly of good help. Faster BATTLE sword is always nice to have. ... I'll get onto that encounter data soon... Haven't checked yet.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
BadPotato wrote:
Could you guys tell me if this: ==> http://www.nicovideo.jp/watch/sm7552129 is the same game? It seem to have the same music, the same (glitchy) engine... !
I don't know how to view it, but this site has a published JPN version. Since the USA version doesn't allow such memory corruption, and even with it, forgoing use of such a glitch shows so much more of the game, enough to hopefully justify publication alongside. I just looked at how much mana your character currently has. Here's the projected values at the start of each battle: 26 - SEI-RYU 31 - 1-3 MOSQUITO 36 - GARLIC + SABERCAT + P-WORM 41 - 1-3 SABERCAT 46 - BYAK-KO You already must get +5 mana at each battle to get enough in time for BYAK-KO. No third POWER available without an extra fight. As for the excess amounts of HP you've been getting... Yeah... Not sure how you got that much, but we might actually need to sell something with that much HP eating up our money just to get the BATTLE sword. Plus the extra damage you did take isn't helping matters. I'll look into the encounter rates for a bit. At least have some numbers we can use to plan out where we want the RNG.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
We use a starter HAMMER from a MUTANT M to get the extra damage we need. No need to buy any weapons beyond the needed BATTLE sword. Or, for that matter, sell anything. I thought I uploaded this, but I couldn't seem to find any links in my previous posts. Well, here's a new upload. ... Clearly, I screwed up the numbers for the RNG... New numbers: C30B:93 (nil, Def+2, ESP, POWER) C30B:94 (Def+3, POWER, BARRIER, nil) C30B:95 (POWER, BARRIER, nil, HP+) Again, to force a single enemy against STEWARD's cronies, C353 must be one of these: 1B 2A 2B 2C 35 4E 52 64 6D 7A 8D 94 97 99 9F A1 A2 A3 A6 A7 AF C4 C5 C8 CA E0 FC FD FE If you're looking for POWER, that 9F is your only choice. Best make sure the mutant you want to give POWER to is in the first slot. The probability ain't bad at 56 possible outcomes. When you have a nice party with the necessities, cheat the random battles away. Unless you really want to experiment with avoiding battles, the point of the test run is to see how we can deal with the battles we must fight. Naturally, you won't be able to record the cheating all that well in a movie, but being able to stroll straight to bosses without intervening battles speed things up greatly. Or perhaps just copy the party straight into the movie I presented up there. Skip the travel time! The state in the linked movie has the BATTLE sword in inventory, so you don't need to hack it into one of your members. Theoretically, this script might work the copying for you:
Language: lua

local T= {} --***************************************************************************** local function GetStats() --***************************************************************************** for i= 0, (4 * 0x1F)-1 do T[i]= memory.readbyte(0xCC00+i) end end GetStats() --***************************************************************************** local function SetStats() --***************************************************************************** for i= 0, (4 * 0x1F)-1 do memory.writebyte(0xCC00+i,T[i]) end end while true do local k= input.get() if k.numpad7 then GetStats() end if k.numpad1 then SetStats() end emu.frameadvance() end
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
As far as I know about the microphone recording here, its only purpose is for the games that need it. As such, it only uses a single bit per frame for the microphone. If you're wondering about full audio from the microphone being played back through the FCEUX emulator, you're out of luck. There's no support for that. If anyone has any corrections to what I said, please state them. I am pretty sure, but not completely positive about it. The auto-hold simply holds down a button for you. Imagine taping down the A button, for example. Typically useful for frame-by-frame playing when our keyboard simply refuses to accept too many keys at once. If you're going to be playing games for Let's Play videos, this shouldn't be of much concern to you.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
At your Mana level, it shouldn't matter how much Mana you gain from STEWARD's cronies, even zero is fine as long as you guarantee +5s from later battles. I'm curious what your plan for SEI-RYU is. Tell me what it is, since I don't see a case here that matches well with my plans.
[...] We could also get a second POWER user! I have a feeling we've talked about this before, but what are your thoughts?
