Portal 2

2011-10-18 07:01:50 pm

Quote from 00Svo:

i figured out how to bunnyhop finally.

anyone wanna set some coop records add me on steam. ID: 00Svo

Yeah, I just figured it out too. I went through like four or five cycles of "Ohhhh, that's how you bunny hop" before I actually figured it out. I'm planning on making a youtube video soon that explains how the hell it works. 'edit this post with the link when that's done.

scurty

2011-10-18 07:34:05 pm

Quote from Negatratoron:

Yeah, I just figured it out too. I went through like four or five cycles of "Ohhhh, that's how you bunny hop" before I actually figured it out. I'm planning on making a youtube video soon that explains how the hell it works. 'edit this post with the link when that's done.

I'm curious to know how as well. Is it a script or precise timing?

wfp

2011-10-19 01:37:32 am

Quote from RøvRapunzel:

Is it possible to submit a Solo Speed Run of the coop maps when cheating is involved?

Quote from Paraxade:

I don't think it would be considered a valid category on SDA.

don't let paraxade discourage you if you wanny do it though, rapunzel, you could still post the runs on youtube and i'm sure there'd be interest from people on this forum.

Paraxade

2011-10-19 02:44:49 am

...I never said don't do them, I just said they wouldn't be a category on SDA. << He asked if he could -submit- them.

wfp

2011-10-19 03:13:42 am

no, i understood what you meant. i just meant that not being able to submit them shouldn't hold back anybody from doing crazy runs like solo co-op.

00Svo

2011-10-19 03:35:05 pm

did this in 2 portals. I know there are other ways but this is the easiest original method we could come up with.

also scurty, i could help you out if you are on PC. add me on steam

Blazier

2011-10-24 10:13:19 pm

Here's a solution of 04 Bridge Gels in 8 portals. It's pretty crazy... Might serve as some entertainment if nothing else.

00Svo

2011-10-25 04:18:42 pm

awesome blazier. btw i finally uploaded our record. however rendering and stuff was taking forever so i couldnt get yours in teh bottom corner. This will do though i think. I made it clear we did it together Wink

mattBLiTZ

Edit history:

mattBLiTZ: 2011-10-25 11:04:31 pm

mattBLiTZ: 2011-10-25 11:04:29 pm

2011-10-25 10:50:30 pm

Pac is Back

Does "You are faster than everyone." mean you're actually #1, or just #1 on your friends list? The fucking console version doesn't seem to have anywhere you can actually pull out the list, it just has the little graph plotting your position against the community. My line is the furthest to the left and I got that sound clip, but surely someone has done better than that run I just did o.O

edit: Ah, looks like not. Guess there's really no way to know on console then. Lame.

Great run 00Svo o.O Sick ideas. Once you optimize, that'll be insane.

Negatratoron

2011-10-26 06:40:07 pm

I've figured out some interesting stuff about wall strafing and how to do it better. Basically what happens is, when you do something that makes you go faster, it also makes you accelerate slower. What follows is a longish post about that, that ends with some conclusions on how to get the most out of your wall strafes.

When you wall strafe, it matters what direction you look. If you look straight ahead, parallel to the wall you're strafing against, your maximum speed is about 232. But if you turn a little bit toward the wall that you're strafing against, you move faster. The greatest velocity you can get by wall strafing is about 262, and you do it by turning about 8 degrees into the wall you're strafing against.

When you look into the wall, your maximum wall strafe speed rises by 30 points (compared to looking straight ahead), but you accelerate significantly slower. I recorded myself wall strafing both ways (used "cl_showpos 1" to display velocity and to verify my strafing angle) at 30 fps, and went through each video frame by frame to get some actual data:

So one way to optimize wall strafing is to begin by looking parallel to the wall, and then turn 8 degrees into the wall after about a third of a second when your velocity nears 232. I recorded myself trying this, went through that video frame by frame, and added it to the plot:

(The yellow ends up lower than the red because I didn't turn quite the right amount into the wall when I did this.)

