Sorry about that, I never bothered replying since there seemed to be a whole different discussion about framerates taking off from there, and since no one else had made a comment about the sound (and I wasn't the only one to notice the sound problem with the original encode), I figured it was evident that it had been fixed.
So the encodes of the first two published Doom movies both have sound problems (of different kinds).
Not to denigrate the hard working people making the encodes, but perhaps it would be better to post the videos in the thread before publication so that others get a chance to review them.
Doom speedrunners don't care about saving bits of a second because intermission screen time is more important for us. Local rules are different and we aren't used to them yet.
Yes, 2 shots seem to be faster because it takes more time to wait for the BFG traces to deal damage than to fire another shot. I'll ask Akse if it's worth fixing.
Yeah, I understand that intermission screen is historically more relevant, because it used to be the most exact you could get for Doom speedruns. When editing movies at a per frame/tic level though, it doesn't seem relevant any more...
According to Doom wikia: "The function M_ClearRandom resets both functions' indexes (P_Random and M_Random) to zero. It is called during initialization of each new level so that demos will be the same each time they are played, and so that multiplayer games are synchronised."
So if the RNG seeds are reset every new level, the final level input from the old TAS could be copy-pasted and still sync. Doesn't seem like too much trouble for someone who knows how.
That information is incorrect. M_ClearRandom is only called at the start of a game.
How many more cells would you need to kill the cyberdemon with only plasma, assuming optimal RNG manipulation? Besides dealing damage faster than the rocket launcher, you'd avoid the weapon-switching animation.
Without doing the calculations, I guess you'd have to collect the extra plasma ammo without losing more than ~4 seconds, which probably isn't possible. The backpack on E2M4 is close but doesn't give enough ammo, and the plasma gun and cell on E2M7 would take too long to collect.
There's also the plasma gun + cell pack near the exit on E2M2... perhaps those could be picked up quickly enough, at least considering that you might save a second by skipping the plasma gun on E2M3 or E2M5? Probably doesn't work out, but I'd like to know that it has been tested.
As for considering this a "new platform" of sorts, I have a question: Is there a way to verify the authenticity of the file? Is it possible to tamper with the file to make it do things that are impossible in actual gameplay? If it is possible, does this mean that we simply have to trust the submitter that the recording has been genuine?
Yes, see http://tasvideos.org/Natt/Doom.html
As long as the movement inputs don't exceed the legal range of values, the demo file simply consists of a sequence of controller inputs, and if doom.exe plays it back with desyncing, then the sequence could in principle have been provided by a player in real time.
As was already noted in the submission, the demo uses "strafe50 on turns", which technically exceeds the legal input range. You can find discussion about this "feature" of the demo format elsewhere (for example, see this Doomworld forums thread). Strafe50 on turns should at most save a few frames compared to an optimal demo that does not use this feature, but using it simplifies TAS recording tremendously.
It might actually be possible to "legally" achieve strafe50 on turns using a custom joystick that connects to Doom's little-known control API. I don't know if this was ever fully investigated. This wouldn't be allowed in human competitive play (ordinary strafe50 without simultaneous turning is allowed, but not custom drivers or controllers that automate strafe50), but might be considered OK for robot players.
Aims for in-game time instead of real time (doesn't care about score tally screens)
Mind if I ask why? It's not like skipping the score tally is detrimental to your in-game time, so as it is it just looks sloppy to me.
By convention, pretty much no one has ever cared about real time in the 20-year history of Doom speedrunning (including 15 years of TASing), and viewers/other players generally appreciate being able to see the individual level times and scores.
Regarding the cyber kill, it's very likely the better option to start with a plasma gun because that does a lot more damage to cybers.
Well, you would fire the same amount of rockets, switch weapons, and finish by firing your cells.
I'm not certain how that would make much of a difference, since he'd still be firing the same number of shots.
Because rockets move slower (slower than the player's movement in fact), you could potentially start firing them earlier and have the cyberdemon walk into them. Plasma projectiles are so fast that you have to run up and get the cyberdemon in your line of sight before firing. That's the theory anyway. It doesn't necessarily save time in practice here.
By the way, if the final rocket does much more damage than necessary to get the kill, you could skip the last 2-3 shots with the plasma gun. Again a very minor thing, and something that Akse already appears to have accounted for, based on his comment.
Interesting with some more Doom TASes!
One thing... Wouldn't it be faster to be closer to the Cyber Demon for the final hit?
Or, could you save a couple of frames by starting with the rocket launcher instead of the plasma gun, firing rockets already from the main building, and making the cyberdemon walk into them?
ALERTS POSSIBLY COMPROMISING MOVIE INTEGRITY:
Skill 3
1 players
What does this mean? The skill level used in the run is Skill 4.
I guess it's the value of the byte encoding the skill level in the LMP format. 0 = ITYTD, 3 = UV, 4 = NM.
The 4:43 timing seems off since the YouTube version is <4 min even with intermission screens.
With Masterjun's latest devilry published (congrats!), I feel that the next step should be more subversive: instead of using arbitrary code execution to manifestly break the game, use it get a subtle advantage (for example reprogram the game to increase Mario's running speed by 10%) and beat the game normally. The goal would be to make it look like a non-glitched run, while actually beating the best possible non-glitched time. If the setup could be disguised well enough, such a run could be submitted as "non-glitched" and potentially fool everybody (a first step on the path towards world anarchy).
Same problem here. I get 20-30 fps watching the demo, 40-50 with rewind disabled, and the sound stutters.
I have a laptop with a 2.2 GHz dual core cpu and 3 GB ram. Not the fastest, but smooth emulation of SNES games shouldn't be too much to ask for.
Very nice. Does this mean that the game can be completed fully with the intended route, or are there any other bugs that you have to glitch around? Either way, I can't wait to see a TAS of this lovely game.
I never really understood why you need to get 15 super missiles for a low% run. Can someone explain?
The only things I can imagine that you need super missiles for are green doors and gates (which only need 1 hit each). Bosses can be killed with the charge beam, so super missiles aren't needed. Seems like you should only need 5 super missiles.
I looked through the tips page for Super Metroid and never found anything, and it's not in the submission text for any of the low% runs that I could see.
I wonder if you could save time by picking up Cutall and entering the Cursed Clock Tower via Garden of Madness.
Doesn't this require the Succubus soul? Though regardless I have no intention of adding out of bounds glitches to the run.
AFAIK you just need to use the Cutall special attack to open gates from behind. This could hardly be called an out of bounds glitch. I haven't tried this myself and could be misremembering how it works though!
I wonder if you could save time by picking up Cutall and entering the Cursed Clock Tower via Garden of Madness. This would let you grab the Wak Wak Tree soul on the way without having to make a detour for it later, and you would only have to climb the Condemned Tower once (perhaps you could even postpone Gergoth until after the Abyss).
Agree about the darkness. It's okay in the small amounts seen here, but would detract from the entertainment value if used more. (It's only used for an extended period of time in the Mine of Judgement, and that's the most boring area of the castle anyway, so not much is lost. )
Nicovideo, seriously? You need to go through a crappy registration process to view videos, and the player doesn't support skipping. Sorry for complaining, but if you're going through the trouble of encoding it, why not make it as accessible as possible?
Registration? I just watched it without an account in this thread.
D'oh! Well, when I clicked on the embedded player, nothing happened for several seconds, so I assumed that it was broken and clicked on the link instead...