Posts for TheAxeMan

1 2
13 14 15
39 40
Posted: 9/17/2010 6:47 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Looks good. It would be nice to know a little more about the glitches used here. Yes vote.
Posted: 9/17/2010 6:45 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Powerful glitches! Wish I knew a little more about how this worked, the submission text is lacking. But the run is good, I vote yes.
Posted: 8/27/2010 2:53 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Those 15 seconds were what held back my vote before so I'll vote yes now. Actually, that's even more than I expected. Pausing to manipulate luck was a great idea, I am wondering if it could be used more. Compared to the last run it looks like you fixed grabbing the ropes. Even if more time was saved in stage 3, stage 2 looks most improved. I still wish the original (U) rom was used but this is probably fine.
Posted: 8/20/2010 6:03 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
I wasnt able to get your test run to compare it, since the link give a 404 (in the old topic in gb forum).
Sorry about that but the link I just posted above uses microstorage and should work. You can grab the rope lower in both of those places by jumping. You do need subpixel optimization but it is not hard. Well, I guess less lag is good, but I like to work with the classic, original versions.
Posted: 8/20/2010 7:27 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Huge improvement over a rejected run is not an immediate reason to publish. The rejected run used the original Castlevania Adventure (U) game. This uses a super game boy (E) version. I think that accounts for a large amount of the time difference. The boss fights look much different and I think the lag pattern is different. There are several places where the rope wasn't grabbed as low as possible. For example, at 5:09 you should have jumped. Same issue at 14:54 too. A while back I made a test run on the original version but only finished the first two stages. The second boss isn't well manipulated. I have no idea how my attempt compares to Phil's or MUGG's. Difference in lag pattern could easily influence some decisions so it's hard to compare to this run. I vote meh because I want to see this done on the original version.
Posted: 8/18/2010 4:18 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Is this specific to NES or should it also apply to SNES and others? I ask because ZSNES has awesome sound, even in linux. My computer is quite old now and is still able to do fine with lower resolutions under windows. I do still get some crackle though. I am trying out Linux again and the Linux version is much worse. It seems to help to up the buffer length. This can completely take out the crackle but then the sound lags. I also tried running the windows version through Wine but my computer is generally not beefy enough to handle this.
Posted: 8/12/2010 7:12 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Ok, I put some work into making a hand axe run. Here's the progress up to Astos. http://dehacked.2y.net/microstorage.php/info/900272950/FinalFantasy-2.1-noluck-duo.fm2 The faster fights against the wizards and Astos are quite stylish. Unfortunately, it doesn't make up for the time needed to buy the hand axe in Pravoka. It's close, only a few hundred frames or so. I had hoped that the change in flow would let me cut a fight out. I can but it's no good because I would see wizards in Earth Cave that I can't run from. The best I can do is to save time by sailing all the way into the docks. I also considered fighting the wizards solo. But it takes time to manipulate the pirates battle so two guys die. Beating the wizards requires visiting the inn and getting the hand axe. Plus the battle takes 3 rounds. So while it would be cool it's not faster. I did discover a technique to start dialogue and enter the menu one frame faster each time. Also investigated whether it might be faster to forgo a point of luck to beat Garland faster. After pirates I am about 300 fr ahead this way. But both of these could easily get lost in luck manipulation. So I might investigate but any improvement is going to be pretty minor.
Posted: 8/6/2010 6:20 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
I tried out coral sword on a few bosses. The extra crit chance might give a small benefit to luck manipulation but because of how the RNG works it doesn't provide any increase in damage. It would add more steps in Earth cave to get it. Another issue is the fact that I need savegame items to manipulate luck. There is a cabin right next to coral sword but there is a cabin and a house in the dragon sword treasure room, plus it's in dwarf cave where there are no encounters. I need both of those before reaching Crescent Lake where I could conveniently visit an item shop. Gaia's item shop is even less out of the way so ideally I'd want to shop there. Anyway, coral sword has long been on my list of possibilities but dragon sword seems to fit in better overall. Elemental weaknesses and monster type weaknesses are two different things. There is a separate byte for each. Lots of weapons have monster type bits set and those don't work. That includes dragon sword vs dragons, coral sword vs sea monsters, etc. But only those 3 weapons have elemental properties. Actually, stunning or sleeping an enemy would be great for luck manipulation in general since I'd be able to manipulate higher damage. But this would require having a mage and buying the spell which takes time and that char will add time running from each battle. Still, it's interesting to consider. One issue is that enemies always wake up from sleep on the next turn and HOLD is a level 3 spell. So this is also a long shot. Another interesting idea is making the second char a thief. Early on he'll help run from battle and later he'll just be there for manipulation. He would probably die on Kraken since he wouldn't be helpful after that. Although he could increase the chances to get a preemptive BANE. I'd need to check if just having a second char is helpful though, it may only be useful if he takes a hit. Problem is that even though it seems like there isn't much difference in ordering two chars to run from battle it adds up over the large number of encounters. He also has to do something during boss fights, adding more time.
