A new trick discovery: zero press!

[j4rio]

it has become a coined term for the trick, no polls needed anymore.

[fx02]

Pfft, just came here to browse a bit after a while and then this shit comes. Well, I hope some new uploads will follow on doom.com.hr :)

Congratz on finding this gem Zero.

[valkiriforce]

I think void switch sounds pretty cool.

[Pirx]

Zero press because you deserve to have your name immortalized this way, and it's outside of doomguy's norm range, so zero press :)

 

What have you done? Found a reason to do all those speedruns again, eh?

[dew]

23 hours ago, rodster said:

What happened to Impse Glide?

It will live on forever in j4rio's heart.

[traversd]

Sorry if I missed it in earlier posts. Is this a vanilla only trick or includes boom/mbf.. other ports?

[Grazza]

On 21/08/2017 at 5:41 AM, rodster said:

What happened to Impse Glide?

That's the name real people are actually using and thus the de facto term for this trick.

[vita]

2 hours ago, traversd said:

Is this a vanilla only trick or includes boom/mbf

The doom2 map 15 impse glide works with both cl9 and cl11.

[RjY]

On 15/08/2017 at 5:26 AM, Grazza said:

I know you're trying to make a point, but I feel the need to mention that almost every factual claim here is incorrect.

Apologies and thank you for the corrections (also to @dew) - I had no idea glides/squeezes went back to 1999. I'm almost certain glides had a huge surge of popularity around the time I mentioned though. I definitely correctly remembered stx-vile thoroughly enjoying himself with glides on TVR!.

 

On 15/08/2017 at 5:26 AM, Grazza said:

I also feel the need to stick up for @cack_handed, as anyone reading this post might think he is some sort of incompetent. He recorded some beautiful demos, and was almost singlehandedly responsible for working out how to make low-speed glides work smoothly. Without his efforts, these glides might have been vaguely the way you describe. To see how he made a work of art of a true glides "assault course", I can recommed his x1t7-036.lmp, which is in the Extremal demo pack.

I think you've invented that interpretation. I meant no such thing. I know cack_handed was very talented and led the the effort to perfect the technique. I was only speaking about my reaction to one particular demo. My intention in naming him as author was to credit/honour him! But it seems it didn't come off that way, so I apologise to him, you, and anyone else who misunderstood.

[Grazza]

I dare say you meant no disrespect and your recollection was honest (if inaccurate), but I am going to stand by what I wrote. Here are the two demos in question:

 

hr02-042
hr02p015

 

 

[Grazza]

Is anyone maintaining a list of maps where the Impse Glide is confirmed to be possible? This might be useful, especially for anyone planning TAS movies, etc.

 

Anyway, on that topic, impse glide confirmed possible for:

Dystopia 3 map06 exit switch (avoiding the need for an archie jump)

 

(There's also Doom.wad e3m6 secret exit switch, but also completely unimportant since it can of course be pressed using a normal switch trick.)

[ZeroMaster010]

Not sure if I've forgotten any, but assuming not, we got these so far:

 

Ultimate Doom E4M2 (facing west, 40 unit gap, through the wall)
Doom 2 Map 07 (facing south, 64 unit gap)
Doom 2 Map 15 (facing west, 48 unit gap)
Doom 2 Map 17 (facing west, 16 unit gap, through the wall)
Doom 2 Map 19 (facing west, 8 unit gap, through the wall)
Doom 2 Map 32 (facing west, 32 unit gap, through the wall)

Plutonia Map 04 (facing south, works upto 69 units)

Alien Vendetta Map 20 (facing west, 40 unit gap, through the wall)

Doom 2 The Way id Did Map 03 (facing west, 16 unit gap, through the wall)
Scythe Map 03 (facing west, 48 unit gap)
 

Facing direction means you walk forwards in that direction and gap is from edge of linedef to edge of doomguy, add 16 units to get distance to center. In all cases the switch is to the left of the player.

 

So far I've seen no cases of pressing the switch while it is to your right (even with the use of -longtics) and no cases of facing north or east. Of course that doesn't mean they don't work, so I should do some further testing to at least determine if they are even possible non-TAS. Not sure if you can hit through wall facing south, or if you can extend as far as 69 units facing west. Also In almost all cases you need to be facing as straight forward as possible, except lv15 (and probably sc03) where you can turn at most twice (TL2), though I am not 100% sure of this.

 

Edit: And according to 4shockblast there's 1 on Rebirth map 06 and they work east and north, I'll see if I can confirm it for myself.

Edit2: I got east to work, unfortunately I can't get it to work with a 32 unit or larger gap which limits its use.

