So are we using top left = 0,0 or 1,1?
Because there can always be improvements on the previous levels that may affect the RNG for the levels after, it seems very useful to have many alternative routes already thought out.
As for the new level 1 route, looks like you could manipulate the trackbot to go left off the ladder next to the purple tp after picking up the gem above that ladder by curving slightly towards the left. Looks like it wouldn't even lose almost any time. I wonder if you could also manipulate the trackbot to go down the first ladder it encounters.
In fact I wanted to test that. Are you guys using the "jetpak15" rar contents unmodified?
Also, Nach, would it be possible to have a Jetpack Game Resources page? Does it take an admin to do that? Didn't see a way.
I'm surprised to see the death exit bug this early by the way. :)
This page should have a poll in the first thread with a list of all the suggestions that have come up so that people can go click on their favourite candidates (games they like/what they think would make a good TAS).
Anyway, these haven't been mentioned yet I think...
Dungeon Master 1/2
Eye of the Beholder
Lands of Lore 1
etc...
Should make for interesting TAS'es because they allow fast fast motion/spell casting etc., but you might also just be able to find sequence breaks for some of the puzzles.
I'd like to add I personally don't think it's a good idea to have more than a 2-3 of levels up on the "bulletin board" at once.
Did you know you can jump at the start of levels even if you spawn mid-air? At least I think you can. Anyway when we have the trick page up it will make this page more neat, because we can easily go see what's already known and what's not.
I agree with Mojo. Going left from where you land after the tunnel looks faster on paper at least.
The hintbook has an older-version picture of at least the very first level (one with the two trackbots instead of trackbot and spinner) so... we'd have to be careful using those even if it's a good idea to use it as a source. It probably wouldn't be a problem to just take screen shots like I did (I did the first ten levels cause after that you can warp directly ahead) and in the process you might even come up with some new ideas.
In fact I forgot to mention: Did you know that the trackbot's AI is not all deterministic? Just watch him move around in different situations, he will eventually break his pattern and go on an adventure. If that was to be manipulable, it could be a way to avoid an obstacle in some level.
Looks like other buttons make the start of the level faster too, but maybe up is the fastest. Also returning back to the "let's use an algorithm"-type solution that I was originally speculating about when this thread was revived, have you considered using simple heuristics for calculating an expected time for each route? All you'd need is an algorithm that determines the time it takes for moving across a certain horizontal distance w/ or w/o jetpack, for falling a certain distance, for phasing a block etc. It wouldn't produce solutions but it might be faster as a method of testing them to some accuracy.
Anyway, I'm sure we'll find more ideas collectively and it certainly makes sense to make it into a more collective project. In fact that's an idea I've had already some time in the past: why couldn't more TAS projects be more communal? You could even have a whole page devoted to a single game with easy-navigation info about the game, a map, relevant pictures etc. Cause that would also leave the all the relevant information available for future runners. So more wiki-style. Maybe there's things like that already going on now. I have made a couple more observations about the game that don't relate to either of the first two levels.
So if someone created the tricks page for this game, we could start listing such things in there as well as mentioning them here whenever they seem relevant. Also making one neat folder of all the route pictures when no-one can optimise them any more would sound smart.
Is there a chance that the simulator doesn't take into account all the weirdness that the Supaplex engine can produce? Did you check out this site? http://www.elmerproductions.com/sp/tricks.html
I'll be very very interested in this run if someone's doing it even if not able to comment on the solutions or anything.
http://www.sendspace.com/file/x0oco9
I created a picture showing the route. No idea why it's not showing.Can anyone name a safe simple file upload site?
The speed of the Internet o_O
Do you think your route is optimal? Would it be useful to create a new picture per route so it's easier for others to comment? Although I think, by the end, you're probably going to find methods to create very good if not optimal solutions that others won't have an easy time commenting on.
By the way, I'm not sure what is more standard in coordinate systems, I was thinking of the screen as having the origin on the bottom left (like the Cartesian plane), but the principle is the same if you place it on top left. And yes I meant using negative numbers as a short hand.
I have the setting turned on to receive notifications for replies, but I didn't see any in recent times, not even in the junk folder, for posts I had been following.
What might be causing this?
Shows 17.92 s, weird how the avi is so much longer. Although the game speed looks fine in it.
Could the discrepancy have anything to do with the way the game's speed controls work? Because weren't those being emulated wrong as well?
Hope it was just the FPU.
I don't know what Ganon's movement algorithm is, but in this video Ganon always fires his fireballs from the same positions in the same order.
Plus the stats are the same."
not having seen the original videos, the guy may not have been claiming to be recording off a console, but rather using a save state to get back to the start of the fight quicker on his emulator. Maybe that would result in the same RNG as well? Don't know how it goes in this game. Or can you save the game next to the boss room?
There seems to be a huge discrepancy between his "outtakes" and the winning attempt. There were no near-misses so to speak. This by itself is not indicative of cheating, it could be just someone thinking lots of attempts that failed early would be a better outtakes reel than a few that made it far in.
However I get the overall idea that it is somewhat reasonable to suspect these could be savestated. Remember the guy that used to stream playbacks of TAS'es and pretend like he was playing? There's plenty enough innocence and naivety out there to make those people prosper.
Firstly, I'd like to suggest using coordinates like these when discussing the levels: (x,y) where either can be negative to indicate starting counting from the right edge / top of the screen. Having a picture is very useful, if it's to show a whole route like BrotherMojo did.