Posted: 8/5/2010 10:57 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Wait, Defense sword can't be equipped by an unpromoted fighter. So that's definitely out. But yes, it's not a problem to simulate both routes or any changes. I still have all the tools I made for the current run to do this. For encounters there are still some details I know about that I haven't seen in any FAQs. Someday I'll get that info out there. Anyway, I can simulate encounters with 100% accuracy so I can figure out how things are affected without having to play through. Edit: A quick go at Kraken with the light axe determines that I could save only one or two turns. There are two light axes in the Sea Shrine and both are about 40 steps out of the way in territory where unrunnable fights are a problem. Of course I wouldn't need to get flame sword but flame sword is only 7 steps out of the way and fits in very nicely to the flow. So I might check how this affects encounters but it's unlikely to work. Another bonus to flame sword is that it hits the elemental weakness for the Blue dragon in Mirage. Monster type weaknesses don't work but elemental ones do. So flame sword does extra damage vs blue dragon while dragon sword does not. Furthermore, resistance to an element doesn't affect physical attacks, just magic. So flame sword does normal damage to Kraken even though he is fire-resistant. Only three weapons have elemental properties: Flame Sword (fire), Ice Sword (ice) and Xcalber (all elements!).
Posted: 8/5/2010 7:09 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Ah, defense sword is worth thinking about. But doing waterfall first may not be practical due to how encounters work out. It's actually very difficult to avoid non-runnable encounters in the Sea Shrine. Difficult to plan, but that route change is likely to add more encounters to run from. If possible, confuse staff could also be useful for manipulation. Light axe could also be interesting but I doubt it's a big enough jump from flame sword to be worthwhile. Sorry, I didn't explain that the table is all raw random numbers. The number from that table is used to get a number up to 200. Here's how it works: 
randInRange(min, max) = ((1+min+max) * rawRand / 256) - min
This is an assembly level function used by just about every random roll in the game. There is a clever sub function implementing 8bit * 8bit -> 16 bit multiplication. Dividing by 256 just means taking the high byte of that result. The hit roll is calculated as randInRange(0, 200). When that number is <= the crit threshold / weapon index then you get a crit. By the way, this same hit roll calculation is done for BANE. The glitch is that a hit roll of 0 succeeds on an immune target. If you look through that table you'll see 00 twice and 01 once. Those are the 3 in 256 values that get you a hit roll of 0 that lets you BANE an immune enemy. Well, it seems like it should be possible to beat Kraken faster but it's not going to be easy.
Posted: 8/5/2010 7:22 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
I am looking at this game again. For one thing it would be nice to get this in FCEUX fm2 format. Also, I have some ideas but they are probably small. Mainly I was looking at the random numbers list: 
   AE D0 38 8A ED 60 DB 72 5C 59 27 D8 0A 4A F4
34 08 A9 C3 96 56 3B F1 55 F8 6B 31 EF 6D 28 AC
41 68 1E 2A C1 E5 8F 50 F5 3E 7B B7 4C 14 39 12
CD B2 62 8B 82 3C BA 63 85 3A 17 B8 2E B5 BE 20
CB 46 51 2C CF 03 78 53 97 06 69 EB 77 86 E6 EA
74 0C 21 E2 40 D4 5A 3D C7 2B 94 D5 8C 44 FD EE
D2 43 00 BB FA C6 1D 98 A0 D3 54 5F 5E DC A8 00
AF 93 A1 E1 6C 04 DE B6 D7 36 16 C5 C8 C4 E4 0F
02 AB E8 33 99 73 11 6A 09 67 F3 FF A2 DF 32 0E
1F 0D 90 25 64 75 B3 65 2F C9 B0 DA 5D 9F EC 29
CE E3 F0 91 7A 58 45 24 1C 47 A4 89 18 2D CC BD
6F 80 F6 81 22 E9 07 70 FB DD AD 35 A6 61 B4 A3
FE B1 30 4B 15 48 6E 4F 5B 13 9C 83 92 01 C2 19
7F 1A 1B 71 B9 3F 4E 9B BF 9E 87 0B 10 57 F2 26
79 9A 05 C0 E0 F7 4D 7D CA 52 9D F9 BC AA FC 8D
7E D1 A5 42 E7 D6 76 A7 84 8E 66 7C 23 88 37 49
D9
Since this is the sequence in which the battle RNG picks numbers it is the key to figuring out what is possible. Low numbers get critical hits, 00 and 01 make BANE work and high numbers do high damage. The game does a hit roll and then a damage roll and repeats for each hit. Early on the best spot to land is 05 for the hit roll and C0 for the damage. That will be a high-damage crit. Later with two hits the best sequence is 01 C2 19 7F. That gives two critical hits and is the only possibility to do so for most weapons. Critical hit rate is based on the index of the weapon, so later weapons generally have higher crit rates. The hit rating has nothing to do with it. Anyway, the RNG is cruel in that I can't land on those awesome values in a one-on-one fight. If I could, most fights could be shorter and I could really save time. I am pretty sure it's not worth it to keep another person alive or revive them just to fight a boss. There aren't many boss fights so it's probably close to optimal. That's too bad because I could take out Kraken in 6 turns instead of 10 and more. So far the most promising idea is to buy a hand axe in Pravoka. The hand axe has one more attack power than short sword and this translates to four more max crit damage. That crosses a threshold that allows me to kill wizards in one hit and Astos in three. I'm pretty sure that's enough to cover the time needed to get it but if luck doesn't work out then it might be close.
