Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
Inzult wrote:
if we're moving to 3 categories, i think it it should be 1) as glitchless as possible 2) anything goes besides l&r 3) anything goes
this also takes a bit of work. As glitchless as possible might well be slower than the published movie, even.
I think if you're going to go this route, 2) should be "anything besides l&r or scroll lock." That way you actually get to show off the small glitches. The big glitches possible with l&r or scroll lock just drown out all the small glitches.
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
In my testing, no. You can't ever change the connectivity of rooms with this. It only changes where you land in the overworld, so you can't skip anything inside a dungeon or other sidescroll encounter.
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
Very interesting that it can be done without fairy. Doesn't change my opinion of the live run category ruling though.
It's fun! I just see it as a wrong warp, since you're getting the game to take you to the wrong tile.
If people really want to run it live, I'm suggesting a new category for it. In fact max, skav, jn, and I just hashed it out: Any% No Scroll Lock. Early hammer, thunderbird skip, these new skips... just not the scroll lock which allows early p7 and the barrier skip.
As for TAS categories I'm not sure...
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
If there's one thing I learned from my forays into emulator development, even from the perspective of someone doing ports, it's that things are complicated for a reason. Optimizing for state saving probably would have costs elsewhere.
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
nanogyth wrote:
Finally found something that changes the number of enemies killed by UNDEAD.
Seems to depend in some way on the overworld tile that the battle happens on.
Now to find tiles that hit all 8 (should be ~%10) and see if it is constant battle to battle.
zip of 2 fm2s with different kills at end
Wow, it's not that each monster gets a fixed chance to survive it? That'd be useful to know!
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
Was trying runs and it turns out that the ship glitch in Ambrosia isn't QUITE as reliable as I thought. It just wasn't glitching tonight, and I had to go to the Shrine of Wisdom and come back before it would. So.... there's still a mystery there.
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
The enemy freezing is not the Time Lord. It's going to the status screen. If you go to the status screen, hit b to exit, then hit the D pad, then enemies on the current submap (only works on submaps such as towns, castles, and I assume Ambrosia) will NEVER MOVE AGAIN until you let go of the D pad. So broken. But broken in a weird way I don't honestly blame Ponycanyon for not catching.
It was only a stroke of luck that I realized the difference between what the TAS does and what I typically do in live runs: the TAS moves optimally. It never let go of the D pad. Once I did the same, boom, jackpot. You can even run into the wall continuously, just standing there banging into it, standing right below a high-level Castle Exodus monster, and never get attacked. Here's the highlight.
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
nanogyth wrote:
presjpolk is working on an RTA of this and he ran into a glitch.
At one point the ship he needs to fight is already defeated.
I was able to reproduce this by casting a spell in the same spot he did.
I have a theory on this: The ships in Ambrosia will not auto-attack you. I suspect they're not 'real' pirate enemies, but specially coded ones. And they have a ship pre-spawned beneath them, that you just can't get on board unless the monster above it is gone.
BUT, if you have the map regenerated somehow, apparently by casting a spell, it flips the order of the items on that square, and the ship ends up on top. Sounds to me like the fixed pirates in Ambrosia were themselves coded to use an in-game glitchy treatment of two things on the same square, and we're just breaking that glitch to act differently.
EDIT: therefore, I wonder if hitting the Status menu would also trigger it. Or... whether casting a spell/hitting Status would influence monster AI in Exodus Castle. Worth trying.
EDIT 2: The game decides which to put on top, the ship or the fake-pirate, when that square is scrolled on screen. Scroll back and forth and it occasionally will give you the ship on top. So stupid simple.
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
Thanks so much all. I think I actually found the code I was looking for. Wow.
It's so handy, this ability to get a log of what was running, dumping the instructions along with notes on what exactly was put into the registers, what addresses were references, and everything.
That made this pretty easy, once I got it going.
Oh man, and I just set a RAM watch on the byte this code leads me to believe was the PRNG result. And sure enough, I'm missing when the code leads me to believe I should be missing. Perfect.
Thanks all.
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
I've written 6502 assembly for the NES actually. Years ago, a little bit, and only emulator tested, but I did it. :)
So great. I know my project this weekend then: explore this tool.
Thanks!
Joined: 9/21/2012
Posts: 11
Location: Northern Virginia
What's a good disassembly tool for NES games? Is that built into fceux?
I'd like to know the combat accuracy formulas for a game, and being able to disassemble the combat code would obviously be a huge help for that.
Thank you!
