I think all complaints will be eventually sorted out when people start taking the "Did you like watching this movie?" question literally, instead of interpreting it to their liking.
TAS refers to "tool-assisted speedrun". "TAS run" means "tool-assisted speedrun run".
But my point is, when a person talks about a TAS on a TAS site and refers to it as a speedrun, why does it have to come down to nitpicking?
Side-scrolling != auto-scrolling.
SMB is side-scrolling, just so you know.
I say this still should be 2-players. They create so much chaos on the screen that's really awesome, who cares if the half an hour long run will be longer by 10-15 seconds because of that?
Ok, two instances of a borderline script used in a hard part of a map among a hundred of more-or-less normal demos, this is so totally like a TAS, and so totally like a run that is based mostly on such scripts. There's an issue of a broad generalization in such logic, don't you agree? There's still a lot to perfect in those runs, especially in 100% ones (they still find minutes of mistakes there).
Just checked the most up-to-date run I have (Qd100Ql2):
E1M1 ends with 14 shells, E1M2 starts with 25 shells;
E1M4 — ends with 35 HP, E1M8 ("Ziggurat Vertigo", comes after E1M4) starts with 50 HP…
I'm pretty sure it was done to give players some leeway in the amount of expendables they have to deal with when running their maps "out of order" (unless it is a Quake feature I don't know about).
All of their scripts emulate real-life conditions, though (unlike inhuman bhops used by Spider-Waffle), so they don't go beyond what's possible in realtime.
Actually, less variety. If one attack does more damage, it will be used predominantly to save time. When many attacks do the same amount of damage, the author can use any of them (more variety!) and not lose time. I think it's good.
And yeah, I'm very excited about this run as well. I see you're being thorough (the number of rerecords exceeding frame count already in the beginning of the run says much) with it and keeping both characters busy with different things, so keep up the good work! I added it to the list of movies I'm looking forward to. :)
Regarding controls…
I use WASD for action buttons and arrow keys for d-pad (which results in some kind of a reversed gamepad). Most of the time I assign them so S is a jump button and D is attack.
Pressing three buttons with at least two of them on the same row poses a problem (until I find myself a good USB keyboard), so, for instance, in Super Metroid I assign attack to numpad 0 to be able to press and hold it with my right pinky.
Frame advance is assigned to right ctrl.
Regarding the prep work…
After all the large-scale preparations (making yourself familiar with the game, route planning, etc.), I would first try to gather all the technical info on character's interaction with the game (a process I'm curently doing for Super Metroid). That includes finding memory values for position and speed (if there's any acceleration), boss HP values and all that kind of stuff. It would save a lot of testing and bruteforcing later on, and also prevent some mistakes (I've learnt it the hard way, heh).
Regarding the process…
Depending on the segment's complexity, I use 3—5 savestates for actual play and up to 4 for referential points from other movies.
F5—F8 — references;
F9 — beginning of a zone/level;
F4 — beginning/ending of a complex segment;
F3 — checkpoint;
F2 — anchor/general duty slot;
F1 — general duty slot.
That's all, I guess.
By funtionality used in TAS production:
I. Engine-based (Doom).
II. Primitive rerecording: savestates + slowdown (Famtasia, old DOS versions of ZSNES, etc).
III. Expansion: bulletproof rerecording + frame advance (FCEU, Snes9x, Gens, VBA, maybe Mupen).
IV. Technological advance: memory watch, ROM disassembly, using bots and macros, easy movie splicing (Mupen lags behind, yet again).
V. ???VI. Profit!
By coverage:
I. Engine-based (Doom).
II. Cartridge-based console emulation with rerecording.
III. Bulletproof optical media-based console emulation with rerecording (first PSX and Sega CD, later other platforms) and arcade machines emulation (Neo Geo, etc.).
IV. Bulletproof computer systems emulation with rerecording (Atari ST+TOS, PC+DOS, Amiga+AmigaOS, etc.).
V. Creating universal enviroments that would allow to TAS virtually any applications?
…Except they actually aren't, unless you mean "they look as if they were tool-assisted". On the other hand, if they manage to find almost a minute of mistakes in QdQwav, it only makes me want to see how much more will a TAS uncover.
Also, don't forget that they have to stitch demos together to avoid inconsistencies, so the amount of health/armor/ammo doesn't transfer exactly between most levels. A TAS won't have this advantage, which will pose as an additional challenge to overcome.
Though I definitely agree (FotF was one of the first games I ever played on PC), I think it'd be much easier to TAS this game on an Amiga emulator, since:
1) we need Amiga rerecording, anyway (LOTS of cool games + high quality music and gfx);
2) it's easier to add rerecording support to an existing Amiga emulator, rather than tune that of DosBox (not implying that it shouldn't be done, though);
3) Amiga has an advantage in picture and sound quality.
Nitsuja had some plans on running the game in this mode, so you might want to contact him for help and/or suggestions, etc.
But I agree that it must be done, GH is a very good game, and it deserves a fascinating run.
Update: Just as I thought there wouldn't be anything "better" than that…
Eye surgery performed incorrectly.
(Not to discourage people who are going to undergo a similar operation, but this vid kinda shows how cautious one should be while choosing the clinic.)
14%, not 17%. It's even. And surprisingly, it makes enough sense for a filename of the smv, if you check.
In the same way, 0% could be called "minimalist" as well, but isn't (even though it's minimalist by definition).
Not counted towards the percent of the items taken.
I've actually been wondering the same. Compare:
SNES "all items" Super Metroid
vs.
GBA "100%" Metroid Fusion,
and
SNES "minimalist" Super Metroid
vs.
GBA "0%" Metroid Fusion.
Consistency, anyone?
Actually, it doesn't change the picture much, since most of the items can still be gotten at different times — it's only a matter of choosing an effective route, which exists for SM100% as well — the other matter is that it's harder to predict the SM routes due to its overall nonlinearity (but the fact that there's only one fastest route for either of the games still stands).