Second name: "よるえき"
Fourth name: "ちごにむ"
1st and 3rd names Any name as long as there is no loss
Open the main menu while walking.
Go up and down the stairs 32 times
When you finish the 32nd time, the movie will end at the frame where you opened the main menu.
Special thanks
advice:pirohiko(working on something faster.)
feos: Renamed the branch to what this is traditionally called.
Samsara: File replaced with an 8 frame improvement (converted to BizHawk), courtesy of Spikestuff and DJ Incendration, who have also been added as co-authors, and accepting. Since the improvement here is directly calling the ending rather than fighting Chaos, the branch will remain "game end glitch", but it will still obsolete the published "stairs glitch" run.
Joined: 10/12/2011
Posts: 6437
Location: The land down under.
While the author writes up information:
Link to videoClaiming for Publication.
Edit: Did the thing again.
- Free frame cause Hawk (which we don't talk about).
- Patched up character naming.
Requesting co-author again I guess.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account.Something better for yourself and also others.
While I am a little upset about my submission not being as fast as possible (and the fact that I didn't do the actual TASing which would make me an author), I am happy that there's a faster set of names here. Yes vote.
DJ Incendration
Believe in Michael Girard and every speedrunner and TASer!
While the author writes up information:
Link to videoClaiming for Publication.
Edit: Did the thing again.
- Free frame cause Hawk (which we don't talk about).
- Patched up character naming.
Requesting co-author again I guess.
While I know the framecount of a movie is based on the movie as played back in emulator, I did want to mention something. To console verify a TAS, the inputs of all non-lag frames must be dumped from the movie. This is done using a Lua script that runs while playing back the movie in the emulator.
I dumped the converted movie from spikestuff, and the original from DJ, and they produce identical dumps. So, as far as the real console is concerned, these movies are identical. Looking at the movies, it is clear that the 2 frame difference is just that bizhawk has 2 less frames at the very beginning of the TAS. I'm interested in knowing why FCEUX has these 2 frames and bizhawk doesn't.
I'm interested in knowing why FCEUX has these 2 frames and bizhawk doesn't.
Spikestuff wrote:
- Free frame cause Hawk (which we don't talk about).
That doesn't explain anything. Plus I talked with Spike on discord after my post. He thought it was related to some kind of extra polling done by FCEUX, once on startup, and once again after the reset. However, as far as I can tell from the console, there aren't any extraneous input polls at those points; plus FCEUX reports those frames as lag frames.
While I don't have a FF cartridge, so I can't test the first frame after boot, I have verified other TASes on real carts from boot, from FCEUX. Extra input frames not seen by the emulator would very likely cause desyncs on console. However, I can see from resets that there's no apparent extra inputs.
FCEUX may interpret something as an extra lag frame, but even if that's the case, perhaps that is actually the more accurate emulation. I don't know.
Edit: Transcript from Discord:
Spikestaff: @Bigbass FCEUX takes an extra frame to poll.
Spikestaff: And then since there's a reset, a secondary frame.
Bigbass: But they are lag frames, there's no input polling happening.
Spikestaff: oh that's how I understood it.
Spikestaff: Cause it was always on boot.
Spikestaff: And a reset just did a second frame to it.
Spikestaff: ¯\_(ツ)_/¯
Spikestaff: (And before anyone asks I have it listed as a 3 way on my end; since this is entirely Umi's input which then has the improvements I made on character naming, which then the bus was picked up and moved around by DJ.)
Bigbass: I could probably check the console to see what's happening, but as far as the emulator is concerned at least the first 10 frames are lag frames (I don't remember the exact number). And I don't think the reset that happens a few frames into the movie, is related at all. The two extra frames FCEUX has are before the soft reset, not after.
Spikestaff: Oh
Spikestaff: cause wait need to double check.
Spikestaff: Nope nevermind.
Spikestaff: When I've done a TAS (which never got released) I actually just removed a frame and it'd poll correctly, so the 2nd frame is weird to me.
Bigbass: Ok so I checked the console polling. The soft reset makes this a little more confusing to explain but here's what I've found.
Even though the soft reset occurs during a lag frame (in the emulator) it still needs to be accounted for on console. In this case, the first "input" the replay device needs to perform, is the soft reset. Thus, the reset will occur on the first input poll.
This is a bit weird because from the TAS editor perspective, the first poll should be the 'A' press. Nonetheless, it still works out anyways. After the reset, the console polls again, and now we give it the first 'A' press. Then a blank frame, then another 'A', and so on.
And just to note, other TASes, including for other games, that contain resets, have been verified using the same dump script.
Bigbass: In other words, what I said above is true: The two frames that bizhawk doesn't have, but FCEUX does, are not input polls.
I'm unable to tell you from this data however, which emulator is more accurate regarding these two lag frames. I am able to count lag frames between input polls, but not between power-on/reset and the first poll.
Someone with NES assembly experience could probably go through FCEUX's debugger and check that way.
FCEUX simply takes 2 frames before turning the ppu on. It's hard coded in there. This is well known to be incorrect, but most games don't do anything before the ppu is ready so it doesn't matter in many cases. It really only matters for games that run their RNG in the idle loops between frame code.
And since NES console verification is mostly poll based, and no polls are happening, it's almost invisible.
Console verification. This uses the most recent movie posted on this thread. While I used the FCEUX version to dump with, as I noted previously, the converted bizhawk version is identical once dumped for verification. For the time listed, I don't know which version is the more accurate frame count, so for now I went with FCEUX's 5,800 frames.
Link to video
AmaizumiUni, do you give permission for the submission file to be replaced with this improvement, adding Spikestuff and DJ Incendration as co-authors?
TASvideos Admin and acting Senior Judge 💙 Currently unable to dedicate a lot of time to the site, taking care of family.
Now infrequently posting on BlueskywarmCabin wrote:
You shouldn't need a degree in computer science to get into this hobby.
