Posts for nitsuja


Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Spacecow wrote:
I'll try to help if I can, but I think marshmallow's run is really the thing to go by here. As far as strategies and routes go I think he pretty much nailed everything. This would change if we chose to do any%, of course, but I personally don't see much point in doing that, since you need a pretty decent amount of jiggies/notes to get to the final boss anyway...
1. Copying SDA runs, but faster, is something we'd like to avoid. Making it any% at least gives some possibility of deviating from marshmallow's run. 2. It could potentially save a lot of time if you skip the right things (the game is long enough as it is that anything to make it faster helps), and the choice of what to skip is somewhat interesting and difficult.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Well, it might as well be added to the non-Win version. Actually, there's nothing platform-specific about the implementation for Windows except that its code would need to be duplicated for each platform (going by the current code structure), but this is implemented fairly differently and possibly makes more sense for this platform.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
OK. Actually I had some questions: I've heard people say that using a 3-button controller setup makes desync less likely. Is this true? I've seen nothing to support that it is, but maybe it matters more for certain games? and about these lines in state.cpp:
		z80_Reset(&M_Z80);
//		main68k_reset();
//		YM2612ResetChip(0);
//		Reset_VDP();
When I comment out the first line, everything still work fine as far as I can tell. Do you know if it was necessary for something? Were the two lines in the middle already commented out, or did you disable them? When I uncomment the last line, the freezes stop happening. (My test was Ecco the Dolphin, loading lots of save states across different sound samples that were playing - each load had about a 5% chance of freezing the emulated CPU... does this happen for anyone else?) But ideally the loading code shouldn't have to reset anything except by loading directly into it, unless... maybe these subsystems can't affect gameplay? I'm not exactly sure what they do, but it seems to be sound stuff.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Bisqwit wrote:
Ps: The vertically scrolling level has some anomaly in scrolling. Probably caused by a bug in FCEU or a broken ROM.
This is a bug in FCEU that has already been fixed. The scrolling was perfectly fine for me when I watched it (in re-recording FCEU 0.98.13).
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Guybrush wrote:
I thought about letting the S-coins be because they are hidden and nobody knows where they are and I don't want to look for them
That's not a good reason at all, it just seems lazy, and you can't call it the "best ending" if you skip them. I think you should try doing a test runthrough of the game just to find them (should be easy because you don't have to do anything perfectly, just test the places they could be - I found 2 by accident earlier but I don't remember where). Then you could start on the actual run once you know where they are. If you're getting all 5 of the S coins and the 2 crystal balls, that's enough that I think it'd be OK to use warps to skip levels that don't have anything worth getting, but warpless would also be fine. Well, it's your choice...
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Bablo wrote:
Well I didn't test normal jump since I found faster way to get on a wall. Anyways I did intro stage again, I don't think it can be faster.
Nice, looks like you managed to save 11 frames on the first level. Do you think this 2nd version won't be finished for a while or will you be canceling this one to replace it?
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
OK, well I don't doubt it happened. I've seen Gens do weird stuff like that before, I think it happened once when I chose Open ROM... and hit cancel. Has anybody here who's used Gens ever gotten the problem where the game freezes (except for the sound processor) when you load too many save states in a row? It's happening really often for me (like every 20 save states I load it freezes and I have to re-load the movie to reset the game), I found a workaround in the loadstate code to make it never freeze but I don't know if it has other side-effects. As for the desync bugs everyone knows Gens has, I'm still getting them sometimes. I found one specific case, where if I load a save state while holding the C button, the game pauses as if I'd hit Start even though I didn't hit the Start button, and when I playback the recording of it, the game doesn't pause there. So maybe pressing a button immediately after loading a savestate is something that can result in a desync because the button can get changed to something else, but what it gets changed to doesn't get recorded or show up in the input display.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
There's no walkthrough, apparently, but there is an actual FAQ, found at the obvious place: http://db.gamefaqs.com/console/nes/file/mighty_bomb_jack.txt (don't click on the link - copy and paste the URL it into your browser instead) It says where the crystal balls are, at least. The S coins are hidden in various places (action stages) just like treasure chest that you need to break a block to reveal. But they don't disappear when you touch them, so I'm not sure if just revealing them is enough or if there's something else you need to do with them.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
OK, well I was asking what part of the size check wasn't working, but I guess you're not sure of that... The first thing that comes to mind is that maybe you have the AVI being saved to some weird pathname that findfile for some reason isn't able to find, or that something is wrong with the code that stores the filename. It works fine for me when I try it... But, anyway, I wasn't aware that plBytesWritten of AVIStreamWrite gives the actual number of bytes written taking compression into account (as opposed to the number of source bytes you told it to write). Everyone I asked said it doesn't, and MSDN doesn't say a thing about it. That's a much preferable method if it works.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Andy Olivera wrote:
In other news, the AVI splitting routine wasn't working
What isn't working about it? You're saying that FCEUI_AviNextSegment works, but the filesize check doesn't work at all? wfd.nFileSizeLow should definitely contain the size of the file (unless it's over 4GB, which it can't be), so I'm not sure what part you're saying was wrong.
