Miscellaneous demos (part 1) [please post in part 3 instead]

[entryway]

I have found trick on "Phobia - The Age" and has finished the second level in "few" seconds. Sorry for the crappy demo, but there is no sense to make it clearly for (G)Zdoom.

map02 @ Phobia - The Age
Recorded with GZDoom 1.0.04

[entryway]

map02 @ Phobia - The Age
Recorded with GZDoom 1.0.06

[Grazza]

I probably spent a lot longer on this one than I should have done, but in the end I achieved my target time:

The Keep by Kurt Dillard, UV Pacifist in 0:07.

[Opulent]

Go for 0.06! :)

[Grazza]

Not right now, thanks. :p

I forgot to mention that the wad is the same as the one used for this demo, so you can update that page with the author's name if you like. Both demos play back with either version of the wad (the one Jason Henry included had only a music change).

[Schneelocke]

Opulent said:

Go for 0.06! :)

0.06, or 0:06? ;)

[Grazza]

Another short demo on a map by AReyeP. The route is based on, and assisted by, monster cooperation.

"The Base" by AReyeP, UV Pacifist in 0:23.

All the monsters behaved perfectly until near the very end. The lost second due to that is annoying, but given that my next best was 0:27, I decided to call it a day at this.

[Grazza]

I noticed something odd in my aforementioned 0:27 on "The Base", so I have included it in the zip (reuploaded) as an egg demo. Specifically, a baron scratches me but I take no damage (just after I teleport back with the red key). Perhaps it is a similar thing to Anima's GlitchRocket on Extremal E2M6, even though it is not an explosion in this case.

[RjY]

Grazza said:

a baron scratches me but I take no damage

This is what happens:

1. Baron goes into its firing sequence; turns to face target (you, the player)
2. Baron continues going through attack sequence; turns to face target again
3. Lost soul behind baron hits it, doesn't cause it pain or interrupt its firing sequence, but does cause it to change target
4. Baron reaches end of firing sequence; scratches target (the lost soul) but doesn't turn to face target again

That is, the baron hasn't attacked the player at all, it's attacked the lost soul behind it.

The interesting part is that it doesn't turn round at the instant of attack (when it makes the scratch sound). It turns out that barons - and hellknights, they both call the same code pointer, A_BruisAttack - appear to be the only monsters which don't turn to face their target on the instant that they attack. The call to A_FaceTarget is omitted from A_BruisAttack in what I can only assume is an oversight by Doom's authors.

Incidentally, if you put the A_FaceTarget call into A_BruisAttack and play the demo, the baron happily spins round on the spot and smacks the lost soul behind it. The change doesn't affect demo sync, at least in this case.

[Grazza]

Thanks for the analysis. My impression from watching with walkcam/slomo was that the LS attacked just after the baron's scratch, but I hadn't worked out the exact distances, etc.

Interesting: I've certainly seen Barons/Knights firing "out of their back", but never noticed them scratching in that way.

[entryway]

RjY said:

Incidentally, if you put the A_FaceTarget call into A_BruisAttack and play the demo, the baron happily spins round on the spot and smacks the lost soul behind it. The change doesn't affect demo sync, at least in this case.

It will affect sync only if attacked target is a spectre. Only in this case P_Random() is called.

[RjY]

Doom Advent Calendar 2005 stuff I've had sitting around for months:

dac11 in 2:15 (doom2 complevel, prboom 226)
dac12 in 3:40 (boom complevel, prboom 23x, plays in 226/241)
dac17 in 3:39 (doomu complevel, prboom 23x, plays in 226/241)

Note: the last two might count as tool-assisted as I made them with my hacked up version of PrBoom 2.3 that isn't available for download anywhere. I personally don't consider them cheated but it's not up to me to decide!

[myk]

RjY said:
Note: the last two might count as tool-assisted as I made them with my hacked up version of PrBoom 2.3 that isn't available for download anywhere. I personally don't consider them cheated but it's not up to me to decide!

What does it do?

[Anima Zero]

Did several demos for "A Monster Too Many", which can be found here:
http://www.doomworld.com/idgames/index.php?id=14365

Demos recorded for this level:

UV-Speed+Pacifist in 0:32
UV-Speed+Pacifist "Natural Route" run in 0:44
NM-Speed+Pacifist in 0:30
UV-Max in 3:06
UV-fast in 2:48

Sending the demos to Ops now.