Ironically, you accelerate fastest when you don't wall strafe at all, but just move forward. I recorded myself moving forward until my velocity was about 175, then wall strafing parallel to the wall until it was about 232, and then tilting 8 degrees in toward the wall, and added that to the plot:

The red line takes about a second to reach a velocity of 261. So if you're strafing for over a second, following the green line and following the red line will get you the same amount of speed; the green line just gets you to 261 faster. Based on a rough calculation, if you wall strafe for at least a second from a complete stop, this will save you about 0.12 seconds. If you have some velocity coming into the wall strafe, obviously the green line will save you less than 0.12 seconds.

If you only have a small amount of wall to strafe against and need all the speed you can get (for example, search on youtube for "sp_a4_laser_catapult 0 portals" and look at the jump at 0:18), this can make things significantly easier.

So to milk wall strafing for all it's worth, the process is
1 (optional): Walk normally along the wall until your velocity is 175 (about 0.1 seconds starting from 0 velocity)
2: Wall strafe looking straight ahead until your velocity nears 232 (about 0.1 seconds starting from 175 or 0.3 starting from 0)
3: Turn 8 degrees in toward the wall to increase your velocity ceiling to about 262

I've marked 1 as optional just because I feel it's disproportionately difficult to do for the benefit that it gives. 3 is also optional on short walls where either your velocity doesn't make it to 232, or your velocity gets to 232 but you don't have enough wall left for it to be worth messing around turning and stuff. On the other hand, turning into the wall you're strafing against sets you up nicely to start bunny hopping.

I've figured out some interesting stuff about wall strafing and how to do it better.  Basically what happens is, when you do something that makes you go faster, it also makes you accelerate slower.  What follows is a longish post about that, that ends with some conclusions on how to get the most out of your wall strafes.

When you wall strafe, it matters what direction you look.  If you look straight ahead, parallel to the wall you're strafing against, your maximum speed is about 232.  But if you turn a little bit toward the wall that you're strafing against, you move faster.  The greatest velocity you can get by wall strafing is about 262, and you do it by turning about 8 degrees into the wall you're strafing against.

When you look into the wall, your maximum wall strafe speed rises by 30 points (compared to looking straight ahead), but you accelerate significantly slower.  I recorded myself wall strafing both ways (used "cl_showpos 1" to display velocity and to verify my strafing angle) at 30 fps, and went through each video frame by frame to get some actual data:

[img]http://img9.imageshack.us/img9/3817/wallstrafe1.png[/img]

So one way to optimize wall strafing is to begin by looking parallel to the wall, and then turn 8 degrees into the wall after about a third of a second when your velocity nears 232.  I recorded myself trying this, went through that video frame by frame, and added it to the plot:

[img]http://img248.imageshack.us/img248/9594/wallstrafe2.png[/img]
(The yellow ends up lower than the red because I didn't turn quite the right amount into the wall when I did this.)

Ironically, you accelerate fastest when you don't wall strafe at all, but just move forward.  I recorded myself moving forward until my velocity was about 175, then wall strafing parallel to the wall until it was about 232, and then tilting 8 degrees in toward the wall, and added that to the plot:

[img]http://img97.imageshack.us/img97/6036/wallstrafe3.png[/img]

The red line takes about a second to reach a velocity of 261.  So if you're strafing for over a second, following the green line and following the red line will get you the same amount of speed; the green line just gets you to 261 faster.  Based on a rough calculation, if you wall strafe for at least a second from a complete stop, this will save you about 0.12 seconds.  If you have some velocity coming into the wall strafe, obviously the green line will save you less than 0.12 seconds.

If you only have a small amount of wall to strafe against and need all the speed you can get (for example, search on youtube for "sp_a4_laser_catapult 0 portals" and look at the jump at 0:18), this can make things significantly easier.

So to milk wall strafing for all it's worth, the process is
1 (optional): Walk normally along the wall until your velocity is 175 (about 0.1 seconds starting from 0 velocity)
2: Wall strafe looking straight ahead until your velocity nears 232 (about 0.1 seconds starting from 175 or 0.3 starting from 0)
3: Turn 8 degrees in toward the wall to increase your velocity ceiling to about 262

I've marked 1 as optional just because I feel it's disproportionately difficult to do for the benefit that it gives.  3 is also optional on short walls where either your velocity doesn't make it to 232, or your velocity gets to 232 but you don't have enough wall left for it to be worth messing around turning and stuff.  On the other hand, turning into the wall you're strafing against sets you up nicely to start bunny hopping.