Edited September 13, 2017 by ZeroMaster010

[Aqfaq]

What about doors and lifts? I suppose there are situations where we can activate them through walls to reduce waiting time?

[Grazza]

Aqfaq: Yes, presumably this can be used with anything that can be "used", though as with any research one must be cautious and not assume anything until it has been verified. I was thinking that there might be doors keyed from one side but not from the other that could be opened in this way. Of course, you need the geometry to be such that you can get close enough to the far side. I have also wondered whether some grabs could be done in a similar way (obviously then you don't need to be mashing the use key at the same time); I do recall some grabs working like this, but didn't note the details.

 

ZeroMaster010: That's a lot of useful info. The useless one in Doom e3m6 is east-facing and 32 units

[cybermind]

Is it possible to activate this switch on Doom2 Map11?

[ZeroMaster010]

Well I just managed to activate a switch facing east while being 69 units away, so lv11 should work, but for some reason the wall infront blocks it. If the distance from the wall in front to the linedef switch is 33 units or less it works, at 34+ it fails while it's 40 units for lv11.

 

I also tried pl28, but blockmap is in the way, but facing east and hitting a switch through a wall with a gap of 32 units should work if there are no blockmaps preventing it.

 

edit: I've edited the first post with what I've got.

 

Edited September 14, 2017 by ZeroMaster010

[ZeroMaster010]

blockmap prevents pl28 from being activated when you are facing east... however that does not happen facing west, so ev10 works :) hopefully someone can do a proper run too.

ev10.zip

 

Also I will just edit the first post with all stuff I have, there's a list there that shows when it could potentially be useful. I have also tested pressing switches the are facing the other direction with -longtics, and the opposite seems to happen where the distance is shortened, though I am not 100% sure. However, it's still possible to activate through walls, so maybe 8-16 unit wide gaps could work if you punch something to change your direction when recording demos (somewhat explained in the av20 demo I made).

Edited September 19, 2017 by ZeroMaster010

[rodster]

Does the complevel matter? I mean does this work in all complevels?

[Aqfaq]

Does this trick work on Hell Revealed II map 04 at the switch behind the blue door?

[ZeroMaster010]

I don't think so, because it's the wrong direction. I also did some testing with -longtics to see if they were possible that way and I can't remember that working at all.

[Grazza]

Another one to report as confirmed possible:

Fava Beans e1m1, exit switch.

 

And a question. Any idea if the exit switch of map05 of Mayberry's Perdition's Gate (the first wad with this name, not the later commercial wad) should be pressable? I haven't had any luck with this one, though I don't see any obvious reason why it should be impossible. This would make all 6 maps in this classic wad viable for NM, which would be nice.

[ZeroMaster010]

from my table I have these values:

 

facing north: 56 (16*) units
facing west: 56 (41*) units

 

Fava beans e1m1 (facing north through a wall of thickness 16), which you've already confirmed to work, should be possible since the gap2 (or wall thickness) value is 16*

Mayberry's perdgate map 05 is facing west with a distance of 56 units, but unfortunately also 56* units from the switch to the outer wall.

 

So I'm quite sure it won't work, at least with TAS it hasn't worked yet so without TAS I'd imagine it could be very time consumming if it should work somehow. Of course I haven't checked every possible position with TAS so who knows if there is some exact spot where it could work.

 

There's also blockmap for maps like pl28 which prevent something that should otherwise work, not sure if that is also the case here even if it was close enough.

Edited December 7, 2017 by ZeroMaster010

[Grazza]

Another one confirmed possible: Ultimate Doom 2 map23, reachable via a void glide (there's an easy enough path through the void, though not the most direct one).

 

Proof demo (not for DSDA 'cos it sucks and there's no txt): ul23o055.zip

[Grazza]

Another one: Dmonfear map05 exit switch.

[Dimon12321]

The possibility of performing a zero press on a diagonal switch through the wall is still questionable, but I think it's impossible.

 

No of the following positions work for HacX 1.2 Map 17. The blockmap here isn't an obstacle.

Spoiler

The is an error in position C. The red line was supposed to go through 2 blue squares, not like I drew.

 

[Looper]

On 11/8/2020 at 3:13 PM, Dimon12321 said:

The possibility of performing a zero press on a diagonal switch through the wall is still questionable, but I think it's impossible.

Not sure if this information is any help but:

You can activate the lowest lift in Doom 2 Map30 from the void through the wall. Of course, the thickness of the wall is '0' but still something to be aware of.