[Opulent]

Anima:
http://www.doomworld.com/sda/dwdemo/amtmdemos.zip

[Donce]

Hi, so I sat down and thought, "hey it's a high time to play some doom at last probably". And I did, at last. Played two Zdoom maps and after that came up with this 1994 map, which is quite good for its era methinks. After playing it through, my nightmare detector played an alarm, saying that this should be beatable in NM, and quite easily, in spite of all those monsters. And that turned out to be the case, since relatively quickly I came up with this demo, which I think I wasn't supposed to, keeping in mind this long no-dooming interval, rustiness and stuff.

[Gusta]

Nice :)
Btw. I got a quicky demo for Europa 3 (eaeuro03.wad by Erik Alm) - UV Speed in 0:28, will be sent to Doug.

[Grazza]

Everyone loves bug demos, don't they? Well here's one:

"The Base" by AReyeP, featuring an intercepts overflow. The demo is here.

The overflow occurs at 0:11.62, and causes a crash if the demo is played in Doom.exe (so Prboom-plus is the only current exe that will play this demo to the end). It would be interesting to know exactly what triggers the overflow here, and why it makes Doom.exe crash (or if this is system-dependent).

The zip also includes a demo showing what I was trying to do at the time, but so far I haven't even equalled my pacifist time with this method.

[Opulent]

Gusta said:

Nice :)
Btw. I got a quicky demo for Europa 3 (eaeuro03.wad by Erik Alm) - UV Speed in 0:28, will be sent to Doug.

http://www.doomworld.com/sda/dwdemo/eur3-028.zip

[entryway]

Grazza said:

The overflow occurs at 0:11.62, and causes a crash if the demo is played in Doom.exe (so Prboom-plus is the only current exe that will play this demo to the end). It would be interesting to know exactly what triggers the overflow here, and why it makes Doom.exe crash (or if this is system-dependent).

Because in vanilla it overwrites memory which contains not only simple variables, but also pointers (activeplats for example) It will cause a crash with a high probability. By the way, this overflow practically always leads to noclip effect because it overwrites blockmap width and height variables (see intercepts_overrun_t overrun array in void InterceptsOverrun(...) function)

[Grazza]

Oh, so when people are playing with Doom(2).exe, the normal result is just an unexplained crash and, if one is recording, that no demo is recorded at all.

BTW, my "dream" idea in this is to find a way to trigger the overflow deliberately, and use it on E1M8 (you'd just run straight to the death room once the all-ghosts effect was triggered). It puts a bit of a dampener on things if more often than not the game will just crash instead.

Given that a crash is a standard result, I wonder if it would make sense to extend this limit in Doom-plus. OK, in principle it breaks demo compatibility, but I only know of one single demo where the overflow has been used in order to exit a level. I don't know; how does one weigh avoiding crashes vs a chance of an effect that is useful only in highly exceptional cases?

[Opulent]

I'd say it should be fixed. It is a bug.
If someone wants it, they can play the real game.

let me put it this way, it is annoying to crash DOOM two hours into a demo. :)
but that is just my 2c

[entryway]

AFAIK, it's very difficult to exceed the original limit with correct algorithm. Of course, it can be fixed, but it will not be "removal of a limit"

[Grazza]

Does it get into some sort of repeating loop, so no increased limit, however large, would be adequate?

I'm aware from cph's article that problems with the algorithms are the reason why the 128 limit sometimes proves insufficient. However, I didn't get a clear sense of whether it just leads to many more "intercepts" than are necessary (and so an increase to 1024 could solve most problems), or if it completely blows up.

Opulent: Yes, I dare say that one on the HR ep1 demo must have been especially annoying.

[Opulent]

Grazza said:

Opulent: Yes, I dare say that one on the HR ep1 demo must have been especially annoying.

not to overmake the point, but it has happened to me on Map11 many times. :)

[ultdoomer]

What determines if an intercepts overflow causes the noclip bug to occur or if it causes a crash? I thought it always led to the noclip bug, but now I'm hearing otherwise.

[TimeOfDeath]

sloth3g MAP03, UV Speed in 1:59 (ZDoom 2.0.63a)

[Schneelocke]

TimeOfDeath said:

sloth3g MAP03, UV Speed in 1:59 (ZDoom 2.0.63a)

It'd be great if you could include text files with your demos - thanks. :)

[TimeOfDeath]

Oops, sorry about that. Thanks for the heads up. :)

[Schneelocke]

TimeOfDeath said:

Oops, sorry about that. Thanks for the heads up. :)

No problem. :) It just makes it so much easier to find out who did a demo and which WAD it was for and so on when you find it again on your HD five years later... ;)