Posts for nitsuja


Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Even just closing the ROM and opening it again causes that memory to leak. I think it's related to the shared memory code (used for Hoe's tool), which I've noticed also causes multiple instances of VBA to not function properly.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
StrangenessDSS wrote:
This reminds me of something you mentioned earlier... that 13 frames was a bit less than half a second. Is the game running at 30fps on the emulator? It ought to be 60fps...
Hmm, no, my mistake, it only runs at 30fps when it's lagging, normally it runs at 60fps. Certain levels such as 2-3 (sink or float) always lag to 30fps, and more of them lag to 30fps whenever there are too many sprites onscreen, such as most of The Day Of. So, 13 frames is a little over .2 seconds unless it's currently lagging like that. And I'm pretty sure it doesn't lag that much on the real system, anyway.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
I found the gameplay a little on the boring side (something about the character...) but the game makes up for it with its good presentation, and it's certainly played well.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
It's too bad some technical issues (including the number of other runs for this game) apparently make it uncertain whether this should be published, because it's really quite entertaining. (Voted yes.)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
pirate_sephiroth wrote:
.m64 files are huge!
It's partly because it has to store 4 bytes of input per frame (from there being so many buttons + analog), and because there is no compression built into the format, and because the emulator was set up to store a frame of input every time the game checks for input. For some reason, Mischief Makers requests input multiple times per frame, so they all get stored in the movie even though the game seems to ignore most of them. The file compresses extremely well, anyway.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
pirate_sephiroth wrote:
I dunno if it's a bug, but I can't make AVIs with these latest releases. The image gets all messy. I though it was some codec problem, but it's not. I thought it was some VBA video setting, but I can make AVIs with the normal VBA (no re-recording). So, might be a bug.
Yeah, I merged some code incorrectly, so it would only work if you set your monitor depth to 24-bit. It should be fixed now if you re-download. (Actually it was fixed a day or two ago but I forgot to announce it.)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Warp wrote:
Yes, we definitely want to hear suggestions on what could be added to the rating system.
It would be interesting to have a "were you familiar with this game?" question, and a way to separately see the ratings given by the two groups. Sorting by the difference between the two ratings would order movies from those requiring in-depth knowledge of the game, to those that are universally equally appealing. Also, I don't know if this was already planned, but it should be possible to sort by individual rating categories rather than by the averaged rating. Technical excellence and entertainment value will probably each give a much different order.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
OK, sorry for taking so many posts to say this, but the trick to get maximum height from Cream's tag action is to activate the tag action as soon after starting a jump as possible (before you even get off of the ground is best), instead of near the peak of the jump. This is because it works by doubling your current upward velocity, and you are moving upward fastest at the instant you start jumping, so activating it then gives the biggest upward boost. That's also why you can use it at the bottom of any upward spring to basically shoot to the top of the level, because it takes the already-high spring velocity and doubles it to something ridiculous. It doesn't matter how fast you're moving horizontally; the only reason that matters on slopes is because it affects your normal jump in that case, which in turn affects the boost. The boost height seems random if you don't always activate the boost at a consistent time in the jump, but this is what I meant by "if you get the timing right". In a non-TAS it's difficult to get the absolute maximum height every time, but it's easy to get very close to it once you know the trick. NOTE: I said it works by doubling your speed, but it also tries adding some small amount to it instead, and it uses whichever one is better. That's why you still get a tiny upward boost if you use it while falling, instead of it making you fall twice as fast.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Chef Stef wrote:
I just checked the readme and it noted that the VBA version was 1.7.2, and the rerecording version was 17.1, so maybe 1.7.2.560 is some hybrid of those?
The v17.1 is the only part of the version number that matters for figuring out which version you're using. 1.7.2 and 17.1 are only coincidentally similar-looking numbers. There is actually a newer version, v19, available here, which is still compatible. Anyway, this played perfectly fine for me as well. It's not bad... I don't think it's publishable quality, but you basically said that already. (So, I guess this is mostly to gauge interest in a run.)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Chef Stef wrote:
I just checked the readme and it noted that the VBA version was 1.7.2, and the rerecording version was 17.1, so maybe 1.7.2.560 is some hybrid of those?
The v17.1 is the only part of the version number that matters for figuring out which version you're using. 1.7.2 and 17.1 are only coincidentally similar-looking numbers. There is actually a newer version, v19, available here, which is still compatible. Anyway, this played perfectly fine for me as well. It's not bad... I don't think it's publishable quality, but you basically said that already. (So, I guess this is mostly to gauge interest in a run.)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
GuanoBowl wrote:
I made a .zip for you all. It includes the m64 and save file.