00Svo

2011-10-26 07:47:41 pm

wow. just wow. awesome work there. i read it all. will be implementung into my playing. just wondering, why is part one so difficult? i imagine it would hardly be different from the normal way. alsoalso you hould check out wall hopping and add it to the comparison

DemonStrate

2011-10-27 06:12:30 am

If we were you, we would quit now.

Negatratoron, thank you so much! I've thought about doing stuff like this before. I wanted to do a graph of speeds for ABHing back in my Portal running days, but I never got around to it. Maybe I'll do that now when I start my new run... it would be useful information. Also, do you think it would be worth it to do the same kind of investigation for bunny hopping? There have been many claims about how turning/weaving increases your speed, but I'm not entirely convinced. Also, do you have an Excel spreadsheet of that data (I'm assuming it is how you populated the graph)? Could I possibly have that (I'm a pack rat for data)?

Negatratoron

2011-10-27 10:42:08 am

Quote from 00Svo:

wow. just wow. awesome work there. i read it all. will be implementung into my playing. just wondering, why is part one so difficult? i imagine it would hardly be different from the normal way. alsoalso you hould check out wall hopping and add it to the comparison

Part 1 isn't that difficult I guess. On a long wall you might as well do it because it might get you 0.05 seconds, and why the hell not... If you're on a short wall and include part 1, you'll indeed be going faster at the end of it, but then you need to mash one more button to make that happen, so depending on what you're doing it might make things harder. I just wanted to make it clear that this is a set of ideas to use however you want, not a step-by-step process that must be followed.

Quote from DemonStrate:

Also, do you think it would be worth it to do the same kind of investigation for bunny hopping? There have been many claims about how turning/weaving increases your speed, but I'm not entirely convinced. Also, do you have an Excel spreadsheet of that data (I'm assuming it is how you populated the graph)? Could I possibly have that (I'm a pack rat for data)?

Yeah, I'm analyzing bunny hopping too. I'll post the results on here when I'm finished, which will probably be some time next week.
Here is the spreadsheet: http://www.mediafire.com/?cu00ghdhgzrsfmq
The first four columns are the data on the graphs. The second four columns are sums of the velocities in first four, so the second four are proportional to the total distance traveled using each method.

[quote="00Svo"]wow. just wow. awesome work there. i read it all. will be implementung into my playing. just wondering, why is part one so difficult? i imagine it would hardly be different from the normal way. alsoalso you hould check out wall hopping and add it to the comparison[/quote]
Part 1 isn't that difficult I guess.  On a long wall you might as well do it because it might get you 0.05 seconds, and why the hell not...  If you're on a short wall and include part 1, you'll indeed be going faster at the end of it, but then you need to mash one more button to make that happen, so depending on what you're doing it might make things harder.  I just wanted to make it clear that this is a set of ideas to use however you want, not a step-by-step process that must be followed.

[quote="DemonStrate"]Also, do you think it would be worth it to do the same kind of investigation for bunny hopping? There have been many claims about how turning/weaving increases your speed, but I'm not entirely convinced. Also, do you have an Excel spreadsheet of that data (I'm assuming it is how you populated the graph)? Could I possibly have that (I'm a pack rat for data)?[/quote]
Yeah, I'm analyzing bunny hopping too.  I'll post the results on here when I'm finished, which will probably be some time next week.
Here is the spreadsheet: http://www.mediafire.com/?cu00ghdhgzrsfmq
The first four columns are the data on the graphs.  The second four columns are sums of the velocities in first four, so the second four are proportional to the total distance traveled using each method.

toten

2011-11-23 11:31:38 am

sp_a2_ricochet:
It can be done faster with better timing with the moving wall.

toten

2011-12-06 09:25:18 am

sp_a2_column_blocker:

Negatratoron

Edit history:

Negatratoron: 2011-12-29 10:42:50 am

2011-12-28 01:50:15 pm

Here's a post on bunny hopping:

Edit: If anything in here doesn't make sense, please tell me so that I can edit the post and improve it.