Oh, wait. You don't seem to recall this is a requirement in my plans. It won't work without a second POWER user, so yes, we absolutely need one! Since the GAZE user has more Agility than the fellow members, I should expect said member to go first without an intervening battle before KINGSWRD. No problems here. All that's happening with the swamps is the fact the encounter rate changed. There's 14 steps of swamp to go through. Just set the RNG so you have a 14-step gap timed for this encounter rate 13 zone. The numbers for encounter rate 13, by the way (hex values): 00 13 2B 2D 38 55 7C 96 A1 B2 BA F9 FF Bold: What we'd find in encounter rate 5 anyway. Remember, resets take a very small amount of time. If you're manipulating a hundred frames for what amounts to roughly a few steps, forget it and eat the extra reset. The leftover steps are still likely useful after the boss battle. I'm finding it difficult to get myself back into gear for this run. Sorry if my posts aren't so full of content like usual. Keep talking, hopefully it'll help me get myself fully aligned again. Finally, thanks for the congratulations. And yes, the team I was in won.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Now that I have a bit more time, I will ask about a few things. Primarily about any%. Any items that are "out of the way" that are worth picking up? I'm thinking a saber from AREA 0, at X 9 Y14. 2 damage per hit from the start with a subweapon is certainly good. The ATTACK up found off a miniboss from AREA 2, at X 5 Y17. This enhances our basic shots forever, and will be of help for corridors 2 and on. Not to mention, even with level 2 saber after completing Corridor 2, that's 5 damage per saber hit (versus bosses) instead of 4. There's another ATTACK up in AREA 4, at X18 Y 1. And just like the other ATTACK up, it's just a single screen away off our path and another mini-boss to kill. However, the boost this one gives is relatively much smaller. Your basic shots would deal 3 damage with 3 ATTACK, and the level 2 saber is no more effective compared to 2 ATTACK versus bosses. Regardless, it's still found with 7 primary corridors left to go. Is level 3 double-saber worth it at all? Sure, we get one level free at Corridor 7, and there's another one almost in the way at AREA 4, but the one in AREA 1 is almost certainly far off the path. The only real advantage of level 3 over level 2 is the fact it destroys a wider range of projectiles (which bosses like to spew out) and deals twice the damage against non-bosses. What about SHIELD? Sure, four pick-ups, and we have increased walking speed. And the other perks with taking less damage is nice. But which four do we take for this walking speed? The earliest AREA we can get this many is 6, and if we skip Corridor 13, it's AREA 7. We've already gone through more than half the game's worth of walking by then. Any thoughts on any% item collection? The saber is certainly a nice weapon, and it's not like you can ever deal more than 5 damage per hit with any subweapon versus bosses.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
0x03000332 - Camera lock flag 0x03000328 - Kills left to unlock scene 0x0300031A - Enemy spawn timer Seeing as I'm familiar with the winning run, there's the breakdown of 1-1:
1-1
 2508 ->  2675  S 1 Input begin to unlock
 2766 ->  2885  S 2 (22)
 2944 ->  3040  S 3 (30)
 3206 ->  3248  S 4 Van
 3306 ->  3654  S 5 3471: boss dead
 3719 ->  3816  S 6 ( 0)
 4077 ->  4846  S 7 4460( 3) Kill counter appears
 5044 ->  5304  S 8 heavy guy
 5368 ->  5528  S 9 (25) 5 frames away, is it possible?
 5581 ->  5707  S10 Can't avoid losing a spawn...
 6187 ->  6487  S11 Boss kill

 7193  Clear score screen
Definition of "scene" in this case are all the locations the camera locks. The first number is when the camera locks, or in the case of the first scene, when the first frame of input is allowed. The second number is when the camera unlocks. The S## is scene number, for our convenience. The (##) for some of them represents what the spawn timer was at when the kill counter starts. The spawn timer is the frame rule of this game. It ticks down every frame. After it hits 0, it resets to some upper value and spawns an enemy. Since it goes through every number from 30 down to 0, there's a 31-frame rule in place. In 3-4, it resets to 10 instead. If the number is at 30 when the camera locks, you've hit exactly the moment, down to the frame, where the game can produce a new enemy. 29 is one frame too late, sorry. This is important as the game usually has an upper limit of 2 enemies on the screen at once. Time a kill right at the perfect moment, then when the enemy finally disappears entirely, it's right at the exact moment where the screen locks. The kill counter already has 1 kill marked off, and a new enemy has spawned because you're now under the limit. The best possible case. 1-1 Scene 1 We've decided on picking up the purple auto-firing weapon. It produces a very short death animation, compared to other weapons. We waste no time picking up the weapon gift, as the enemy spawning won't begin until it's picked up. We wait until tossing grenades after the weapon pick-up. There is, after all, a building ripe for dropping what we want: More grenades! We're stuck waiting for agents anyway, so why not? The kill counter is set at 1 in this scene. We kill the first agent as early as we think possible -- The moment his hitbox accepts an in-flight bullet. 