Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Yes, I forgot to mention that it's possible that the camera speed is limited for the technical reason that loading/displaying level tiles any faster would cause lag. That's not necessarily the case, though.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
I don't think this is a good idea... there aren't very many significant differences from the SNES version, and the screen is smaller, and VBA has some pretty obvious problems with emulating this game (flickering background and slowdown).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Bisqwit wrote:
Now, there are two players, and I'm not sure what to do.
I don't see why the number of players that made the movie should make a difference about what the movie obsoletes. The number of characters, maybe, but how many players there were shouldn't matter.
(I haven't watched it yet but if it's over 2 minutes faster than the published run I'll probably vote yes when I do watch it.)
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Here's a new FCEU version.
- Fixed the reset bug with the MMC1 mapper saving
- Added auto-hold key functionality, mainly to make 2-player runs easier to make (set it in the input config dialog)
- Fixed the splitting of large AVIs (thanks to Andy Olivera for the fix)
- Made frame counter and input display status saved across FCEU sessions
- Increased the version number to 0.98.15 since leaving it the same was generating confusion
windows binaries:
http://www.filespace.org/nitsuja/fceu.zip
mirror: http://www.savefile.com/files/4359024
source code:
http://www.savefile.com/files/7104156
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
I just realized that there is, in fact, a way to set an A+B key in FCEU separately from A and B. Since you can assign multiple keys to a button, all you have to do is assign 2 keys to A and 2 keys to B with one of those keys overlapping, and that key becomes the A+B key.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
For all 3 of those examples you could just check if link's health went down (some addresses of things like health/weapons/position were already posted on the first page of this thread... EDIT: and a link to them is also posted by Hoe in the very next message of this thread). To actually detect when they happen and which one of them happened you'd probably need to monitor the PC register and know what code corresponds with those effects, but depending on your idea that might not be necessary.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
m.r.bob wrote:
This is just a matter of pressing one button though.
It's one button + the frame advance button, which makes 2 buttons at once. That's more than some keyboards can handle, if it's the wrong 2 buttons that is. (You're using the input display to check what was pressed, right?)
m.r.bob wrote:
/me makes topic
Go ahead and make it if you're going to - I wasn't sure if I should post this reply here or wait for a new topic to appear...
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
m.r.bob wrote:
Damn, I've figured out how frame advance works, but I can only get the arrow keys to register while paused, not the jump key or anything.
Try changing what your jump key is. The keyboard has limitations on what keys can be pressed at once, so it'll take some experimentation to make sure all the key combinations you need to be able to press will register.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
OK, I think it's fixed, at least in MMC1 games. It was a problem with the saving, so to test it with that savestate you had, load the state then save it again, then test with that newly saved state.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Highness wrote:
Woulden't it be faster to "camp" under the eal in "Will the eal come out and play" quest, instead of camping over it? Since the star seems to land "below" the eal?
I think if he camped under it, he would have to be moving upward when he tagged the eal's tail, then turn around to go back down to the star, in which case starting above it to be already moving downward would be faster. In any case, I think he said he doesn't plan to redo anything in JRB.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
I like this much better than the glitch-through-everything version, but it looks clumsier than I thought it would... Also, a complaint I have is that you only play one set of levels, instead of playing all of them - I think it's better to play them all for this game because otherwise you miss out on a bunch of unique levels and the run is perhaps too short.
EDIT: This can still be done much faster, even without glitching through walls. For example, here is my first attempt of the first level of this run, it's about 2.4 seconds faster, and I can tell from the inconsistency in the movements I made that it should be even faster than that if fully optimized.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Well its place seems to be
http://tasvideos.org/movies.cgi?id=250
(but if it doesn't belong somewhere on the site then probably no movie of this game does)
Did you watch Adebis's movie? Yours is clearly much slower (well, technically it's a little faster, but only because you didn't get the mushroom, which will cost extra time later) and you did much fewer tricks in the level.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
I have various problems with the sound in FCEU off an on, with no direct causes that I can tell. I think the sound code is pretty unstable/buggy and is likely overrunning some buffers on playback... Sometimes it changes when I change FCEU versions, but it seems they eventually happen in any version. The most common problem for me is that the sound lags behind a lot (mario jumps, and 3-6 seconds later I hear the jump sound), but I haven't heard of anyone else having that problem. When I get cracking and popping sounds, I have the most success getting rid of it by turning the sound buffer really low (after turning it high and restarting FCEU and playing some game for a few seconds). EDIT: I found that actually increasing the buffer size fixed the delay problem, but the amount it needs to be increased to to avoid it isn't always the same. Changing the buffer size shouldn't cause this sort of behavior...
Phil wrote:
That's good another bug that will be fixed soon
Not quite yet, but hopefully soon, yes. The code determining what happens when it reads the memory is pretty complicated, but I can't find anything wrong with the way it's set up, so I haven't found what exactly is making the value be wrong. It might still be some unsaved variable I haven't found/noticed, since the bank-switching doesn't appear to happen until after it screws up, and if you advance to the same frame as the save was on and load it, it works fine, even if it crashes or resets when you load the save from any other frame.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
josh l. wrote:
I can agree less!
I'm just kinda concerned, because this site has become pretty useless now. But if nobody cares, then whatever.
The El Viento run was pretty fun to do, and I'll probably try to do another improvement someday, regardless of whether or not it gets published.
Just because it's often slow to have a run published on it doesn't make it useless... Also, one thing that Bisqwit didn't bring up here, but if none of the judges think they are are familiar enough with the game to judge a movie of it it, they'll have to rely more heavily on the votes to do so, and if there aren't many votes yet either it's understandable that they'll put it on hold for longer.
