Why did you choose the GBA version?
There's already a very similar run on the SNES version:
[1588] SNES The Legend of Zelda: A Link to the Past (USA) "warp glitch" by Tompa in 03:44.65
It has better gfx, doesn't waste time on a three-minute intro and does at least show the full boss room before glitching to the credits. I don't really see a reason to favor this version over the other.
I enjoyed it, but it isn't a completed game - so it really can't be a completed TAS.
Why does it matter? "Finished" is an arbitrary decision the publisher makes on reasons more related to monetary optimization than game quality. It's as relevant to the entertainment of the TAS as the color of the game disc.
If noone had ever told you that it's unfinished and the crash had been replaced with a message saying "You've won the game", would you still have voted no? Even though the gameplay portions of the TAS would have looked exactly the same?
Also, [516] SNES F-Zero "first track" by Saturn in 02:22.03.
No, there's no justification for this; if it weren't for that barrier then you wouldn't need the Grip.
Have you ever tried doing the suitless sequence without using the power grip?
There's no bomb jump and no way to backtrack and acquire the grip. You'd be stuck a couple of screens in, maybe even before entering the ship.
Finding a glitch or a loophole in a commercial game that went through professional QA is a feat worthy of display.
Finding a glitch or a loophole in an amateurish fan game is not surprising at all.
Ok, there's some nontrivial luck manipulation involved, but King's Bounty did it faster, and Monopoly did it better.
No
Adding another category to a game that's been done a million times would require showing something new. The run had a few clever parts (moonwalking, bullet bill, ..), which is good, but large parts of the run seemed too similar to existing runs, just with less coingrabbing and more bumping into walls.
Another issue is that this type of run requires a specific encode with button presses and subtitles to enjoy, and I'm not sure that's a good requirement to have on a published movie. We admittedly have several runs that require subtitles (or reading the submission text) to understand, but extensive OSDs are new.
I don't mean to sound overly negative on an interesting optimization problem that's obviously enjoyed by several players. But since this topic is asking about publication, I'm not sure what to vote.
While I loved the excellent item collection, I will have to vote No because the movement looked improvable. Leaving unoptimized movement frames in the run is a giant waste of the viewer's time and should not be tolerated.
Looking forward to an improved submission in the "100% rings" category!
III) It is possible to download the videos from Youtube using Realplayer.
youtube-dl even allows you to select the quality before downloading.
I started watching, but I guess I'll switch to natt's encode when available. Anything that wasn't up- then downscaled again is bound to look better.
(this is taken from Dada's somebodys WIP encode, thus it has some artefacts. It's at 3:12:56 into the video, at the end of the second gargant ride, before meeting ramuh.)
Lacking impressive action scenes, I've been searching the transcript for some speedrunning-related quotes. Not sure how many are actually visible in the run though.
During the gargant ride:
Zidane: "See? It can go faster. Good job!"
Beginning of disc 4 in black mage village:
Amarant: "...Can lost time be recovered?"
Lindbulm, after dagger turned silent.
Blank: "About time. Yo!"
During the play, when garnet's disappearance is discovered:
Zorn: "We must hurry!"
Thorn: "Hurry, we must!"
Chocobo's Forest:
Mene: "You in a hurry, kupo?"
I doubt newer version of phpBB will solve the issue if it still ties viewing to sessions. We're currently having an issue with MySQL which affects PHP sessions which I doubt has little to actually do with the version of phpBB.
phpBB3 has an option to save the view information in the database.
Unfortunately, phpBB3 is more resource heavy than phpBB2, and with the amount of customization done to this forum, upgrading is anything but trivial.
btw, how can mysql affect php sessions? Aren't you using the "files" session handler?
It wouldn't be acceptable to improve a speedrun of a PC game by deleting the level files beforehand, so I'm not sure why it should be acceptable to improve a FF9 run by effectively making the fmv files unreadable. In other words, I fully support keeping the FMVs in.
Lil_Gecko, do you have any plans for adding subtitles?
That looks very ambitious. Nice!
Two questions though:
* can you elaborate on the origins of the cores? Which of them were developed from scratch, which weren't? And what are the origins of the re-used ones?
* are you planning to provide linux support eventually? Would it be a matter of porting the GUI, or is there still windows-specific code left in some of the cores? Has the new code been developed with portability in mind, or did you introduce new dependencies?
Not sure what stats to upgrade after the hangar yet. Speed is of course very favorable throughout the entire round. Any thoughts on stats-distribution would be very helpful.
I think you'll need some Air for two cash pickups in New York, the one outdoor cash icon in Skatestreet Ventura and for reaching the ziplines in Bullring. Unless higher values make your jumps any faster, just grab as much as needed for the tricks.
I'm not sure about Hang Time. If it just causes you to float in the air like a racoon plumber, then avoid it. Dito Ollie, lower stats might be faster for the quick rail to rail jumps. (grinding was faster than rolling or jumping, right?)
Spin seems to be useless. You don't need to cram more spins into each jump, since scoring is not a problem. Even in the competitions, spreading the cash across two runs seems faster, so you'll just score while you're at it.
Balance? Probably not.
You can buy 9 stat points after the first level. Enough to max speed and spread 5 points between air, ollie and landing. If you're lucky, you won't need to visit the stat screen ever again after that.
The run you linked is 22:32, it had later been improved to 21:57 by the same runner. Since neither of us found any working public mirrors, I'm sending it directly to highness now.
Is it feasible to hexedit the run to sync on the US version prior to publishing? Padding input for the cutscenes would be easy, but I don't know how much luck manipulation is going on. Are there any further version differences that could cause desyncs, like different amounts of lag, slight variations in level geometry or different bugs?
Is there a rigorous definition of "macrostate", or would you just pick any macrostate you like?
For example, with the card deck we might also observe "cards angled slightly to the right when dealt" and "cards angled slightly to the left when dealt". We can use your formulas on that to get an entropy value, but it doesn't tell us anything about red/black any more, but about the dealer's hands. Can both be the entropy of the system, despite being different?
For the gas, you're naming pressure, temperature and volume. Do you observe them separately? Are they a combined macrostate?
Also, how would you determine the entropy of a card deck dealt inside an ideal gas?
Can we even say "The entropy of this system is X", or do we have to add "..with respect to macrostate A" every time? And if so, how do we avoid comparing apples with oranges when dealing with entropy?
(this is taken from 