First of all, it looks like Portal 2's physics engine only knows whether you're on the ground or off the ground; horizontal velocity is computed the same way whether you're jumping, falling, or crouch flying glitched. So everything in this post applies both to both bunny hopping and to crouch flying glitching.

There's some math in this post, so here are some definitions I'm going to use:

v for velocity = your horizontal velocity. It is a vector with an x component and a y component. As far as I can tell, horizontal motion and vertical motion do not affect each other at all in Portal 2 so v does not have a z component.

h for heading = direction you're accelerating in, as a unit vector with an x component and a y component. It doesn't matter if you accelerate by hitting forward, backward, left, or right. (You could bunny hop just as well by turning 90 degrees and pressing forward and backward instead of left and right)

a for angle = direction you're accelerating in, as an angle in degrees or radians (take your pick). That is, h = <cos a, sin a>

So here's how Portal 2 figures out your velocity when you're bunny hopping or crouch flying glitching:

v can be broken down into two components: one parallel to h and one perpendicular to h. Here are two examples:

and

Portal 2 calculates the component of your velocity in the direction of h, which is the red line. If it is less than 60, you accelerate in the direction of h until your velocity in the direction of h is 60.

For the math geeks among you, this means that your velocity can be approximated with the differential equation

whenever you're actually accelerating. If 60 - h v is less than 0 then the equation does not hold; you never slow down in the direction you're trying to go.

I don't want to clutter up this post too much, so I'm throwing a graphical example of how bunny hopping increases your velocity into a hidden content box. Click the button if you want to look at it. Otherwise, below are some implications about how to bunny hop in Portal 2.

You're moving with velocity old v and you strafe in direction h like so:

Portal 2 calculates the components of your velocity with respect to h:

Portal 2 looks at the component of your velocity in the direction you're trying to move and determines whether it's less than 60:

If it is indeed less than 60, you gain velocity in the direction you're trying to move until you're going 60 in that direction:

Your velocity is now new v

Here are the implications about bunny hopping:

Implication 1: The amount of velocity you gain when you bunny hop is directly related to how much you turn and weave. It doesn't matter if you turn in a big circle or if you go left-right-left-right, but you're only getting faster if you're also changing direction.

Implication 1 disclaimer: Portal 2 only does its physics calculations a few times a second. If you turn so fast that v and h make an angle of greater than 90 degrees, you'll just slow down. That is, you slow down if the dot product of h and v is negative.

Implications 2, 3, and 4: (2) If the direction of h is too close to the direction of v, you won't accelerate at all (specifically, you only accelerate if the dot product of h and v is less than 60). (3) If you know how fast you're going, you can calculate the size of this dead zone. For example, your velocity is 175 when you're just walking normally. Suppose you want to jump, turn d degrees to the left and then strafe right to start bunny hopping. Your speed in the direction of your strafe is given by 175 * sin d. If this is greater than 60, you won't accelerate at all. So if you set 175 * sin d = 60 and solve for d, you get d = arcsin(60 / 175) = 20.05 degrees. (4) This means that if you walk forward and start bunny hopping, you should turn a maximum of 20.05 degrees before your first strafe; turning more won't help at all.

Here's a picture of implications 1-4:

Implication 5: You gain exactly 0 speed when h is at the top of the green arc in the picture of implications 1-4. You gain the most speed when h is at the bottom of the green arc. You always gain the most speed if h is perpendicular to v, but then you also end up turning the most.

Picture of implication 5 (Of course, 60 * h should be 60 times as long as h. None of these pictures are really to scale):

There's still much more research to do here. I've only found a way to mathematically describe bunny hopping. I'll work on ways to use this stuff to actually optimize bunny hopping at my leisure, and post anything interesting I find. Feel free to bang on my Steam account (id: Negatratoron) if you have questions, comments, suggestions, or just want to talk about general abstract nonsense.

Here's a post on bunny hopping:

Edit: If anything in here doesn't make sense, please tell me so that I can edit the post and improve it.

