If you are recording a movie, you should not use run cd or run bios
The problem is FFT is not possible to play at all in pcsx-rr without launching via BIOS.
arukAdo wrote:
what was you going for ?
I really just wanted to play around with it. I found the bugs discouraging and haven't done much since my last postings other than work out a skeleton of a crit script.
arukAdo wrote:
movie = 0 entertainment and probably unwatchable, its like playing chess.
Pretty much anything you do with this game is going to be "boring" to people who aren't fans of it. FFT literally is a chess game and its most ardent fans tend to be chess players. Whatever I end up doing, in the unlikely event I complete it, certainly would not be accepted by the site.
arukAdo wrote:
this is a game where you can disable [...] movements on map
False. Map movements cannot be disabled.
arukAdo wrote:
theres even a setting for text speed
Which is irrelevant. Pressing circle or square for "turbo text" advances dialog at the same speed regardless of setting. Any dialog that cannot be accelerated is similarly unaffected by the text speed setting.
Sometimes it gets stuck at the wrong rate. I have luck by briefly hitting Tab (fast forward). It seems to recalibrate upon releasing the button. Sometimes I need to tap it several times. However, my FPS problem is random. I haven't run into a situation where it veers off course consistently.
It looks like running the game via "Run CD" vs. "Run CD Through BIOS" may be the culprit. Can someone confirm? I'm using version is 1001 if it makes a difference.
It seems this is an emu bug after all. But now I can't get Gafgarion to move at all no matter what I do. What the hell. I was playing the game normally just last week. I can't even get it to work with the regular PCSX now. Works fine in pSX, though. Sorry for the false report, guys. I'm stumped, though. How irritating.
You got me. Anything's possible. I can say the bug persists regardless of memory card presence and that it's not down to an exact frame. FFT seeds well in advance, though, without algorithmic knowledge, you can't see the results until showtime. My hunch is that first battle stats are seeded during the FMV "castle raid" intro and that my interrupting it on the first possible frame is an unchecked race condition. All conjecture until I get back to it, though.
EDIT: below was determined to be due to an emulation bug. It can be worked around by launching Final Fantasy Tactics via "Run CD Through BIOS" rather than "Run CD".
Starting the game too quickly triggers a bug which freezes the AI on Gafgarion's starting turn:
EDIT: links removed.
I have not yet investigated the cause, but it seems the game fails to load fully either the AI code or the battle stats required for AI decision making. I plan to pick this up again once issues with Lua are worked out with pcsx-rr. I can say the failure point is prior to Orbonne Monastery.
You probably want to use "VSync Advance" instead, but the downside of that hotkey is that you can't save states between frames
I'm not quite following. I thought every rerecording emulator offered only frame precision in save states. If I'm losing every other frame at times, then actions between frames are diminished, no? Anyway, I'm not too concerned about losing a frame or 20, I'm more concerned about reproducability.
mz wrote:
Your issue with pcsx.pause() is the expected behavior too, I believe, since it probably just waits until the next frame to pause.
Waiting until the next frame to pause I can deal with. pcsx.pause() altering script execution is what I have a problem with. Are its side effects known? It seems to add a wait frame, maybe two at places, but it's hard to tell when PCSX advances two frames at a time.
mz wrote:
You can upload your scripts somewhere, if you want, so I can take a better look at this when I get some time.
How do you guys usually handle this? I know TASvideos doesn't have much in the way of hosted space, so I guess you all are creating your own websites. Who's good these days?
mz wrote:
The next version should work as every other emulator (besides Mupen64 and FBA), so this won't be an issue anymore.
[...]
Anyway, the Lua code in the SVN is already much better than the last official release, but a new version can't be released until zeromus finishes his work with his new SPU.
Maybe waiting for the next version is the way to go. Thanks for your guys' work.
After playing around with scripting Final Fantasy Tactics through Lua for a week or so I'm about ready to say that PCSX-RR's Lua behavior is not fully deterministic. I haven't quite pinned down the behavior, but I believe the crux of the issue to be pcsx.pause() not performing exactly as intended. In the dialog advancement scripts I've created, I can cause them to desync in different spots just by moving pcsx.pause() statements around then continuing when the script pauses. I also want to say that, absent external influences, Lua scripts don't always execute identically, but I can't say this with confidence. I've also noticed that frame advance doesn't always increment by 1; it often advances the counter by two. I haven't noticed a pattern as to what determines the behavior. I'm using v0.1.2 and frame skipping is off. Turning it on doesn't change the behavior, by the way, nor does any fiddling with the dithering, stretching, or sound options.
If anyone else encounters these behaviors, my workarounds have been the following:
* Keep scripts small, switching over during long waiting periods
* Keep the debug pcsx.pause() statements in, since they influence script operation. The downside is heavy babysitting of the Pause key when recording.