That defeats the purpose of zipping it, since the save file is huge and barely compresses at all. They should be separate downloads, with only the .m64 file being zipped.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
bobxp wrote:
Not really that good of course, but would I be right in saying that this is the best way to beat the Zone 1 Boss, or have I missed something?
I would guess that this is the fastest way to beat the zone 1 boss. Which means act 2 should be the first act completed, unless it somehow turns out that the Tails/Sonic team isn't at a disadvantage in that act.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
gbagcn wrote:
I tested the vertical height differences between using the Sonic/Cream team and the Sonic/Tails team tag action and this is what I found. The height from the Sonic/Tails team is about 1.8 times the height that the Sonic/Cream team gets. This height difference is so significant that I think that it is worth the small times delays associated with using it.
No, it's definitely not 1.8 times higher. I get that Tails only throws you 1.27 times higher than Cream. On an upward slope, I got 0.81 times (Cream went higher), and on a downward slope I got 1.35 times. (For slopes it depends on how fast you're running and how steep the slope is though, like it does for normal jumps.)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Hmm... now I'm unsure how far back I should go to redo things. (Hex editing works well for some levels, but not for minibosses or when there's any interaction with the enemies.)
StrangenessDSS wrote:
Maybe you'll hate me for saying this, but I'm glad the boost grabbing works so well... it'll leave more fun stuff exclusively in the domain of us speedrunners.
No, I agree on that; it's at least something to differentiate the TAS from a speedrun. (And it is fast, after all, so I really shouldn't complain.)
StrangenessDSS wrote:
There are a couple things I'll have to see if I can duplicate in real time.
Note that it's possible to step through and look at the input if you're not sure what buttons I pressed for any particular part. (BTW, I've noticed that any throw move can be made much faster than normal by double-tapping the B button and holding it down the second time, was this already known?)
StrangenessDSS wrote:
Against Marzen 64, did you try letting a bomb guy get out? I seem to kill him faster with one of those guys helping than just by throwing enemies at him as soon as they appear.
I didn't see one come out until I waited quite a while, and by then I could've already killed him, but maybe it can be manipulated to appear sooner.
StrangenessDSS wrote:
You're only one frame faster than me on the 200M dash. ^_^
Well, boosts can only happen every so often, and the game queues up the next boost if you press the button for it early, so it doesn't take perfect timing to nail every single boost as soon as possible (except for the very first one). Not that I could get close to that time normally anyway...
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
P.JBoy wrote:
vbm part 1 works, but not part 2 of dark knights wip, but they both work on v17
OK, bug fixed, see first post for v19.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
mwl wrote:
AKA and Acryte, those comments should have been said earlier, not when the run is already 25-30% finished. I guess you two can take the time to make runs that do have all those improvements.
Wasn't this intended to be a test run? Better late than never, I would think.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Spoony_Bard wrote:
A few of those 'war' levels in the third area lagged the movie to hell with the explosions in the background. Didn't desync it though.
That's actually the video plugin doing something computationally expensive. Some of the plugins are a lot more efficient at whatever it is than others (they might cache the results so that only the first explosion is slow). It's not actual lag of the kind that will show up in an AVI, which unfortunately happens pretty often in other levels, more often than on the real system. (I try to minimize it, but it's often unavoidable that the framerate will drop to 15 FPS.)
Zurreco wrote:
My only concern so far is whether you could manipulate the Migen Jr fight. Last time I played against him, I would always take a punch in order to make him do a shorter second combo. Any thoughts?
You mean purposely get hit? I tried it but it made his next catchable punch happen so much later that I don't see how it could be worth it. I could've missed something, but I did manipulate luck on every punch in that battle to make the next catchable punch happen as soon as I could make it.
Spoony_Bard wrote:
Wowzers. That boost grab trick works wonders for this game. It even allows you a certain level of 'flight', bypassing many of the game's obstacles and netting you some insane times.
I actually dislike how it makes everything so easy. Before, I was able to bypass most of the same obstacles anyway in trickier ways (to get slightly-slower times). Now there's not as much use for grabbing an enemy to act as a mid-air shield since I can just fly higher to dodge the attack, or grabbing a bomb to use the explosion knockback to get over a wall, since I can just fly over the wall, etc.
Spoony_Bard wrote:
I love how you owned Cerberus Alpha, you made fighting it look easy. I always had trouble with that boss, ALMOST as much as Phoenix Gamma.
It's definitely my favorite boss fight -- really fun to play, and with so many crazy things going on in it. There's not much of a chance to see that here though, since I make such short work of him.
Spoony_Bard wrote:
So seeing how awesome this trial is turning out, how do you think the prospects look for a 100% run, collecting all the Gold gems?