Your El Viento improvement should probably have been looked at by now since it's a clear improvement by the same author, but you have to admit it didn't get very many votes or otherwise much attention drawn to it, and there are dozens of other submissions...
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
kooz wrote:
the sound skips/studders terribly. I know it is this specific version because I reverted back to the previous one and the issue went away. Did something change in the sound code with v6?
(At what speed is this happening, btw?) I have this problem in older versions too sometimes, it depends a lot on the sound settings, like what you set the sound buffer length to and what the quality/rate/force 8-bit options are set to. But the code did change somewhat, notice the option called "Use old update code" in the Sound Configuration, that puts it back the way it used to be in case the old code worked better for you. (If that doesn't fix it to put the code back the way it was, then it's probably a problem with the settings.)
Phil wrote:
know it happens in Battletoads which is made by Tradewest like Double Dragon but battletoads is AOROM not MMC1. I know it happens in other games but don't remember which ones. Thinking of that it also happens in Legend of Zelda( MMC1), when I was doing my version.
I'm a little closer to tracking down the problem now... A memory read instruction (at 0x8003) is giving a different value than it should when this happens. I think it's somehow related to memory bank switching not being fully loaded, because the memory bank information isn't stored in the save state and it is instead the responsibility of each mapper to restore it properly, and both MMC1 and AOROM have some bank switching (affecting the range after 0x8000) that can happen when the game's running, or at least that's happening when it shouldn't be to make it crash/reset like this.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Baxter wrote:
and nice [rerecord count]/[amount of frames] ;)
Hopefully you aren't seriously taking that into account, though. If I were to start from scratch now using the same paths, it'd probably take less than 20k re-records, but this is higher because it includes all of the route experimentation and refinements and whatnot. And the number doesn't mean that much - there are lots of ways of affecting it regardless of the amount of actual progress being made, even without hex-editing.
Anyway, I'm glad people are enjoying this. I wasn't sure if it'd be too hard to watch or impossible to figure out / focus on what's going on, or something like that.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Bag of Magic Food wrote:
Why isn't your version the latest beta?
The source code for VBA-1.8.0-beta3 hasn't been made available, for one thing.
gbagcn wrote:
If there is a problem with the camera in sonic advance 1 why don't you just make a hack of the rom that makes the camera move faster
Well, I have no idea how to make such a hack (unless the camera position is stored relative to Sonic's position, which I doubt). Is there a known game shark code that does something like this?
Speaking of codes, what do you think about using the (built-in) "Sonic and Tails" code to have Tails tag along? It's be like Sonic 2 (except Tails would be even more worthless than usual).
Bag of Magic Food wrote:
It would be cool if the movie made with the hacked rom still worked with the unhacked one...
Yeah, but it would probably change the gameplay unless it's done in a sophisticated enough way, because normally the enemies/springs/spikes don't exist when they're far enough offscreen.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
This is weird - after I load that state, both the memory and the registers are completely identical no matter whether it's going to reset or play normally afterward. And nothing seems to be wrong with the savestate loading or saving, even when it makes the game crash. I think the problem is some variable that isn't part of the savestate (format) but should be.
Do you know of any games besides Double Dragon that may have had this problem before? I'm wondering if it's specific to this mapper (MMC1).
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
If you load that state you gave, then step 1 frame, then load it again, then unpause, it always resets the game. So maybe it's not the save state that's being transformed to reset the game, but it's actually certain times in the game when you can't load other states without messing things up. Anyway, this proves they're related.
One question: Why is the status info at the bottom of the screen shaking like that after loading the state? Does that normally happen, or is it something weird with the savestate that's making it happen?
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
I found the movie Halamantariel had, and put it here. It doesn't look like anything special since it's just one track that's completely flat, but I think that was it.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
Acryte wrote:
Nitsuja, exactly how long (if you are willing that is) would it take to get a proper start menu for OoT? If it took a while now and I recorded stuff and then switched to a quick one wouldn't the rec files be completely off?
It can already be sort-of fixed by using certain settings in one graphics plugin (the "Copy Framebuffer to RDRAM" setting in Jabo's D3D8 1.6). It has to transfer a lot of data every frame which will probably make emulation on your computer much slower, but it mostly gets rid of the pause menu delay. The resulting movie can only be watched using the same plugin and settings, however. It's a closed-source plugin and I'm not exactly sure what they did to fix it, and I'm a little suspicious of it since the graphics plugin shouldn't even be able to alter the timing like that. It's pretty impossible to predict how long it would take to find a better fix for it (one in the emulator that doesn't rely on that plugin), maybe some day I'll suddenly find something that fixes it, but I wouldn't count on that happening. It seems more likely that Hacktarux or someone that knows how the emulation core works will fix it in the process of improving the emulation accuracy in some future version of Mupen64.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
That means the file is corrupted or something, you could try re-downloading and unzipping it again, but... Check the workbench/submission list, it's there now.
EDIT: Oops, I just realized what you meant by "the latest beta". You can't watch it with that version of VBA. Go Here to get a VBA you can use to watch this type of movie file.
Emulator Coder, Experienced Forum User, Published Author, Skilled player
(1300)
Joined: 12/21/2004
Posts: 2687
OK. I asked for an uncorrupted and corrupted version of the same frame because then it would be easy to compare the files to find out what exactly became corrupted in the save. But this is better, a way to actually reproduce the bug.
