I'd recommend you just TAS for fun initially, e.g. to make funny superplays that have unusual goals or just goals of entertainment, until (or if) you have the knowledge to compete for time. Just enjoy yourself, or aim to find new tricks perhaps.
I wonder if there's a way to intelligently analyze the ROM contents (both the state transitioning of the ROM as well as the memory contents) to pre-determine if a certain sequence will lead to an bad-route (without actually running the emulator)?
Sounds to me eerily similar to the so-called halting problem. If that's so, then no dice.
The halting problem is only unsolvable on turing machines (machines with unlimited storage space).
Also, many goals can not be realized as trivial increments of identical subgoals. In super mario bros if you run right, avoid death and grab the flagpole/axe you'll beat every level...until you get to the first castle that loops you forever if you don't take the correct path. Or 8-4, where you have to take a certain path through the pipes to reach Bowser.
I think most of us have understood the fact that brute forcing is not feasible within our lifetime... But I wondering about smart pruning.
Let's stop thinking about input. Let's switch our usage paradigm - look at the ROM disassembly itself.
I wonder if there's a way to intelligently analyze the ROM contents (both the state transitioning of the ROM as well as the memory contents) to pre-determine if a certain sequence will lead to an bad-route (without actually running the emulator)?
Would it be possible to generate all sets of the "state of the game" with a indicator of if the route will be good, undetermined, or bad? Of course this should be on a game-by-game basis. I know this sounds impossible in our lifespan as well, but I can't wrap my head on if this would be more computationally complex than simple brute force?
This algorithm has already been invented! It's called 'the human brain after learning how to TAS' ;)
Just because a problem is NP (or harder) doesn't mean that the problem is intractible to solve, or even to get close to a good answer. Sudoku is NP-complete - but your computer can solve sudokus in the blink of an eye, because they are so small. On the other hand, TASing almost any game is an incredibly hard complexity class - yet TASes are frequently made, because the TASer merely has to do as well as previous work on the game, not perfectly.
The constructions featured in the paper are pretty amusing :)
After some experimenting I've found out that it's my operative system (Windows 7) which makes the armour refill ships appear randomly at every playback, but the Windows XP on my laptop doesn't have this problem at all, probably because of Hourglass not completely supporting Windows 7. I would've like to try TAS this game, but my laptop is sooo crappy and won't emulate the game perfectly so I can't bother trying.
If anyone else wants to have a go at this, then be my guest and try it out. I'll help out with whatever I can if you wish too.
That's interesting. What function determines when the armour refill ships appear?
I'd have a look through the source if I wasn't about to leave the house
I know you're not completely happy with it, but it's still better than anything I've made in FamiTracker, probably because I haven't found any good tutorials for it.
I didn't read a tutorial before starting :D I still haven't read one, though I'm going to before I write another
I like the song. Are you using it for a game you're making or just making it because you feel like it?
Melody came into my head. If I get good enough I'd be happy to make music for people's games.
Does anyone better than me at getting games to run in Hourglass know how to run this one? Trilby: The Art Of Theft
When I try it and turn logging on, it ends in a loop like the following:
00016648: (f=0, t=6000, c=00000204) DeterministicTimer::GetTicks(0x0) called.
00016648: (f=0, t=6000, c=00000284) MyQueryPerformanceCounter called.
For thousands and thousands of lines.
I think it would be interesting to TAS, since playing it fast requires running past guards and cameras and not giving them enough time to spot you. Plus, there's a system where you trade RPs (essentially EXP) for skills and passive abilities, so some route planning would be required. There could be an any% run (just beat the final boss) and a 100% run (all trilby grades? all RP in the game acquired?)
I also know of two useful glitches:
1) Causing in game processing to pause (e.g. by going through a doorway) on the same frame an alerted guard/camera would cost you an alarm causes them to instead freeze and never actually do it. Perhaps this would be useful in a TAS.
2) If you roll past a screen boundary, you roll much longer. Obviously useful.
Also, an RP maxed file seems to have 8152 RP, unless I missed something. You have to unlock EVERY door from both sides, for example.
Link to video
EDIT: Trilby: The Art of Theft is a game built on this engine: http://www.adventuregamestudio.co.uk/ If it helps
I feel I can say 'yes' to both, because I've done both, but it feels like a phony answer for some reason. Do you know what I mean? It's all work I've done for university, it's gone on for thousands of lines of code, done what it's meant to do, etc but it's not a polished, loved piece of work I can give to other people, just something I can talk about having done. I know what I need to do now, though - program something from start to finish, and blog about it as I do. Have something concrete that I'm proud of, and documented that it's done by me. I have something in mind, too, I've just never had the motivation to do it since there wasn't quite enough of a need for it.
Sorry if I'm clogging up your thread with this, I just have to get this out somehow.
I know java jsp C C++ php javascript and the linux command line, but I have no commercial experience with it (I'm fresh out of university). Or is that okay? I can send my resume if it is.
Have you done any real programming work?
What qualifies as real programming work? The most complex thing I've undertaken is (figuring out how it should work and) making an automated difficulty agent for an online flash rhythm game.
I actually found the first video kind of dull, just because of the difficulty mode making every level dark. You only see a fraction of what's firing at you at any given time.
If you reduce the graphics setting to the lowest settings, the levels are no longer made dark.
I know java jsp C C++ php javascript and the linux command line, but I have no commercial experience with it (I'm fresh out of university). Or is that okay? I can send my resume if it is.
I have no idea what you're talking about. I've played the game through under these conditions before, and you can most definitely lose your rings when hit.
You might be correct. I only know how the debug mode works in the PC version of Sonic 3 and Knuckles, and it made it impossible to lose rings.
In all honesty, I think the best way to reconcile is just to make a sub/splinter site with greater emphasis on entertainment, while tasvideos keeps itself devoted to "gotta go fast". Would that cause huge problems for people?
Hell, you could just have a YouTube channel that hosts entertainment runs. The community could run it and preserve runs that are decided entertaining but not worthy of a new category or obsoleting an existing run.
Basically, I just see a viewpoint that majorly differs from the site's intended goal and could be satisfied with minimal hassle. Let the official runs be as fast as possible, and you can publish entertaining runs for fun, informally.
the ending of Perilous Pit where you decide not to die
The "die and complete level" trick can only be used on levels that don't have oracle, and PP has one.
In general, using glitched level exits on levels with oracle in will make the game uncompleteable.
Also, dying and falling on the oracle won't work.
There is also another known glitched level exit: Enter invalid door (if door is on air, it is most likely invalid) and keen teleports to (0,0) and the level will end.
I didn't mean it like that, just the way the flames don't seem to kill you despite clearly going through you.
There are lots of little things about this that I love. Shooting things because you can, riding the Bouncers, the ending of Perilous Pit where you decide not to die, that one tiny zip, racing upwards through vertical shafts, the many narrowly avoided deaths on Isle of Fire, the expert pogo manipulation. It's a short and sweet run.
Could you do Keen Dreams?