It's probably worth doing, not sure when it'll happen though... BTW, does anyone know if the age affects anything besides the ending? Because I'm starting to suspect that entering 90 has made the game's random number generator effectively biased toward unhelpful results. Especially the "ball" event -- I've never seen such a high ratio of bombs to white balls before, and I kept getting a score of 1 or 2 until finally luck-manipulating it up to a mediocre 6.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Shadow Byrn wrote:
So does this mean that this game going to get done?
Hopefully... I'm more than halfway done already (up to 4-1 now). The WIP is located here. I haven't updated the table of times yet, but they're almost all much faster now. The levels I'm not so sure are optimal are 3-1 and 3-4, and to a lesser extent, 2-11, 3-10, and 3-12. They're good enough, I think, but making them even faster would be nice. Except 3-10, because the festival is so incredibly tedious. The way minibosses take damage in this game is really strange... Each one seems to have a different nearly-impossible-to-replicate circumstance that deals far more damage to it than normal. I wouldn't be surprised if it is possible to OHKO Taurus or even Lunar.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
No, I meant Sonic/Cream. The upward boost is almost as good as Tails' if you time it right, and it can be done more often, and it doesn't slow you down horizontally, and it has no animation time to wait for, and it gives an absolutely enormous boost on springs. Sonic's air-dash move (like the fire shield from Sonic 3) is somewhat useful too. There's the obvious disadvantage of not being able to actually fly, but it's likely that Sonic/Cream would be faster on acts where that's not necessary but where Tails/Sonic wouldn't be able to move upward quickly enough. (Maybe a good choice for Chaos Angel 1?) That said, I hadn't really thought about it but Cream/Sonic seems to be better than Tails/Sonic in every way, making it the team of choice when it's possible to get away with only double-jump + flight instead of a high-jump move. EDIT: Well, not every way - Cream slows down when flying, whereas Tails doesn't slow down at all, and Cream doesn't have any mid-air tricks.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
upthorn wrote:
Okay, Gens+ has some pretty substantial SMS emulation flaws (like completely freezing up on certain games), so I'm not sure if there's any good reason to make it re-recordable at all.
Well it played Game Gear games well enough when I tried it. Also, since it supports Sega CD already, it might be worth seeing if it doesn't have the same desync problems that Gens has. EDIT: or maybe it would be easier to merge in SMS/Game Gear/some of the Sega CD support to the re-recording Gens?
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
If you do keep it static, reserve a lot more bytes than 4. I agree that it's nice when the data always starts at the same offset for the purposes of hex editing, but in that case there's no good reason to be miserly about the header size. For the ROM filename, usually a ROM has its name stored inside it, and that's better to use than the filename. You could store both, of course. For the checksum, you could look at the code that any of the other emulators use to do this, and might as well just reuse that code since calculating a checksum of a Genesis ROM is the same as calculating a checksum of anything else. Or, even just making something up (looping through the ROM data and adding or XORing every x bytes together) would be better than nothing, albeit probably not as reliable, so I wouldn't recommend it if the other option is doable. Also, if GIR is being updated, it should have the entire movie stored in it somewhere, for the bulletproof re-recording feature. EDIT: Oops, I meant, stored in the Gens+ savestate.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
That's assuming it's already uploaded somewhere on the internet that you can link to. If not, you might want to upload the WIP here: http://dehacked.2y.net/microstorage.php, and then post the link it gives.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
OK, thanks, I think I've figured it out now. Basically, the trick is to boost up instead of sideways in that extended slide jump, and grab as soon as possible after that boost, then wait 13 frames (about 0.43 seconds) and repeat, while holding the directional pad up and sideways the entire time. It decreases your horizontal speed slightly each boost if the timing isn't perfect, with more of a decrease the larger the gap is between doing the boost and canceling it. In the test I did, I think I overused it to the point of slowing down a bit, but I still got 6'37'' on 1-1 with it. It's actually easier to perform (probably in real-time, and definitely in a TAS) than the normal extended slide jump, because you don't actually have to time it exactly in that sweet spot part of the jump, it works basically anywhere in it. EDIT: This move pretty much breaks the game... It also has the advantage of not making Marina yell repeatedly, but it's a little boring that it's almost always the fastest thing to do. Still, I don't see how it's possible to get sub-14 seconds on 3-1 with this. Sub-17, yes, if you start flying immediately and use it to fly over the whole level...
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
upthorn wrote:
After that, it's okay to use savestates you made during the movie to resume recording, or to go back and correct mistakes. This can be done in two ways: 1) Loading the savestate while your movie is recording.
I believe that should be "1) Loading the savestate while your movie is playing and not in read-only mode."
AKA wrote:
The speacial stage is irratating as hell
I believe that should be "The special stage is irritating as hell." (sorry)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Are you sure Tornado is definitely faster than never using it? Because I remember that the run using Tornado was losing a startling amount of time at boss fights compared to the one with the Hammer (or the weaponless one, I can't remember which).