I read about people reinstalling their OS regularly, and it just baffles me that that could be necessary just to keep the computer moving at a reasonable speed. I've been running the same install (OSX 10.4) for almost three years now with no noticeable slowdown. What happens to Windows (and, apparently, to Linux) that necessitates all these reinstallations?
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Gah, that'd really hurt my head.
Except that I never actually look at my desktop these days. Here's what it looks like, but this is what I'm normally working with.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
I use XP at home, and macs at school. They run shitslow, and always crash. and as far as I'm concerned, that's all that matters. I can't work on a mac.
macs suck, because of osx
Q.E.D.
Any OS running in a school environment sucks. They get loaded down with crap you don't need in a single-user environment, their maintainers may not be particularly skilled (IT, as a cost center, has difficulty justifying the budget needed to hire skilled workers), and as an added bonus they tend to be out of date. In short, the OS in an academic environment (particularly in highschool, as opposed to in college) is not a valid comparison to the OS in a personal setting.
In any event, of course the OS you choose to use is going to seem better to you than the OS you're forced to use. I have a Windows box I bought basically solely because I need to test Fusillade on Windows, and you know what? It sucks! Nothing's where it should be, the UI is all wrong, the commandline isn't *nix, etc. But none of those are valid complaints because I haven't said anything specific. They're just things being different. You need to learn to separate "different" from "worse".
A valid complaint would be, say, "I don't like how Windows doesn't differentiate between a window and an application. In Windows, if I want to switch to my coding terminals (of which I normally have four open at any time), I have to hit alt-tab a whole bunch of times to bring them all to the front, and if I ever switch away to a different application, I have to hit alt-tab a whole bunch of times again, since they're each different applications. On a Mac, I have to hit command-tab until the Terminal program comes to the fore, and then they're all available (and I can easily switch between them with Command-`). Much fewer keystrokes."
See? Specific, outlines what is done differently and why the alternative is better, at least for me. I'm sure there's legions of people out there who hate the OSX application/window decision. But at least I have a specific reason here.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
I would think that a 100% TAS would have as its top priority reducing revisits to the stages. Given the complicated dependency chains here (most of the upgrades require other upgrades to get), figuring out a route will be nontrivial. But the more time you spend planning the route now, the less time you'll spend re-doing your run just because you realized something halfway through.
A quick look through a GameFAQs guide shows the following. Some of these may not apply in a TAS.
Blizzard Buffalo:
* Heart tank requires a ride armor
* E-tank requires leg upgrade
* Has the leg upgrade (no requirement)
Tunnel Rhino:
* Heart tank requires Triad Thunder and the arm upgrade
* Head upgrade requires Triad Thunder and the arm upgrade
Gravity Beetle:
* Heart tank requires killing Blast Hornet.
* Frog ride armor requires leg upgrade
Blast Hornet:
* Heart tank requires leg upgrade
* Chimera ride armor requires Tornado Fang
Crush Crawfish:
* Heart tank requires a ride armor
* Hawk ride armor requiries Triad Thunder and arm upgrade
Volt Catfish:
* E-tank requires a ride armor
* Armor upgrade requires Gravity Well and arm upgrade
Neon Tiger:
* Arm upgrade requires Tornado Fang and leg upgrade
Toxic Seahorse:
* Kangaroo ride armor requires Frog ride armor
Anything not listed (E-tanks, heart tanks) can be gotten without special upgrades.
In short, you need to get the leg upgrade to get the arm upgrade, and the arm upgrade to get the head and armor upgrades. You need the Frog armor to get the Kangaroo armor, and a bunch of tanks require ride armors as well (Chimera and Hawk will probably be the favorites for dealing with those). IIRC you also must get the Chimera armor to be able to use any armor, even though the Frog and Hawk armors don't have it as a prerequisite.
As a first pass, I'd say the following:
Tunnel Rhino (get Tornado Fang)
Blizzard Buffalo (get leg upgrade)
Blast Hornet (get heart tank, Chimera armor)
Neon Tiger (get arm upgrade)
Gravity Beetle (get Gravity Well, heart tank, Frog armor)
Volt Catfish (get Triad Thunder, E-tank, armor upgrade)
Crush Crawfish (get heart tank, Hawk armor)
Toxic Seahorse (get Kangaroo armor)
Blizzard Buffalo (get heart tank, E-tank, exit)
Tunnel Rhino (get heart tank, head upgrade, exit)
Doppler stage 1 (get golden armor)
This uses just two stage revisits. It's also possible to backtrack in Blizzard Buffalo after getting the leg upgrade to grab the E-tank; I don't remember the stage layout at all well enough to know if this'd be preferable to getting it during the revisit.
