The Smash Lab: What is it?

[leafgreen386]

Honestly, I would be surprised if the hitbox data for attacks wasn't located in folders specific to the characters, probably in the same location as the model data. And I can pretty much guarantee you that it isn't located on the flip side of the disc, for whoever suggested that (all characters function properly with just a single-sided disc; it's all the SSE data that's held on the second side).

As far as research is concerned (albeit not related to the topics set forth so far), kupo15 and myself spent some time to figure out how hitstun was calculated a few weeks ago (you can see our findings here), and are currently investigating how the knockback system works. So far it seems to be the same as melee, but we still need to figure out how the stale moves system factors into it.

 

[Mmac]

Geez! Don't you guys ever sleep?!?


Yeah, Steel2nd pretty much got it right. If only I could stay up to 5 am....

 

[Pr0phetic]

Honestly, I would be surprised if the hitbox data for attacks wasn't located in folders specific to the characters, probably in the same location as the model data. And I can pretty much guarantee you that it isn't located on the flip side of the disc, for whoever suggested that (all characters function properly with just a single-sided disc; it's all the SSE data that's held on the second side).

As far as research is concerned (albeit not related to the topics set forth so far), kupo15 and myself spent some time to figure out how hitstun was calculated a few weeks ago (you can see our findings here), and are currently investigating how the knockback system works. So far it seems to be the same as melee, but we still need to figure out how the stale moves system factors into it.

If i were to take a gander, when moves stale its like a negative value on the knockback, i wish I can look into it. I'm getting LoZ:TP soon so ima get into this.

Geez! Don't you guys ever sleep?!?


Yeah, Steel2nd pretty much got it right. If only I could stay up to 5 am....

Its around 4 pm by us since Steel is in NJ(i would think.)

 

[phi1ny3]

Odd, because the DTilt doesn't change really much for the pushback when stale. Maybe it's so small that it isn't really noticeable.

Ya, it's certainly not as noticeable as a smash. I would think that there is a connection between the two.

 

[Mmac]

Its around 4 pm by us since Steel is in NJ(i would think.)

I mean during the time he posted it... Although being on the East Coast probably explains things.

It seems to me that in simplistic terms, The Shield Stun and Hit Lag always gets reduced by roughly less than half when staled, no matter how small the knockback is. The more extreme the Knockback, the more quickly it diminishes.

 

[Adapt]

As far as research is concerned (albeit not related to the topics set forth so far), kupo15 and myself spent some time to figure out how hitstun was calculated a few weeks ago (you can see our findings here), and are currently investigating how the knockback system works. So far it seems to be the same as melee, but we still need to figure out how the stale moves system factors into it.

I can tell you all about knockback and what stale moves does to it. This research has already been performed by ph00tbag, FadedImage, and me.

Knockback has 2 components (it increases in a linear fashion y=a+b*x)

a) Initial Power - The theoretical knockback at 0% (this can never actually be achieved because knockback is calculated after damage is applied) This is your baseline.

b) Knockback Growth Rate (KGR) - This is the amount that knockback will increase with damage percentage.

Initial Power is unaffected when a move becomes stale. The only thing that changes is the KGR decreases. KGR decreases by around the same % that the damage of the move decreases by. Moves with higher KGR are more affected by stale moves. Similarly, moves with zero KGR are unaffected by stale moves (wrt damage and knockback anyway)

(look below for an example of the KGR decaying)

As a side note, a fully fresh move actually does 105% damage and knockback compared to the same move in training mode.

In the meantime, I've recently heard that stale moves have less shieldstun and pushback. Is this true? And if so, how does it work? Is the reduction substantial? Is the reduction gradual as the move stales? Does it ever get back to the original shieldstun before staling? If you're going to analyze specific attacks, then I recommend using MK's Dsmash, Dtilt, Fair, and/or Dair, as those are some of his more popular shield pressure attacks.

I want to add a bit to what Steel did.
I do not have the hardware necessary to count frames, but I was looking at the distance you slide back once you clear the stale move queue.

It does return to full distance after you stale the move completely and then clear the queue with a new move. I would guess that everything else returns to the original values, but I can't prove it without proper equipment. The pushback also seems to stale in the same way knockback does which is completely unsurprising. It's obvious that it is related.