Is full-speed flying faster than running on regular bricks? I think Mojo is correct about landing to reset momentum. Could it possibly be faster to:
a) turn back after picking up the "2" gem (from Mojo's drawing) and going under the platform to get the next gem (at coordinates 6,8)
b) turn right or burrow up to after picking up (12,-3) to go to (14,-1), use the conveyor belts to the next ones on the left quickly and picking up the one that's left afterwards.
c) phase through (-3,4) instead of going round the platform after picking up the tank.
d) lose height, when necessary, either all on one go or in smaller quanta? E.g. the last flight to the door.
I don't actually think any of these would be faster but it's the only ideas I had.
I agree with Radiant about not using the powerups at all in case they can't be fully manipulated. That means that you might get unlucky and have one spawn on your path though.
I have never seen anything of the sort in the game, so that anomalous disappearance must be an emulator problem. Hope it doesn't start cropping up!
Hmm... I'm really at a loss as to how to use the later examples given in Cheat Engine.
I launched some of the scripts provided with JPC-RR. I discovered that the game runs at 70 fps. ... otherwise I haven't got a clue why this 5-line code I took (as a test) from the site's Lua Scripting section doesn't work. It gives "Kernel: Unprotected error in script: attempt to index a nil value"
I'm guessing this is a message from the emulator itself. Somehow the code isn't compatible with JPC-RR? Yeah I'm pretty hopeless with this stuff. If there were tools (scripts I could easily modify to search and display the right addresses or to get to perform simple algebra) I would take a look into some of the basic mechanisms of the game, but as it stands I don't have the skills or the energy to acquire the skills that it requires. I'm sure that was always the hard part anyway.
local str="test"
while true do
gui.text(0,0, str)
emu.frameadvance()
end
By absolute memory space you mean the memory that the game data itself is loaded into as opposed to where JPC-RR keeps its own stuff?
Yes. And also the bus address space of emulated PC.
This is the address space memory read/write Lua functions use.
So wait... a simple Lua algorithm would find the necessary memory locations because it starts its search directly from the game data? Or is the BIOS data also searched? Meaning autoexec affects it? But how about if I always launch the BIOS the same way wouldn't the memory always be the same as well?
Is there an example of how the algorithm should look? Preferrably as complete as possible :) i.e. one that actually also displays the values it finds in the designated memory locations.
Warepire, I'll have a look at that Dolphin example. If that doesn't work, I'll have a look at the Lua option if it's not too complicated for me.
I once searched some of the memory addresses in absolute memory space:
25C83: X position (Word)
25C81: Y position (Word)
27020: Fuel left (Word)
By absolute memory space you mean the memory that the game data itself is loaded into as opposed to where JPC-RR keeps its own stuff? How did you know/did you just try different options to find out that it was indeed a word and not something else?
Why does the game reserve a whole word for representing those things? It seems to me 2 bytes would be enough.
Depending on the loading, you might need to add or subtract some constant multiple of 16 from those addresses.
How big of a gap can it be? Is it always in the same "neighbourhood" so you could just browse the memory and be able to spot it if you knew whereabouts to look?
I was trying to use JPC-RR in conjunction with Cheat Engine to find some memory locations and store them. However, it seems that the program assigns the game's data to a different place in memory each time so it's not enough to find the correct addresses once. Is it even possible to do what I'm trying to do?
The game runs okay in the emulator, but I'm getting all sorts of crashes from CE and now the emulator froze as well...
Yeah I had the German version for whatever reason.
Yeah enemies, block breaking and traps at least create complications, but the initial TSP solution(s) would serve as a starting point. In the first level it's unlikely to run into the two "joysticks" by accident, but elsewhere that stuff would certainly come into play.
Hmm... I don't know what version the screenshot is from, cause my version has two "joysticks" and no spinners.
I understand such brute-force calculations can become very computationally costly. There's 15 gems in the first level, but really only 13-ish such that have to be considered separately. But if the level had 20+ gems, would it still be within computationally reasonable range? Or is it not as bad as I'm thinking?
The first level (I wanted to include a screenshot I'd taken myself but didn't understand how) seems to be something in the vein of the Travelling Salesman's Problem. I bet there's software out there created for solving the problem. If it was versatile enough, you could maybe "solve" the more open-ended levels using such software.
You start at around where the "02" is in the bottom bar. There's a double fuel tank to your right.
I think you could try different methods to simplify the equation, e.g. the two gems inside the room made of stone would be considered one node. You could also try to sever off connections that seem too distant and impose other restrictions and rules, again depending on the versatility of the software.
I'm not skilled in maths or programming but I'd be interested in hearing whether this would be the general sort of strategy that anyone who's done TAS'ing would use.
I wonder if the random item spawns would turn out to be manipulable. I think probably they'd always be on a timer of some sort, but maybe where they spawn is.
I always thought Keen was a boy as well, but you know how you sometimes call ships and whales "she" without actually having took the time to find out? Oh and Dwedit's post in this thread is only going to encourage others to call all male game protagonists "she" just in case Dwedit's reading their comments.
In all fairness, not all languages have gender, so Finns must be excused.
The run is very clutch.
I'm guessing the save file is created in-game after all. At first I thought there was a pre-existing save file in the "rom" you were using but then I realized it's a pc run so you can type fast.
Nice run, I'm starting to see that PC games are way too well programmed compared to a lot of console games, i.e. seems the glitches aren't as spectacular when you're used to seeing the backwards super jumping of Mario 64. But definitely a yes.
Whoa I just realized what you did in the Perilous Pit :)
You start at around where the "02" is in the bottom bar. There's a double fuel tank to your right.
I think you could try different methods to simplify the equation, e.g. the two gems inside the room made of stone would be considered one node. You could also try to sever off connections that seem too distant and impose other restrictions and rules, again depending on the versatility of the software.
I'm not skilled in maths or programming but I'd be interested in hearing whether this would be the general sort of strategy that anyone who's done TAS'ing would use.