* Pad actions with extra frames. This will naturally de-optimize your movie, but what else are ya gonna do?
Since no one has mentioned this, I'm gathering either I'm doing something wrong or PCSX's Lua functionality is not heavily used. Either way I'd love feedback.
Are you sure it's dual core? I thought it was just hyperthreading enabled.
Every piece of literature I've come across specifically mentions a second core. Unless reviewers are conflating dual-core with hyperthreading, then, yes, I am sure. Furthermore, Intel's spec site states 2 cores and 4 threads, with the inference that each core is hyperthread enabled.
I've seen ROM dumps where certain bugs were added, taken away, or both.
The "bugs" were introduced in the localization process. 7th Saga, the Western version, nerfs level ups when compared to its Japanese counterpart, Elnard. But it actually gets worse. Monster stats were not adjusted at all, which is bad enough, but, most importantly, hostile apprentices gain levels as if they are in Elnard. This has the effect of making the game more difficult the more levels you gain. So, in parts of the game there is pressure to keep levels low, but, counterbalancing that are encounter after encounter that play out like boss fights. Running is risky in that a failure can deal enough damage to the party where the enemies have a permanent (and fatal) advantage. The plus side is that apprentices gain levels Elnard style when not in your party, so switching regularly can be beneficial, especially when done after a grindfest, which should always be done solo.
I wouldn't call 7th Saga overly difficult, though, but you do need in depth knowledge to play it properly.
The MKV gradually desyncs at an approximate rate of 0.1 frames per second. By 20 minutes the audio is two seconds ahead of video. Tried with Media Player Classic and KMPlayer with identical results.
Even with dialup, the difference in 5 MB ceases to matter when discussing ~100 MB files. Whether the transfer finishes in 6 hours, 15 minutes or 6 hours, 35 minutes, the recipient has gone to bed hours ago.
As much as I am jealous, I'm obliged to keep the file sizes as small as possible.
Nothing personal. This site attracts obsessive compulsive types with the TAS frame wars and whatnot, so it's not surprising when it spills over into everything else. Anyway, whenever I find someone interested in optimizing PNGs I compare the results to whatever process I'm using at the time to see if I should be switching to something else. Since 2007 or so I've used something called PNGSlim for this purpose. The basic premise is that it pits all the freeware PNG crunchers against each other and returns the victor. I think it has some extra logic on top of basic file size comparisons. It's "open source" in that the core is a win32 cmd file, but I've never taken more than a cursory look. A warning: it's ghastly slow and exponentially so as resolutions increase. Anything less than a Pentium 4 2.5 GHz or so and you'll need to take lunch before this thing finishes.
Flygon wrote:
the 2004 encode possibly looked rather bad to compensate for older Internet connections back then
That and MPEG4 ASF (DivX) still dominated in 2004, which is literally half as efficient as MPEG4 AVC (H.264). ~415 kbps for combined video and audio for a Sega Genesis source seems perfectly reasonable. There are numerous 16 bit publications in the same range.
I don't think this movie was seeded even once beyond the original uploader. I've been trying to download it since a few hours after its publication. I think your options are to track down the movie's creator or encode it a third time.
Hitting a lightning bolt instantly slows the car down to 30 mph.
The whole game they throw people, trashcans, whatever at you, and the one stage where that would kinda make sense? Lightning bolts. The very thing you're supposed to get the car to connect with.
I spent 20 years repressing this game's music. Thanks a lot. The jumping sound is the same as Shadowgate's "scream" noise. I never could line myself up right in the cafe. That's as far as I ever got. Looks like I didn't miss much. That reverse-cafe heart catching game is lame as can be.
Anyone else get the impression that LJN's Back to the Future was a pre-programmed response to skateboard fad games 720° and Skate or Die? They probably had this game ready beforehand and slapped together some sprites and mini-games for the final product. Funny that the glitching actually makes the game's premise more canon. I mean, Marty's warping is the only relativity-breaking thing going on in this game.
I like how they couldn't get the rights to Johnny B Goode and replaced it with a similar sounding--but lame--ripoff. Marty kinda looks like a rocking Hitler in the scene. Achtung Baby!
And why are you avoiding the lightning bolts in the final stage? Isn't that the exact opposite of what they were trying to accomplish in the movie?
I have to agree with the others. While I enjoyed the glitching, the game's far too repetitive, and I needed 150%-300% to make it through. Meh.
I enjoyed watching this far more than I expected, especially the first level. I never expected Pugsley to kick so much ass ass and all with a shit-eating grin on his face. This game looks Nintendo hard. I was impressed with the production values. The backgrounds and music samples are excellent.