 

[A_CyberAttack]

On 8/14/2017 at 8:59 PM, Linguica said:

For the Map17 trick at least, the issue appears to arise out of P_InterceptVector. When the player presses "use", the function correctly detects the solid wall intersecting the "use" trace at 16.04 units from their center (since the player has a radius of 16 units, right in front of their face, in other words). The switch line is detected as being crossed at... 10.2 units, or inside the player, in other words.

 

The reason *why* this happens is pretty confusing and obscure, because the algorithm to detect the line intersection is sort of black magic, and it further messes things up by bitshifting parts of the values to be smaller.

 

I was looking into this and I think I have a pretty good idea why it happens. This is probably already known but just in case someone stumbles upon this topic and is curious.

 

P_InterceptVector is a function to calculate the distance of intersection between two lines/vectors, the first parameter is related to the player position and angle and the second parameter is related to the line i.e. the switch. The parameters are set up in a way that the function returns a value <1 in case it's a hit, i.e. the switch can be used.

 

I don't see anything wrong with the formula the function uses per se, however, as mentioned by Linguica, the function does some bit shifts (>>8), I believe to avoid overflow issues. If you look at those bit shifts from a mathematical point of view as if they were divisions by 256 the result of P_InterceptVector wouldn't change, because they all cancel out. But since those are used over DOOM's fixed point numbers, a bit of the precision of the number is thrown away. Additionally, since for example in DOOM2 MAP07 the switch is hit almost perpendicular, both the "num" and "den" variables that are computed by the function are very small numbers. This causes the division that is done in the function to amplify the numerical error caused by the bit shifts. For example in the demo by Looper (no07-113) he would not have hit the switch if the P_InterceptVector calculated results in a mathematically accurate way.

 

However this is not all, for example in the demo by eLim (lv07no457) even if P_InterceptVector calculated the distance correctly, he would still have hit the switch. So there must be something else. I am not 100% sure about this, but I believe the problem is in P_UseLines, it uses the "finecosine" and "finesine" arrays to calculate the pair of points that are used for the vector of the player's position and angle. I think it's known from squeeze glides that those tables are inaccurate. This also causes more numerical error which also contributes in actually hitting the switch (I think this actually helps more than the error in P_InterceptVector, I believe).

 

It would be really interesting to see if other numerical errors could be useful!

 

Edited June 6, 2021 by A_CyberAttack

[A_CyberAttack]

BTW I can't be the first one to notice this and this probably isn't useful, but I noticed the following code in P_PathTraverse:

 

    if ( ((x1-bmaporgx)&(MAPBLOCKSIZE-1)) == 0)
	x1 += FRACUNIT;	// don't side exactly on a line
    
    if ( ((y1-bmaporgy)&(MAPBLOCKSIZE-1)) == 0)
	y1 += FRACUNIT;	// don't side exactly on a line

    trace.x = x1;
    trace.y = y1;
    trace.dx = x2 - x1;
    trace.dy = y2 - y1;

 

It appears that this code gives you an extra unit (which is absolutely microscopic) of reach when trying to use something, as long as you are completely aligned (to an insane level of precision) with one of the blockmap axes.

 

[A_CyberAttack]

As a test I made the code above unconditional, and while it doesn't have any noticeable effect on uses, it has pretty noticeable effects on movement, for example momentum is always preserved when running against axis-aligned walls and wall-running is very easy.

 

I haven't looked much into it but my guess is since walls are 0 units thick, shifting the x or y coordinate a mere unit makes various checks in the movement code fail to detect the wall.

 

As I said it's probably nothing, as forcing this code to always execute is very artificial, even if it were useful you would need help from the map's architecture and either to find a way to get consistently aligned with the blockmap for various tics or find some place where failing a single check could be useful.

[Looper]

I modified doom2 map07 by extending the exit switch/block by 16 to every direction, so that the center of the player is exactly 64 units away from the exit linedef(s), assuming the player is as close as possible to the exit block. This allowed you to:

1. Hit the exit switch while facing west, in similar fashion as the normal map07 іmpse glide

2. Hit the exit switch while facing north, in similar fashion as the normal map07 zero press

3. Hit the exit switch while facing east, in similar fashion as the normal map07 switch trick

But those angles felt a lot more precise (in position) than the usual map07 phantom press, still a tiny increase to the 'use' distance.

 

I also tried to long switch the exit by using long tics while sliding to the _right_, but couldn't do them, except from the usual angles.

I also tried to activate the exit by using short tics, but placing barrels into the map, forcing the slide to go to the _right_ instead to the left, but no success. (So the difference is that you don't use longtics to get the wrong slide incase shorttics makes some calculations go bad.)

 

It seems like you can fool the game only when sliding to the left, unless there's some strange angles in around 180 degrees _away_ from the exit switch with the wrong slide.

Edited June 7, 2021 by Looper