I've basically ignored boss weaknesses here on the theory that dealing 4-5 damage per strike vs. 2-3 damage (with the X buster) doesn't make all that much of a difference compared to avoiding a stage revisit. However, it may be possible to keep the number of stage revisits the same while nailing weaknesses better. Tunnel Rhino in particular will be annoying with his lengthy invincibility. Fortunately Neon Tiger is hit before Gravity Beetle.
Again, this is just a rough first-pass route; I haven't put much thought into it at all.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Looks nice! I'm no expert on MM TASing, but the only things I noticed were that you ended with 2 Bombchus which could have been used to speed up a bomb hover (no big deal if you need the chus later), and that while flying with the flowers, you didn't take a straight path.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
It's been a busy few days. Beta 6 is out, with a whole bunch of changes. You can download it here. Please let me know what you think!
Edit: couple screenshot links: One and two.
Pirate Sephiroth: I suppose you didn't read the game description, since it explicitly states that the purpose of the game is to survive for as long as possible. You're welcome to think that it's not a "real game" if you like, but that's one feature I won't be adding. My next game will have plenty of killing. Not this one.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Thanks for the encode, ShinyDoofy!
One of the things I appreciate about this run is that Dezbeast didn't start charging weapons until the frame they were needed -- even if this meant going back two screen transitions. I like that level of precision. Nice work!
Of course, the Koken makes the endgame a bit predictable...oh, well.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
I think the best thing about this run isn't so much the run itself as the WTF nature of the game. You're playing as a knight in plate armor, armed with a pistol, hopping about a castle being chased by skeletons on motorcycles, playing hide and seek with children.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
It's not the fall that hurts. It's the sudden stop at the end. If you can slow that stop down, then the instantaneous force at any given moment (which is what causes the damage) is reduced. The same principle applies to crumple zones in cars.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
I guess an alternative way of putting that is that bullet hells require you to look into the future about a second or two, but to see that future very clearly ("That knot of bullets over there is going to clear up and make an opening, so I need to move there"); traditional shmups require you to look ten-twenty seconds into the future, but see it relatively vaguely ("There'll be a wing of fighters coming in over on this side soon; I should switch to a fast-firing weapon"). These are very different skills.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Thanks for the list! I'll see about giving some of these a shot.
I spent some time working on the game yesterday, and added a bunch of stuff. Then I accidentally made this (2MB animated GIF link), which I'm going to try to incorporate into one of the new patterns.
I had a long rambling paragraph here comparing the different genres of shmups, but it really didn't accomplish anything. Oh, well.
Gradius is a weird example because the fully powered-up ship is vastly more capable than the starting ship, to the extent that when I play Gradius, I tend to go many stages before dying -- and then, having died once, be utterly unable to continue. Part of the reason I like Einhänder is that there isn't such a vast difference in power levels. Right after dying, you're almost as capable against most threats as you are if you'd come all the way from stage 1 (the big difference being the amount of ammo you have, the secondary difference being which gunpods are available, but that rarely makes or breaks a fight). I've always hated that "Oh, you died? Well, now the game's twice as hard!" philosophy that some shmups have.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
I suspect the problem with Phoenix is one of timing -- the wall boss has lengthy periods of invulnerability that would prevent you from landing the maximum number of hits with Phoenix. You'd probably only get one "pass" with Phoenix, and it costs twice as much as Comet Flash does. I admittedly don't know how the damage ratios work out.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Sure, a list of games be helpful -- though short of pirating them via MAME, my ability to find these games is limited.
Re: the lives business, how many bombs does your average shmup start you with, on each life? And bombs are generally available during the level to replenish your supply. In any event, shields and lives are very, very easily tweaked -- they're both parameters for the course.
Anyway, here's my plan of attack:
* Reduce the scoring penalty for getting hit, and improve the bonus for using the inflector. I've tried play with a reduced-strength inflector and I feel like it ends up being too uninteresting -- you can't affect the bullets' paths enough. But I've reduced its maximum range so it shouldn't pull bullets in from all over the place now. The big problem with the inflector is making it only rewarding if it's used skillfully.
* Add multiple bullet sizes and shapes (shape being just for flavor; all bullets will still use circle-based collision detection). Reduce the bullet hit radius (on the default size, it was a 10px radius circle; now it's 8px). Maybe add a laser.
* Replace the less-interesting patterns with better ones. Move some of the Mania and Extra patterns into the main set; I think my skill at making patterns has steadily improved over time and it's a shame that the patterns I've made more recently aren't available to unregistered users.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Thanks for the honest and detailed feedback. I really appreciate it. Now, you've put a lot of thoughts out here, so I apologize if I don't get to them all.