1-1 Scene 2 On the way to this scene, a grenade pack is dropped. The timing these two agents were killed were such that they were counted when the screen finally locks. Not a lot to say except that the last kill needed to end the scene was done with a bullet killing that agent at the moment he spawns. 1-1 Scene 3 ... And this is why we needed that exact moment. When this scene locks, the spawn timer is at 30. A single frame of wait, and that delays clearing this scene by 31 frames. Thanks to this, we also have a one-frame window to kill an agent to coincide with the screen lock. We hit that window. We made sure to keep using the auto-firing weapon right up until we picked up the spread gun. The short death animation is certainly nice, and we get to keep all the spread shots for later enemies. This is also the first instance where we begin charging up and keeping our charge while still shooting enemies. Maybe a tad late, since we reached 9 damage charge, not quite 10. 1-1 Scene 4 The FBI van. We used the car so that we can fire while still moving at a decent speed. A short delay was taken here so that our shots have a longer travel time, and thus can shoot earlier. We still lock the camera at the same frame anyway. Poor FBI van barely even had a chance to get any screen time. We jumped out a frame or two early (I forget now) so that we have a slightly better height against the following mini-boss. 1-1 Scene 5 The robot. If we kept the car, it would promptly explode when this boss lands. Then we're launched upwards for a position that is no good for grenades. However, I'm pretty sure the flame or ice weapons would be more effective against this boss. Far worse start-up compared to the spread, so maybe manipulate an agent on the scene to drop a flame weapon. Whatever the case, any new strategy here must be good enough to beat this scene at least by 30 frames to beat the frame rule. 1-1 Scene 6 The spawn timer was 0 when we reached this scene. Not bad, any improvements means that at least 30 frame must be saved to have any hope of getting through this scene earlier. I really should experiment with the flame/ice weapon sometime. Probably flame, for a reason involving death animation length. In any case, the spread gun's death animation is far too long. Knife, grenades, anything that has a death animation shorter than roughly 60 frames. Else, we lose a spawn. We saw a grenade drop, but I said something to the effect of "forget it and move on." The building was used for a weapon change to the purple auto-fire weapon. 1-1 Scene 7 Off camera, I was manipulating two grenade packs. On camera, it's very obvious I'm shooting numerous times while still charging. When you roll or backflip, releasing B does not dissipate your charge. On the frame you finish rolling or backflipping, you can press (and hold) B to shoot your gun without losing your charge. Not shown here, you can also jump to keep your charge -- On the frame you press A, release B. Hold B again on the following frames and you shoot. This also works on the frame you land. I left the scene with maximum charge of 10 damage. Not bad. 1-1 Scene 8 Somehow, this guy takes double damage from grenades. Sure, 6 damage per grenade? I'll take! 50 HP disappears from 10-damage charge, 36-damage from a 6-pack of grenades, and the remaining 4 from our gun. You can dig down and bury him to skip his death animation, but this involves killing you and all the lost weaponry this implies. While you can reach the next scene earlier, you most certainly have sub-optimal weaponry that will eat up any time this would save. 1-1 Scene 9 The spawn timer was at 25 when entering this scene. Now, where can I save 5 frames to beat the frame rule? Gah. It's also why we're technically behind Team 1 for this scene -- We've been frame-ruled. Not much else comes to mind about this scene, though. Never saw the agents come from anywhere except by rope from above. 1-1 Scene 10 Another unremarkable scene. That first agent also insists on coming by rope. Only when the scene locks does it allow agents coming from either left or right sides, by my tests. 1-1 Scene 11 For those curious, the flower seen while rolling to the boss scene is an extra life -- Dig under it, and you can have it! Not that it'll help, but still... Anyone knew of this little secret? There are more flowers like it in later stages. I suppose the only scene of 1-1 not commanded by the frame rule would be this one. As for weaponry, I'm pretty certain the spread gun is the best choice. We can try flame/ice, but that tiny hitbox coupled with the fact this boss only allows one hit per frame makes it less effective. Spread is likely the way to go. Also, that stray enemy bullet could have been manipulated away, I believe. Eh... Anyone want to see if there's any way to improve what we got for 1-1?
