The thought occurred to me that, if you're dedicated to only having seeds possible normally (after the current version was released), you could still use the entire seedspace thanks to the eventual overflow.
This would be slightly more interesting if you posted the odds of generating a blasting artifact on the upstair. Break it down if need be
an item on the upstair (...items/squares in rooms?)
a weapon(*10%)
is it an artifact?(*5%)
which artifact (8/19 will kill a neutral, if I understand it rightly, one of which(Stormbringer) bypasses the 1/4 chance by being intelligent. Grayswandir would then be 1/19)
(I'm not intimately familiar with the item generator, so this may be wrong-ordered, but spoilers are well-written.)
am I reading correctly that Excalibur can't be randomly generated? It would make it an even 20 artifacts generatable.
For added value to other watchersreaders of this run, please note that most Nethack players consider Grayswandir to be the best weapon offensively in the game, dealing double damage to everything, as well as silver damage (+d20) to most demons and undead.
Well, whenever you throw a barrel, it will go to set positions, but it doesn't change when the barrel has been thrown, so you have some few frames to choose between while the X-position of the barrel (or in this case, the bonus) change.
If I remember correctly, two-player co-op mode doesn't really do anything special but I think I'll take a look, the BADBUDDY code might be interesting as well.
(snip)
Here's a list of all the exits
It was linked in the submission..
Checking what happens if you throw and can switch while the barrel is in midair seems necessary, aye. It seems like the most likely way to get to +E0 if it adjusts X position to the new active kong (have diddy half a screen behind?).
BADBUDDY...makes it so player 2 can kongswitch? The description is kind of vague.
second thought: did you determine that +E0 doesn't crash by hex editing?
Oh and I forget exactly how it works but if you beat expert does is carry on straight to master? If so I'd love to see a run of expert and master combined.
SMB1:
Finish Expert w/o continue => game proceeds in Expert Extra
Finish Expert Extra still w/o continue => game proceeds in Master.
Finish Master w/o Continue => ??? (I don't think it's been done. Master 3/9!!!)
So, pretty much any Expert run will need to go all the way through.
And unlike in SMB2, you can't begin from Master even once you've been.
Score would be boring often (and would necessarily be Warps). Now, interestingly, for Play Points it is more efficient to NOT warp.
I forget how bananas score, but time score iirc is always 2x hundredths if you have more than half, and 1x else, plus a big bonus on warps.
If I remember correctly from a discussion on SCUK (samus.co.uk, no longer available) there were nontrivial numbers of enemies taking nonstandard damage from various attacks, having weaknesses, strengths, etc.
Of course you should remember to boostball Meta-Ridley when appropriate.
I don't think 21% has actually had a completed run, even if every room has been shown to be doable.
Looks good, I see how heavily inspired by this game La-Mulana was.
Then I remember the difference between that team and Konami when suddenly BULLET-SPEWING MOAI HEADS.
What other subweapons are there? just different flavors of arrow?
Is the Holy Water in L1 out of your reach?
The bonus token room you mark as debug is still accessible through a cheat, mind.
So if the X vs character is what determines the exit...I'm not entirely following on why you can't throw to have this reach positive E0 or whatever for Necky's Revenge Because it activates as soon as you let go of it?
help me out here.
More important edit: what happens if you perform two-characters-active glitch when in two-player co-op mode?
For some reason I recall reading that the RNG for this game acted differently on different GB hardwares (GBA vs GB vs GBC, which GBC palette you chose too) due to pointing at an area of memory that acted differently between them. I was going to do this because I had a save right before Dunatis, but I got a little too happy when I rolled up Teleport, so I got a dragon-raced save.
notes wrote:
GBA
3: +HP (9)
4+2=6: gain Gaze (xD0)(18)
7+4=11: gain nothing (29)
12+2=14: +hp18 (38)
39+6+4=49
+6+1=56: gain Cure(x99)
57+6+3=66: gain Teleport!
So, getting +HP after 3 attacks killing Dunatis (thus, 6 rolls for turn order = 9 since poweron), followed by a reset without walking, load, attack without walking.
battle 2 having two attacks (the 4 is before I was considering the turn order, but counting the roll-for-ability as 1), gaining Gaze
3 had 4, gaining nothing
5 had 2, gaining 18hp
6 had two combat rounds, in which there were 4 and 1 rolls, after which I gained Cure (which ability's value is 0x99)
And, dunatis battle seven took three attacks, after which I gained Teleport.
edit: This was counting how many events should be calling C0A5 according to the Game Resources page, i.e. the turn order (6/combat round), how many damage events occured, and then adding one after each battle for the Mutant ability. (hmm, problem: Does it roll the "which ability" roll on C0A5 if it doesn't roll that an ability is gained from C0A6?)
Party composition was robot-monster-monster-mutantF, if that helps. I didn't note abilities, unfortunately. (C-class monsters only, though.)
