TAS (tool-assisted) demos: part 2

On 5/8/2018 at 11:43 PM, rodster said:

Chillax Map01

  Interesting choice, I played that map quite often in a multiplayer years ago and at some point thought that it can be beaten solo. Actually I don't remember the tactics after the RSK and in the round hall with the spiral stairway afterwards. In a word, no ordinary prb UV max in DSDA and it probably explains that I was wrong.

  In the demo monsters are so helpless in all this craziness. The mancubie gets in the knockout after the single jab, lol

34 minutes ago, Player177 said:

  Interesting choice, I played that map quite often in a multiplayer years ago and at some point thought that it can be beaten solo. Actually I don't remember the tactics after the RSK and in the round hall with the spiral stairway afterwards. In a word, no ordinary prb UV max in DSDA and it probably explains that I was wrong.

  In the demo monsters are so helpless in all this craziness. The mancubie gets in the knockout after the single jab, lol

Thanks! 😄

I wanted to know how a fully built demo would look like on a Chillax map. 😄

There is an old Chillax demo thread with Map 01 beaten on regular play, but it was on Chillax v6.9. Not sure how much changes there are on Map01 between the two versions.

https://www.doomworld.com/forum/topic/56348-chillax-demos-complevel-2/

9 minutes ago, rodster said:

Not sure how much changes there are on Map01 between the two versions.

 

https://www.doomworld.com/forum/topic/56348-chillax-demos-complevel-2/

the 4 viles on the pillars were added making my route not possible, you would need to punch a lot more stuff than I did to have ammo for the viles.

THISSUXX UV speed in 1:58. It has been hoarded since the end of February 2017.

30tsx158.zip

Edit: Andy, please replace the 30tsx159 run with this one as it's included in this zip.

Baby's first attempt at TAS.

UV Speeds for rush.wad

MAP02 in 0:48 ru02x48.zip

MAP03 in 0:54 ru03x54.zip

MAP04 in 0:56 ru04x56.zip

MAP05 in 0:27 ru05x27.zip

MAP07 in 0:19 ru07x19.zip

MAP08 in 1:25 ru08x125.zip

MAP09 in 1:25 ru09x125.zip

MAP10 in 0:57 ru10x57.zip

MAP11 in 0:59 ru11x59.zip

MAP12 in 3:44 ru12x344.zip

Tools used were 50% slow motion and savestates so there are a couple of glides that are a bit slow.

I didn't do 01 or 06 because it seemed like they wouldn't be too different from non-TAS.

Definite room for improvement left on some of them.

Had some more fun with slowmo.

Cybercontrol

D2ALL UV Speed in 27:28

cyberconx2728.zip

Chillax Map 02 UV Max in 10:40 fully built.

Thanks ZM for uploading.

ch02x1040.zip

impressive work ZM! hopefully someone will soon try to beat the 14:02 d2all speedrun including your and a few other awesome tricks discovered in the meantime!

Teeth.wad UV-Max TAS in 1:04.

te32x104.zip

Swim with the Whales

Map 01 UV Max TAS in 0:11

Posting this old demo since a few people posted 13 / 12 demos for this map recently (non tas). This was the beginning of a max tas I started working on after finishing crumpets, but I never got around to it. You can see the start of map 2 max in the demo as well. Maybe one day I will get back to it...

sw01x011.zip

TNT

Map 05 Pacifist TAS in 0:23

~ A L S O   R E A L I T Y ~

ev05px023.zip

1 hour ago, ZeroMaster010 said:

Doom 2 map 30 uv-pacifist in 9:21

 

youtube

last vile spawns at like 8:45, so skip to there.

 

 

 

pa30x921.zip

Very nice!

I think this deserves an own thread :thinking:

The final frontier. Incredible.

The cinematic quality of the final scene is out of this world. A crowd of unearthly creatures rises out of the abyss and performs a mysterious ritual resulting in something not fully comprehensible but so grand... I'm in awe.

Very impressive! How long did this take to do in the end?

1 minute ago, Dragonfly said:

Very impressive! How long did this take to do in the end?

I have no idea honestly, 25 hours sounds reasonable for something like this, but I think it's around 50 hours. Time goes by quickly and some of the viles just refused to spawn... arch-vile number 13 I remember was especially hard to get, had me convinced that there was a 12 vile limit, lol. Then of course I had no idea how many viles were enough, so a lot of time wasted trying to just damage the icon of sin. There were a lot of other smaller issues that I had to figure out, but the solution was "just spawn more viles", if I had stuck to that and just gone from 0 to 30 without messing around it would have been easier.

Incredible level of ingenuity and patience in that Pacifist, ZeroMaster010. Your demos have a kind of emotion only a legendary speedrunner can create.

36 minutes ago, ZeroMaster010 said:

 

I have no idea honestly, 25 hours sounds reasonable for something like this, but I think it's around 50 hours. Time goes by quickly and some of the viles just refused to spawn... arch-vile number 13 I remember was especially hard to get, had me convinced that there was a 12 vile limit, lol. Then of course I had no idea how many viles were enough, so a lot of time wasted trying to just damage the icon of sin. There were a lot of other smaller issues that I had to figure out, but the solution was "just spawn more viles", if I had stuck to that and just gone from 0 to 30 without messing around it would have been easier.

Saw your videos on YouTube before I joined Doomworld and I just want to say that you're a legend man.

Great work & congrats to defeating yeat another legendary "problems of Doom". I wonder at which point we can declare the game "solved", heh.