Posted: 6/29/2010 8:42 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
I have been making steady progress and reached the first mountain area. Biggest change is that instead of visiting Brynmaer right after the sealed cave I go straight to the statue and walk back to Brynmaer. Also found ways to take out those enemies faster and have much less lag. Gained about 5 sec in this area vs test run for a total gain of 1400 fr over test run.
Posted: 6/18/2010 4:28 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
This has come a long way since your first attempts. I bet there's still some improvement possible with better manipulation but this is easily good enough for a yes. The Dragon Warrior cameo is classic, thanks for putting that in.
Posted: 5/20/2010 5:38 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Ok, as I said I have been making slow but steady progress on the final version. Here's how it goes up through the sealed cave. http://dehacked.2y.net/microstorage.php/info/75625300/Crystalis-Final-toBrynmaer.fm2 A big reason why it is taking me some time is because I found some new strategies in this beginning area. This puts me 17 sec faster than the test run which was already 17 sec ahead of the published run. About half the improvement is in getting the xp for level 1 and half is in getting the xp for level 2. I am more confident in the test run flow later on so it should be faster and more straightforward. But I'll continue to look for new strategies and tweaks. At the very least I am saving frames everywhere due to less lag.
Posted: 5/2/2010 1:42 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Awesome job, this has it all. I always thought the airwalking could be better.
Posted: 4/17/2010 5:22 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
I've started but not gotten too far yet. A little bit faster with some ideas that could save more time later. But it should continue smoothly as I get into things. Thanks for checking up.
Posted: 3/28/2010 5:37 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Well, there are definitely limits to how well some of these can be manipulated. Each character who uses multiple elements in a battle seems to always get at least one going up. If they use two then it seems possible to manipulate both going up. I haven't checked whether you can get 3 elements going up for one character. With two characters each using two elements it seems difficult to get all four elements increasing. So I wouldn't expect to use many elements in each battle and get all of them increasing. Even if you could you might have to set up the final round of battle to be really simple to manipulate it. In calculating steps to next battle I mentioned there are two components. The part that can be manipulated only varies by a small amount, less than 10. Plus, this amount is based on the seed at 0024. That doesn't change during the battle. So on the instant each battle spawns all of these are set at once: -Enemy group -Element of surprise -Steps to next battle It is probably impossible to optimize all of these for every fight. But you could get more control by waiting some frames at the end of a battle and maybe some more outside of battle.