First of all, it looks like Portal 2's physics engine only knows whether you're on the ground or off the ground; horizontal velocity is computed the same way whether you're jumping, falling, or crouch flying glitched.  So everything in this post applies both to both bunny hopping and to crouch flying glitching.

There's some math in this post, so here are some definitions I'm going to use:

[b]v[/b] for velocity = your horizontal velocity.  It is a vector with an x component and a y component.  As far as I can tell, horizontal motion and vertical motion do not affect each other at all in Portal 2 so [b]v[/b] does not have a z component.

[b]h[/b] for heading = direction you're accelerating in, as a unit vector with an x component and a y component.  It doesn't matter if you accelerate by hitting forward, backward, left, or right. (You could bunny hop just as well by turning 90 degrees and pressing forward and backward instead of left and right)

[i]a[/i] for angle = direction you're accelerating in, as an angle in degrees or radians (take your pick).  That is, [b]h[/b] = <cos [i]a[/i], sin [i]a[/i]>

So here's how Portal 2 figures out your velocity when you're bunny hopping or crouch flying glitching:

[b]v[/b] can be broken down into two components: one parallel to [b]h[/b] and one perpendicular to [b]h[/b].  Here are two examples:
[img]http://img337.imageshack.us/img337/8857/bh2.png[/img] and [img]http://img502.imageshack.us/img502/4799/bh3.png[/img]
Portal 2 calculates the component of your velocity in the direction of [b]h[/b], which is the red line.  If it is less than 60, you accelerate in the direction of [b]h[/b] until your velocity in the direction of [b]h[/b] is 60.

For the math geeks among you, this means that your velocity can be approximated with the differential equation [img]http://img849.imageshack.us/img849/7677/bh4.png[/img] whenever you're actually accelerating.  If 60 - [b]h v [/b] is less than 0 then the equation does not hold; you never slow down in the direction you're trying to go.

I don't want to clutter up this post too much, so I'm throwing a graphical example of how bunny hopping increases your velocity into a hidden content box.  Click the button if you want to look at it.  Otherwise, below are some implications  about how to bunny hop in Portal 2.

[hidden]
You're moving with velocity old [b]v[/b] and you strafe in direction [b]h[/b] like so:
[img]http://img824.imageshack.us/img824/9452/bh7e.png[/img]
Portal 2 calculates the components of your velocity with respect to [b]h[/b]:
[img]http://img803.imageshack.us/img803/5570/bh8.png[/img]
Portal 2 looks at the component of your velocity in the direction you're trying to move and determines whether it's less than 60:
[img]http://img209.imageshack.us/img209/6271/bh9.png[/img]
If it is indeed less than 60, you gain velocity in the direction you're trying to move until you're going 60 in that direction:
[img]http://img683.imageshack.us/img683/5663/bh10.png[/img]
Your velocity is now new [b]v[/b]
[/hidden]

Here are the implications about bunny hopping:

Implication 1: The amount of velocity you gain when you bunny hop is directly related to how much you turn and weave.  It doesn't matter if you turn in a big circle or if you go left-right-left-right, but you're only getting faster if you're also changing direction.

Implication 1 disclaimer: Portal 2 only does its physics calculations a few times a second.  If you turn so fast that [b]v[/b] and [b]h[/b] make an angle of greater than 90 degrees, you'll just slow down.  That is, you slow down if the dot product of [b]h[/b] and [b]v[/b] is negative.

Implications 2, 3, and 4: (2) If the direction of [b]h[/b] is too close to the direction of [b]v[/b], you won't accelerate at all (specifically, you only accelerate if the dot product of [b]h[/b] and [b]v[/b] is less than 60). (3) If you know how fast you're going, you can calculate the size of this dead zone.  For example, your velocity is 175 when you're just walking normally.  Suppose you want to jump, turn [i]d[/i] degrees to the left and then strafe right to start bunny hopping.  Your speed in the direction of your strafe is given by 175 * sin [i]d[/i].  If this is greater than 60, you won't accelerate at all.  So if you set 175 * sin [i]d[/i] = 60 and solve for [i]d[/i], you get [i]d[/i] = arcsin(60 / 175) = 20.05 degrees.  (4) This means that if you walk forward and start bunny hopping, you should turn a maximum of 20.05 degrees before your first strafe; turning more won't help at all.