* Play for pure survival vs. playing for score: this is what the inflector is for. You mentioned being bored on the initial loops; that's when you should be using the inflector to pull bullets in and increase your score multiplier. Ideally the game is always as challenging as you're willing to make it, up until the point that the patterns become so hard you can't use the inflector at all. Whether or not this works in practice is a different matter of course.
* The scoring system: I welcome any suggestions you have for improving this, as I admit it hasn't gotten very much attention. Most traditional shooters I've seen either reward you for large hit chains (which incorporate not getting hit, since if you get hit, your chain is broken), or grazing bullets, which is basically done in a more relaxed manner by the inflector. Do note that your score increases purely based on how strongly you affect bullets' paths; whether or not your ship actually contacts the bullets is meaningless. Incidentally, you get a natural score multiplier based on which loop you're on; if I increased the "strength" of that multiplier you'd have more of an incentive to keep playing after getting hit once.
* Bullet hitboxes: I can make these smaller. That's definitely a reasonable suggestion. Do note that they're circles, not boxes, but I suspect that's true for most shooters. :)
* Lives: is this so different from your average shooter? Three lives, two or three bombs per life is pretty standard. The relatively high availability of new "bombs" is unusual in Fusillade, as is the auto-use of bombs (Personally I've always found non-auto bombs annoying), but the actual count isn't really all that far out there.
* Invincibility periods: bear some consideration. The main thing I wanted to do here was keep you in the action even after you get hit. That's why the shield effect just destroys nearby bullets instead of wiping the entire screen. Normally you shouldn't get hit in rapid succession unless you're parked right next to the emitter (though the ChargeLaser pattern breaks this).
* This game vs. various free games: ultimately if you don't think the game's worth the money, then don't buy it. My goal is to create something fun and accessible (even to people who aren't familiar with the bullet hell subgenre), that doesn't require a significant time investment to understand, but does have enough depth to be worth playing more than once. Yes, I'm competing with a large body of free work; the question is if I'm bringing enough new content and ideas to be worth buying. You evidently don't think so; others may think differently.
My scores: unfortunately, I tend to blow away my scorefile regularly while testing. I just replayed the Beginner course, and while I didn't get a particularly great run, I ended at loop 6 (time: 7:14.56) with 233460 points, having died against the Crossfire pattern. On Intermediate, I ended at loop 6 (time: 7:09.96) with 320222 points, dying on Bigspiral. These aren't my best, but they're pretty representative. I tend to overuse the inflector, causing me to get hit a bit more often than I otherwise would -- but giving me more shields to get hit with.
Some of my favorite shmups:
* Einhander (PSX) takes first seat, no question. The componetized bosses are a great concept that's executed magnificiently, the difficulty is good barring a few hiccups, and the soundtrack is rocking.
* Super Aleste (SNES) is reasonably high up there. Here I enjoy the strategic approach to levels, which is similar to R-Type (know which enemies will be showing up when) but much less punishing -- memorization is not a requirement for most of the game.
* Dragon Spirit (NES) is one I've always enjoyed, though it's not particularly noteworthy gameplay-wise. I guess I just have a thing for dragons. Or else it's nostalgia talking. Some of the soundtracks in Dragon Spirit are really excellent, though.
I'm sure there's others that deserve to be on this list, but it's not like I keep a checklist handy. These are just the most memorable, for whatever reason.
To be honest, I haven't played most bullet hells enough to get a good feel for how they stack up against more traditional shmups in my little pantheon. I'm not very good at bullet hell -- playing DoDonPachi costs me a lot of quarters. :) The reason I made Fusillade is because there's a few moments each time I play a bullet hell game where I'm thinking "Yes! This is awesome!" And then I die, and the moment's ruined. I'm trying to capture that moment.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
I saw a 2P TAS WIP of Perfect Dark's first level (sadly, I don't seem to have the file any more), and it was absolutely impossible to follow. It's difficult enough tracking a singleplayer run even if you know the game well, because of all the glitches, but two players just causes information overload. While it's technically more impressive and faster in many cases (since you can parallelize objectives), I think it's more entertaining to publish a TAS where you can track what's going on. It'd also be a lot less work to make!
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
Er, right, sorry, I meant video, not TAS.
Pity it's not feasible. Oh, well. Don't suppose there's any way to wrong-warp up there? I don't know the MM map all that well.
Pyrel - an open-source rewrite of the Angband roguelike game in Python.