Oh and for those who wonder about repeating this feat non-TAS... let's see. ZM is manipulating the look-up table, so there's no such thing as probability for him, just a lot of patience and the occasional ssg/cg shot to move the LUT index where he wants it. But the index position is opaque to a human player running around the map, so we can treat every cube spawn as an independent random draw with mutually exclusive results - at least from the moment the monsters start fighting and moving the index on their own. To the calculator-mobile!

The cube spawn is decided by the current index position in Doom's look-up table that approximates true RNG. There are just two indices that spawn a vile, so we're at 2/256 = 0.78%.

ZM says he went for 30 vile spawns, because anything less would be much harder to coordinate for a successful jump that also keeps the attack going long enough to kill Romero. Spawning 30 viles consecutively would get us to (2/256)^30 ≈ 6.0772e-64 (using probability between 0 and 1). For comparison, the winning odds for EuroMillions are 1 in 139,838,160 ≈ 7.1511e-9.

But not even ZM's TAS goes for a perfect vile spawn rate. In fact, there are 22 other monsters alive by the end (if my eyes don't deceive me) and the tally screen shows 10800%. That's 107 killed off cube spawns, discounting the sole original revenant. This gives us considerable slack in repeating his feat: getting exactly 30 viles out of 159 spawns equals to (2/256)^30 * (254/256)^129 * binomial(159, 30), bringing us to nice 4.934e-32. But why stop there? Who says we can't get luckier than ZM in our real world run? After all, we can get up to 159 viles! So we sum the previous equation as (2/256)^i * (254/256)^(159-i) * binomial(159, i) over i = 30 to 159 and improve our chances to... uh, 5.101e-32. Fuck.

But wait... what's that sound I'm hearing throughout the entire demo? Oh, it's the sound of no lost souls around at all. That's right, girls and boys, ZM also manipulated the IoS to not spawn a single pain elemental. Even a single PE would be game over, because it'd spam the map with lost souls, which would make the terrain more chaotic, harder to navigate, and the skulls would probably randomly charge into some of the viles and constantly disrupt their attention. So we gotta detract that from the adverse result in our formula. Luckily, the PE spawn chance is the same pathetic 2/256 (iirc). We get: (2/256)^i * ((254-2)/256)^(159-i) * binomial(159, i) over i = 30 to 159. And with that, we crater to 1.8403e-32.

And that's not all! See anything else peculiar in the demo? Yup, all the viles get spawned on the top platform. Even one of them spawning elsewhere would very likely lead to a quick demise of a non-cheating player, because the vile would unpredictably limit your safe zones and revive killed off monsters. You also can't have any non-vile monster spawn on the top floor, because it'd instantly hurt the dearies and they'd kill themselves in close combat retaliation. What does that mean? There are 13 cube spawn targets on the map, 3 of which are on the top floor. We need at least 30 viles spawn upstairs and everything else downstairs, with no PEs at all. So I think that is: (2*3/256/13)^i * (252*10/256/13)^(159-i) * binomial(159, i) over i = 30 to 159. Here's the wolfram alpha input. The result is 2.8269e-66, which brings us even below the initial pessimistic outlook and it might be less probable that the immediate end of the universe by vacuum decay.

I'm probably wrong. I haven't done this crap since uni. And I don't have the balls to factor in stuff like cacos spawning on the top floor, because they can safely float away before joining the shennanigans. Anyway, good luck, marine. You might need it.

I believe there were only 120 enemies spawned prior to execution of flight into the Romero room, which is bare minimum of enemies required to spawn to gather 30 AVs on topmost platform. Still, the chance of pulling it off in non-TAS run is ridiculously low.

awesome work ZM! once I manipulated the RNG to let the first 3 monsters to be AVs and it was a pain... very astonishing accomplishment!

I wanted the first ones to be AVs such that I can shoot rockets to romero after double arch vile jumps until I figured out that one can do this faster with pain elementals which was too difficult for me to create... maybe it would be a challenge for you? let the first monsters be pain elementals such that you shoot at their ghosts with an angle directly to romero. what crushed me back then was that I wanted to do it with 2 rockets...

all in all this would be way faster than the usual route, if anyone is interested, I still have the demo somewhere around of my attempt (if I remember it correctly I even posted it somewhere here too).

edit:

@rodster found the thread (a demo is in my first comment):

@termrork

I'm interested. Well, I'd probably won't be able to finish anything huge like that but I'd still like to see the effort.

An amazing achievement, realizing a long-standing dream for many Doomers. I believe it was Cameron Prosser first raised such a possibility, but mistakenly thought it could be done with revenant missiles. That at some point led on to the vile blasts, but I think everyone imagined it would remain an undone task for ever, even assuming it might be a theoretical possibility.

And, uh, I can't help but mention another motivation:

18 hours ago, Grazza said:

And, uh, I can't help but mention another motivation:

Not sure whether should we say about sexor motivation in this case without all sorts of clarification 🤣

like artificial act of sexual intercourse in most perverted form between human and the Icon with long foreplay?!

Is there a category that's about no shooting and punching at all? Are there maps in the original IWADs that are literally impossible to finish without shooting or punching?

A Zen category?

1 hour ago, VGA said:

Are there maps in the original IWADs that are literally impossible to finish without shooting or punching?

Doom2@Map18 immediately comes to mind.

Doom 2 map 7 pacifist, NM speed and nomo in 2.49.

lv07_249_tas.zip

8 minutes ago, 38_ViTa_38 said:

Doom 2 map 7 pacifist, NM speed and nomo in 2.49.

lv07_249_tas.zip

Welcome back!