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
The bulk of my plans are here. More RNG details. Although, these were before the discovery of the agility oddity. The mana bonuses in the TELEPOR route are almost immaterial. At this point in planning, they give practically zero benefit to A. The only real benefit is that we can get away with is smaller Mana gains later. Also, you messed up starting from the very first battle. You don't fight a LIZARD -- They have 40 HP. A GOBLIN has 20 HP, and is oh, so much more capable of dieing in the first round. If C343 was 0x58 from your title screen reset, possible but difficult, a GOBLIN would have showed up both in the plains and the forest. Again, I'm thinking we walk a few steps first, then reset. We have a few spare steps we can pick to try to mess with the RNG here. It's a pretty tight set of rules that needs to be followed. I'll be looking at a few things myself and see what I can come up with. At least remember stuff and get back into it.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Here's a script for you, guys: http://paste2.org/p/1085691 It's tweaked for 1-4, but remove certain comments and comment out some others, and it can work for other stages. For those who didn't bother trying to figure out the script, it is capable of detecting hitboxes and predicts the RNG for us.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Wiki: MoviesList Thanks. This is a good start. And apparently useful in catching errors in names already. Sometimes I feel the only way to get things done is to do it yourself. Maybe what I tried to do myself here got someone to notice, but whatever the case, again, thanks. Now we have something. Even if it is a bit ugly now, that can always change. These lists are far faster to browse through than the full-blown movie lists. I can say I can browse through the NES list without thinking about the hundreds of KB worth of images coming my way. Though I don't get to see the interesting descriptions or images, dropping those was pretty much the point. As for any suggestions I have, my initial thoughts are some sort of table, but I'm not sure what to put in this table. Maybe non-obsoleted published movie count? Shortest reported time of such movies (even if it's not indicative of what the other movies bring)? Would a table add a bit too much vertical space? Might not hurt to experiment with tables.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Yep, this "Frame advance" is the next best thing since we discovered save states! First, we can go back and fix our mistakes without rewinding all the way from scratch, and now we can even stop the game until we want it to run! Is there nothing we can't do at this point?
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Nintendo 64 * 007 - The World is Not Enough * Banjo-Kazooie * Castlevania - Legacy of Darkness * Chameleon Twist * Diddy Kong Racing * Duke Nukem 64 * Gex 3 - Deep Cover Gecko * Gex 64 - Enter the Gecko * Glover ALL * Goldeneye 007 * Legend of Zelda - Majora's Mask * Legend of Zelda - Ocarina of Time ALL * Mischief Makers * Mortal Kombat 4 * Mystical Ninja Starring Goemon * Rocket: Robot on Wheels * Super Mario 64 ALL * Super Smash Bros. * Tetrisphere * Tony Hawk's Pro Skater 3 * Turok - Dinosaur Hunter * Turok 3: Shadow of Oblivion * Wetrix 64 ... Why isn't anyone taking action? Is my suggestion not worth considering? I am taking matters into my own hands at this point. I will keep a manually updated list until someone tells me otherwise. I will produce other lists eventually. For now, the N64 list is yours to play around with.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Okay, now that my activity with the contest is starting to die out, what with its closure coming shortly, I will have time to finally finish this TAS once and for all. ... But I need to refresh myself on a few things, first. And especially work on the recent discovery on the fact the fourth character needs to be slower than the first, regardless of being alive, to work around the crazy turn order stuff. As for naming, so far the votes are pretty unanimous towards having some form of recognition rather than a nameless crew.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
Netplay is certainly nifty to have, but I want you to try a different project, a broadcasting system. It's simply a request, but one I hope is an intermediate step on the way to full netplay. One-way data transfer might be easier to handle than two-way. We still might have a problem with things like SRAM or converting savestates around. Maybe tell the client user to start a new movie, from power-on, clear SRAM, and just send the entire movie's worth of input over. The best case would be if one can broadcast a TASing session. The server doesn't send state saves/loads, just tells the client to produce the appropriate saves. After all, if the input matches, both server and client should be able to produce the same states. Low bandwidth usage would be nice. If you're going to use FCEUX as a starting base, use emu.registerbefore for a function involving joypad.get on the server side, and "boundary" for joypad.set on the client side. That should avoid any nasty problems with input coming a frame late, particularly since we have multiple emulators running and we can tell them to delay as much as we want until the next frame is sent. While I like to say I'm decent at using lua scripts, I haven't dealt with networking. I might want to see if I can at some point. Would love to help, but I'm not sure where to start.
Editor, Experienced Forum User, Published Author, Skilled player (1173)
Joined: 9/27/2008
Posts: 1085
What's that?! Someone reached the title screen already? It's clear now, this is an indication the other teams are doing something! We need to hurry!