Is this useful for deciphering the initial state of RNG on actual hardware? Any suggestions on how to further test it? (Saving in front of a boss battle, esp. Dunatis for being an easy B-class monster, seems like the best recipe, because it doesn't rely on random-battles, and gets the full range of results for the after-battle roll.) Clearly I should better-document it, who goes when for how much damage (or miss), what stats and slots were. And do at least two trials per system, though anecdotal evidence (normal speedruns) suggest this not to be necessary as it's static per system.
even more edit: I could potentially test all of GB(two of 'em), GBpocket, GBC, GBA, GBA-SP, SGB, GBPlayer, though I'd probably want a new set of GBA values at a comparable save.
Also, I'm wondering- once you finish the run, I can't think of any technical barriers preventing a human player from duplicating it. It is worth producing a minimum-turn run somehow though.
Does nethack save the RNG state when you save?
Right, that. Can't adjust vertical position, but I expect that's faster than using ability-movements. Would need some luck manipulation or just good timing to maximize fish value... Pterogliding is less useful but might be handy on Neo Star 1.
Comicalflop wrote:
Ice+Rock= great for downhill, it will run out of speed so you have to restart the spinning sequence again
indeed, it's faster than running if you dash-jump curl. You can cancel fast enough to still do it repeatedly in, say, the falling platforms level on Neo Star, the only level I think you'd definitely use it in, as you need drill(spike+stone) and can switch to curling stone(ice+stone) after getting that shard without leaving.
Jumping into something seems to do a half point. Do any other attack modes vary in terms of damage output? I haven't seen anything else that seems to do less than "kill normal enemy" damage (1).
(Not including the inability of vanilla kirbymoves- slide/puff- to damage bosstypes)
The RNG advance without step/encounter appears to be any terrain that you could walk on, if something were different (being on a lower level for water, the rock being destroyed).
Enjoyable. Would be more enjoyable if the music didn't restart after every battle.
What happens that completely changes all the memory blocks after you expand the inventory well into the characters' data?
I'm still convinced that the "End Battle" command is a debug command, otherwise inaccessible that you reach by using garbage character classes, since it performs a full restore on your party (more visible in the incomplete testrun,than here, which has multiple character deaths in the Dark Crystal World[name?]).
It's great that it has a "Light" Warrior (those four old guys) for overworld sprite, so a "Light' warrior beats the game.
(yes vote)
According to my calculations, Santa Claus would need to complete his evaluation, manufacture, and delivery for an average of 220 people a second to have everything delivered before the next Christmas. More like 68400 to do it all on Christmas (Remember, there're more than 24 time zones).
Guy needs a bit more than a wallzip glitch to pull it off...
Merry Christmas to All, and TAS all a good night.
Oh, useful! Thanks. Of course, this doesn't address the Pokémon version (and, say, Hitler no Fukkatsu/Bionic Commando) issue.
I'll be snookered, http://tasvideos.org/movies.cgi?name=mega+man appears to catch them all. I editors fixed up the tagging issues, or I needed to put in the search differently. Suggests that region differences will be caught successfully, though in the HnF case you'll just pick it up by obsoletion chain.
A thought occurs to me, though NESVideoAgent's auto-ROM-matching is gone, that the ROM classification databases be used for the purpose of "which games are the same". Probably overkill. (And, mightn't catch Pokémon Yellow with R/B)
That isn't a legal maze layout. NetHack's maze generation algorithm is rather restrictive.
How about making the letters separately? The point of those stairs being there is as a reminder that there is one level where you will be three locations at least (Vlad branch). I know narrow S is possible, and a T should be doable if just by lighting hijinks.
is uppercase A impossible? I suppose a narrow one wouldn't work either as pillars won't generate, but an umber hulk or woodchuck could fix that quick in a pinch.
Would a lowercase one, a la
.....#
####.#
.....#
.###.#.
.......
be valid to generate? (Dark spaces won't matter)
edit: Hmm, any benefit to getting the amulet-bearing monster (if you get one to take it off the High Priest) to nab a stack of C!oGL? Free evasion of mysterious force?
Watched, was mildly interesting. Youtube encode desynced some clear minigame-to-the-music in the second part, but this may just be GNASH doing stuff.
Strong meh for triple-fetch-quest platformer.
Kirby will get the power-up back in his mouth, which can't be done without this trick.
Actually, when holding something above head, pressing down will put it back in mouth. Probably slower than using the room transition like that, yes, but stilll. And, when an enemy is held overhead, you can throw it forward or up.
Many enemies have interesting functions when pulled out and held overhead, such as shooting projectiles, causing a jump, giving gliding-fall motion (pterodactyl)...I don't recall any being TAS-useful, but I haven't played in a very long time.
Oh! I know what you can do to be entertaining. Manipulate one of the dungeon levels to generate looking like a word. (Probably use a MMap scroll on it.)
Is that even possible? I can't think of any dungeon generation algorithms that would produce anything looking even approximately like a word,
T A S should be easy enough. Branching corridor, corridor up to room on each side with corridor below room, winding corridor+room. Or, in maze,