Posted: 3/28/2010 9:22 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Ok, I got around to doing some analysis on this game. Since there's no debugger on snes9x I couldn't figure out exactly what is going on but I was able to learn a lot with RAM analysis. RNGs: There are at least two different seeds used in random number generation. One is at 0023 and one at 0024. 0023 is most important seems to control everything in battle. 0024 doesn't change in battle but does change in outer world. The seed at 0023 increments every frame when ordering your guys around but during the battle it only increments as needed to get random number for hits, damage, etc. In the outer world it and 0024 change every 17th frame when 1884 hits zero. But that change doesn't seem straightforward and I can't see the pattern. Encounters: Set a memwatch at 7E1868. As you walk around this value decreases. When it hits zero, you get a fight on the next step. This gets set when you leave a town or just before getting control to move on after a battle. I can't figure out the exact algorithm for how this gets set. It seems to have two components, one more random than the other. The more random part uses the RNG seed at 0024. The less random component seems to cycle through a list of numbers around 10, 20, 30, 40 each time a new step count is generated. Spell affinities: When you use multiple elements in one battle, the increase at the end of battle is random. Unfortunately the determination happens just before the "enemies defeated" message and so the luck for this is tied to the luck in the last round of battle. Still, you should be able to exert some control over what stats go up. For example you can make sure fire always goes up to power up most of your attack magic. Other manipulation: The random counter does increment during the messages after battle. That's why you can manipulate levelup stats so easily. You might be able to use this time to manipulate your luck for the next battle. You would have much finer control due to the fact that luck changes every frame vs every 17th frame. But there seem to be some other factors in how the enemy group and surprise value are chosen. During the Dal Gren battle where you die in the beginning you don't gain fire affinity for using magic. I didn't look through all the battles for this though.
Posted: 3/22/2010 7:51 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Excellent ideas, I hope this works out. If it compares to FFL1 then running is faster than doing a reset, but not by much. Also, I found that timing the soft reset gave me more options in manipulating luck. It might be necessary to get some things to happen. You may also be able to spend a few frames to manipulate longer periods between battles, saving lots of time overall. Manipulating the mutant's abilities is going to be difficult but should be doable. I understand most of the monster changing chart. In FFL1 the system can be abused to get higher level monsters from any level meat. But this system is different. You can't get Evil Eye without meat of a high level monster. To go from Baby-D to Medusa in one meat you need meat of Silver or Griffon. Jaguar won't work because of the A/B/C modifier. Similarly, Medusa becomes Evil Eye with meat of Demoload or Sylph, both endgame enemies. What about the Xcalibr from Final dungeon? It always hits too and affects a whole group I think. You can buy catclaw in Final town. Best way to get the gold for it is probably selling treasure picked up in dungeons.
Posted: 3/12/2010 9:17 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Janus, you are a beast! This game is awesome and it would be great to have a run for it. There are some technical details that would be interesting to know. Especially how to manipulate to encounters. On gamefaqs the savestate hack guide has some nice info on where stats are stored. But when you look directly in the memory the offsets are offset. Anyway, 7E1612 has your fire affinity and 7E1617 is sp affinity (For first char. Other chars will be after this. These affinities are 8-bit values and it seems that they always go up by 2. Now it looks like when you use fireG you sometimes get fire and sometimes sp. Sometimes both go up. I wonder if you can always manipulate that? If you have to fight a lot of enemies, manipulating higher spell stats could make Chezni quite powerful. The final boss would also be a lot easier since only his magic does any damage. I'll see what I can find out about encounters. I haven't done assembly analysis on SNES before though.
Posted: 3/12/2010 8:14 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
I was glad I watched this again with FreshFeeling's commentary, it was really insightful. He answers boct's question and noticed the same thing as me about the final boss's final form being female. But mostly his explanations of the battle tactics, glitches, strategies etc was very interesting. Probably more interesting to a TAS addict than a casual player though. By the way, the recommendation to turn on the input had me noticing that turbo was used on dialogs. So another improvement would be to optimize that. After hearing all of the details I'm even more amazed at how prolific Janus has been. His runs are quite high quality especially given that he isn't disassembling RNGs.
Posted: 3/9/2010 7:48 AM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Excellent. I haven't played this in a while and for some reason I was entranced by the imagery of strong women. The writers and artists definitely have some sort of fixation on this. It starts out with a big sister figure for the hero. Then there's a super-strong bio-engineered chick who tears enemies apart with just her claws. There's an android chick with a bazooka. To top it off, the most powerful final-final-final version of Dark Force has a feminine body after every previous incarnation was male.
Posted: 3/7/2010 8:10 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
Now that I think back, the TAS is actually much less long and boring than playing the game normally. I played through this myself a while back and it is really just one levelup with monotonous dungeons that you really can't get through without a map.
Posted: 3/7/2010 8:04 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
This is awesome, as usual. I do have a question though. Why don't you get Brwn in Melenam? It's not far out of the way and would up your defense, perhaps increasing your possibilities on some of those boss fights.
Posted: 3/7/2010 1:44 PM
Quote
TheAxeMan
Ambassador, Experienced Forum User, Published Author, Experienced player (698)
Joined: 7/17/2004
Posts: 985
Location: The FLOATING CASTLE
The place to put the organized info is in the Game Resources. If you make up some notes we can put them up on a SoM page there. By the way, what version should I run HHS's WIP? 1.43v16 and 17 both desync with default settings.
1 2
13 14 15
39 40