Here's a picture of implications 1-4:
[img]http://img833.imageshack.us/img833/8056/bh5.png[/img]

Implication 5: You gain exactly 0 speed when [b]h[/b] is at the top of the green arc in the picture of implications 1-4.  You gain the most speed when [b]h[/b] is at the bottom of the green arc.  You always gain the most speed if [b]h[/b] is perpendicular to [b]v[/b], but then you also end up turning the most.

Picture of implication 5 (Of course, 60 * [b]h[/b] should be 60 times as long as [b]h[/b].  None of these pictures are really to scale):
[img]http://img46.imageshack.us/img46/9451/bh6v.png[/img]

There's still much more research to do here.  I've only found a way to mathematically describe bunny hopping.  I'll work on ways to use this stuff to actually optimize bunny hopping at my leisure, and post anything interesting I find.  Feel free to bang on my Steam account (id: Negatratoron) if you have questions, comments, suggestions, or just want to talk about general abstract nonsense.

Amoirealinge

2012-01-08 01:28:46 am

(new acc. drMalcom)

Awesome work Negatratoron. If I understand this correctly I can get the most acceleration in the direction perpendicular to the one I'm facing before the strafe ?

Negatratoron

2012-01-08 12:32:05 pm

Quote from Amoirealinge:

Awesome work Negatratoron. If I understand this correctly I can get the most acceleration in the direction perpendicular to the one I'm facing before the strafe ?

Yeah, exactly. You generally get the most acceleration if you strafe in a direction perpendicular to the one you're flying in.

It's possible to quantify bunny hopping, but I doubt people are likely to think to themselves "Okay I need to turn 20 degrees then 10 degrees then 15.2 degrees....." and do a better job for it. In my opinion the most useful thing to get out of that bunny hopping post is that weaving = speeding up. I think a good bunny hopping strategy would be to keep that one thing in mind and let your brain take care of the rest. It's good at learning skills after all, and just thinking "weaving = speeding up" gives it a nice push in the right direction.

CBenni

2012-01-11 02:09:27 pm

I was wondering if there has been any findings on glitches to get out of bounds by now? At the moment speedruns practically consist of bunny hopping, button save glitches and excellent portal shots (and reportal in rare cases); OOB stuff would do a lot Cheesy

I've been playing a bit with those glitches and im extremely impressed with the stuff you guys find...
Keep it up!

cheers, Benni

RøvRapunzel

Edit history:

RøvRapunzel: 2012-01-12 02:00:38 pm

2012-01-12 01:59:45 pm

Well, out of bounds glitches has been found but i dont think it's connected to anything usefull yet.

CBenni

2012-01-12 02:07:54 pm

Oh nice, i bet DemonStrate will find something epic with this Tongue

I love how the graphics of Portal 1/2 glitch out completely when being out of Bounds^^ in other games you just fall down forever and stuff like that, but just look at that rainbow madness in Portal 1 Smiley

Thx for sharing that with me!

cu, Benni

DemonStrate

2012-01-13 05:31:34 am

If we were you, we would quit now.

I'll be back, don't you worry. I'm letting BananasaurusRex blow his load on the game. I will not compete with him for the challenge rankings, but I will use his strats and hopefully improve upon them for a full game speedrun. The one level I've seen him do that reminds me most of myself is this one:

Plus, I want to look for glitches in the non-DLC version, but everyone is playing the DLC game currently. I would run the non-DLC for obvious reasons.

Blazier

Edit history:

Blazier: 2012-01-13 08:17:33 am

2012-01-13 08:16:01 am

For OoB glitches there's this:

Let's hope it's useful somewhere.

Also, DemonStrate, good. I'd love to see you continue with your routes for the non-DLC game.

Darkevil

2012-01-13 12:56:54 pm

Negatratoron

2012-01-13 02:00:31 pm

Quote from Darkevil:

Oh god it's genius. Out of curiosity, did you know about the old button save glitch or did you just randomly find this?