Andy Olivera wrote:
On the downside, there's no easy way to compute a file's length in C.
No, actually this is quite easy with fseek and ftell. The real problem is that there's no easy way to compute the length of a file that the OS is denying both read and write access to. The Windows API has a way around that, which is to match the file with a findfile call and ask the size of the file, which works even when opening the file with read access would fail.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
qFox wrote:
(btw, if you're up for suggestions, enable the frameskip key to be held ;))
This is already in there... there's a small delay in case you only wanted to go 1 frame, but then it starts to repeat.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
ventuz wrote:
I like this new change especially TAB turbo, so I can skip boring parts in any games.
Actually you could (and still can) hit F4 and F2 to come close to that before - this is just an alternate way of doing it that I find easier.
ventuz wrote:
Except, I don't know if this is bug or not. When you play a replay and say you wanted to know how long and how many re-recording count and click "Play/Resume movie" button (during replay is still playing). Click cancel, replay stopped, but check mark is still tagged there.
A bigger problem in your case is that Play/Resume movie won't flush the movie to disk until after you hit OK, so the frame count/preview information it shows in the playback dialog of the current movie will be wrong/outdated unless you stop the movie first. EDIT: I read what you said more carefully, and no, that doesn't happen. Clicking cancel does NOT stop the replay, that's why the check mark is still there, because it's still playing. EDIT2: Argh, I found a crash bug with recording a new movie from start... EDIT3: Nevermind, actually recording works fine. It's just a bug that was already there where if you specify an invalid pathname to the new movie it crashes.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Bag of Magic Food wrote:
"Work In Progress". Yeah, I think that term gets overused here.
Especially when someone calls their finished test run a "Completed WIP" or "Finished WIP" or something similar - aren't those oxymorons?
qFox wrote:
My version 0.98.13, yours is the same. Date does differ though (nov 19, dec 4)...
It's only 0.98.13 because that's the 'official' version it's based on. Since the official project is dead, maybe I should've started making up new version numbers after that? About this run - as far as I can tell, it's still up in the air as to which abilities are fastest to use. It's pretty much been decided that wheel is not the way to go because tornado is generally faster than it, but it could be that another ability would save more time than it loses in movement if it's able to kill bosses faster. And BTW, at the very start of the game, walking to the first level door is faster than running to it. (Well, not by more than 1 frame, but still...)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
I made a patch to Gens movie test 9f that fixes some things that've been bothering me about Gens for a while. It's mostly for (my) personal use (so I can stand to watch or make Gens movies) but I thought some other people here might be interested in it. download: Windows Executable (version 9h2) older download: Windows Executable + Source Code Diff (version 9g) changes:
  • bulletproof rerecording (out-of-order loading shouldn't cause desync anymore)
  • I think I caught a desync bug where a chunk of memory at the end of save states was never getting loaded...
  • continuous frame advance (and no pause required, with sound)
  • ctrl-T to toggle read-only (for switching to/from playback while recording)
  • 0-9 keys to load save states 0-9 (not F0-F9 like several other emulators, because the F keys were used for too many things already)
  • shift+0-9 keys to save save states 0-9 (ctrl+0-9 to only select, now)
  • =/- for speedup/slowdown (automatically switches in/out of "slowdown mode" as necessary)
  • hold Tab to fast-forward (ctrl-shift-R to reset, now)
  • it no longer uses 100% CPU when it doesn't need to (this is a big one for me)
  • gave messages an outline (if you didn't set them to transparent) so they're not impossible to read on some bright backgrounds
  • allows you to choose record or playback while a movie is already recording or playing, which properly stops the current movie when you hit OK
  • added autofire and autohold support (configure in Input settings dialog)
  • added toggling graphics layers support (in graphics menu, submenu Layers)
  • added numeric fame counter option (in Misc. dialog)
  • added sound soften filter option (lowpass filter)
  • added high-quality PSG/DAC options (they were already in Gens but were not shown in the menu before)
  • added an option to not use the number keys 0-9 for loading (in Input settings dialog)
  • made movie playback dialog fill in a default filename
  • includes Upthorn's changes to 9h, which are (primarily) multiplayer input display, controller settings loaded from GMV, and somewhat-complete SegaCD savestate support
Except for the first 2 changes, these are windows-only usability improvements. The new keys aren't configurable (unless you're compiling it), so just make sure they don't conflict with other keys you have set (for instance, if you have '-' or Tab set as the pause or frame advance or slowdown mode toggle keys, that's a conflict).
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Well, it should be fixed now (updated in the links above), so try it again.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
It doesn't run at all? What the hell? OK, well there's an easy way to work around it: Copy your fceu98.cfg file from your previous version of FCEU into the directory of v6, and then it should run fine. You'd probably want to do that anyway so you don't have to reconfigure all your settings.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Genisto wrote:
I'm doing a fast as possible version in Jurassic Park II, but there are random desynchros during certain rooms loading. I tried everything, but there is nothing to do.
For a (temporary) workaround, make sure the emulator speed is at 100% or slower and that you don't use fast-forward, especially not during the loading scenes (both for playback and recording). I'm guessing you tried all the other suggestions already (VEHR off, fake mute on, etc.) which should be done in addition to that. I'm not able to get your movie to download at this time, though...
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Bag of Magic Food wrote:
Hmm, does the frame count disappear after a few seconds too? I think it would be good to leave it up in case you start to forget what it was, but let you turn it off without having to advance a frame in case you want to see the screen under it.
I missed this post, but no, the frame count / input display doesn't disappear after a few seconds (it used to, but it was annoying so I changed that in an earlier version).
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Update (v6):
  • fixed 1-frame input/command lag at low speeds (<25%)
  • fixed messages staying too long at low speeds
  • made the sound less irritating at low speeds (there's still an option to put it the way it was before, for anyone that actually preferred it)
  • re-fixed AVI audio/video sync
  • fixed directories code so savestate directory customization works
  • removed input-display-in-AVI
  • removed a few confusing and mostly useless hotkeys
  • hopefully fixed the non-compiling/non-linking code issues that were brought up
Windows Binaries: http://www.savefile.com/files/9280207 Source Code: http://www.savefile.com/files/6606095
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Apparently that means this is the first time anybody has ever tried compiling FCEU with SDL since before Blip's first 0.98.12 version was released, or at least the first time they've complained about it here... Most of these need to be either duplicated from the windows source into the SDL source (such as SetEmulationSpeed) or added as a dummy function to the SDL source (such as HideMenuToggle).
Andy Olivera wrote:
There's still some desync in the AVIs, though much less severe than before. I think it's caused by the values in FCEUI_GetDesiredFPS().
Actually, the AVI code didn't use FCEUI_GetDesiredFPS(), it used exactly 50 or 60 FPS. As it turns out, this is what was causing the audio/video desync all along - I tried changing it to use FCEUI_GetDesiredFPS() instead of 50 or 60 and now it's perfectly in sync.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Bablo wrote:
I should jump all the time, and I lose frame everytime I jump. It's faster to take that delay at start than to jump around 20 times.
But there was a wall coming up that you would have to climb anyway. Since you reach the wall faster by jumping at the start, I don't see how it could be slower, unless you are saying that the jump delay is saved even after climbing the wall. (edit: Hmm, it is saved if you climb to the top, but it gets set to no delay if you jump from the wall.)
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Bablo wrote:
I just tested that if I jump at the very first frame I get control in intro stage and climb on first wall, I have to jump there again, or it has that delay.
What's wrong with jumping again? It should still be faster, if the first jumps are made short enough (and frame-precise).
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Everything is improvable, not just rooms 8 and 11, and there are more interesting goals to aim for in completing this game. This helped him get familiar with the tools a little but it shouldn't have been submitted in its current state.
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
Baxter wrote:
The best would be when the game speed was displayed a certain, short, period of time, independent of the frames, so that it is readable, but doesn't get displayed too long.
It is a period of time, independent of how many frames go by. But, apparently only if you pause it. Angerfist seems to be right that at low speeds when not using pause or frame advance there is an extra 1-frame delay of input relative to the display update, I'll try to fix that...
Emulator Coder, Experienced Forum User, Published Author, Skilled player (1301)
Joined: 12/21/2004
Posts: 2687
OK, there was a typo in the filename, I meant this one: http://dehacked.2y.net/microstorage.php/info/405/