Here's an example of decay that I have in my inbox:

Dedede's Fsmash:
1st hit: KGR - 55.75 - 100%
2nd hit: KGR - 48.37 - 86.60%
3rd hit: KGR - 43.95 - 78.80%
4th hit: KGR - 40.01 - 71.76%
5th hit: KGR - 36.57 - 65.59%
6th hit: KGR - 33.61 - 60.28%
7th hit: KGR - 31.16 - 55.89%
8th hit: KGR - 29.19 - 52.35%
9th hit: KGR - 27.71 - 49.70%
10th hit: KGR - 26.73 - 47.94%

This is DDD's fsmash decaying (research done by Faded). The most decay happens at the beginning, which is similar to how the shield pushback stales.

 

[M.K]

Here is whats going down:

Currently Ankoku has been added as a moderator for the Smash Lab, and he'll be helping me direct the room once it gets started. This week I have finals. So the official opening of the Smash Lab will be early in the week of the 14th. That means one more week to go.

In the meantime, I've recently heard that stale moves have less shieldstun and pushback. Is this true? And if so, how does it work? Is the reduction substantial? Is the reduction gradual as the move stales? Does it ever get back to the original shieldstun before staling? If you're going to analyze specific attacks, then I recommend using MK's Dsmash, Dtilt, Fair, and/or Dair, as those are some of his more popular shield pressure attacks.

Remember that the admissions list isn't finalized just yet, so you still have time to make an impression.

I just quoted the post to answer the question, and use as a reference.
I do not have access to picture or video-capture tools, so please bear with my cute little graphic organizer!

**This test was completed on an Infinite Time Brawl, not Training Mode. Therefore, Move Decay still applies**

In this test, I created a custom stage consisting of blocks. There were about 10 blocks in each direction.
I positioned a Dedede on the last block on the right. I positioned another Dedede on the line that seperates Block #1 and Block #2.

I used a Non-Staled Dedede FSmash on the other Dedede while it was sheilding. The Dedede traveled 1 and 1/2 blocks to the left of it's starting point and lost most of it's shield.

After using the FSmash on a spare Mario 3 more times (for a total of 4 times), I once against positioned the Dedede on the rightmost block, and the other Dedede on the line seperating Block #1 and Block #2.

Using the Dedede FSmash on the other Dedede while it was shielding resulted in a Shield Push of about 3/4 of a block from the starting point. The shield still lost most of it's fullness, but there was less hitstun than a non-stale FSmash.

Conclusion

Stale moves cause less knockback and shieldstun than non-stale moves, resulting in a shorter period in which the opponent being attacked is able to regain control of their character and counterattack, escape, jump, etc.​

 

[ph00tbag]

I wasn't suggesting they'd be scattered around, but rather that they may be hidden inside of some larger file structure.

Ah, well then that's my mistake.

Although at the very least, I'm pretty sure I recall some hackers finding hex commands to induce certain animations. I dunno, it may take a great deal of brute force trial and error, but I feel like if there's an object in the game, it should be somehow modifiable through hax. I'm admittedly far less along than you in programming, so perhaps you can clear up any misguidance I'm weighted with.

 

[Pr0phetic]

Good stuff Meta-Kirby, that ontop of Steels stuff is nice information.

 

[leafgreen386]

stuff

Ah, I didn't know the research had already been done, but yeah... we came to the same conclusion that knockback growth is linear. I had already assumed that stale moves would affect only the incremental portion of the kb, and not the base kb, seeing as that was how weight worked.

Do you have a link to your research thread? I searched the tactical forum and found this thread by Mr.Silver, but that's just the knockback formula, without the effect of stale moves on it, and I think his fall rate data is wrong.

I did find something else interesting, though. The best I can tell, there is no actual acceleration due to gravity in brawl. When I tested it on a custom stage 4 blocks tall, and then dropped, it took the same number of frames for mario to travel the distance of the first block as it did to travel the distance of the last block (they were each 24 frames). The last block should've taken less time to travel if there was acceleration due to gravity in brawl. As for why it appears there is acceleration in brawl, I offer an explanation on how I think it works in this post, and have more about how I think it works in other posts in the last page or two of the gravity mod thread.

The ultimate goal I'm looking to atm is cracking the game's units and the momentum calculations. I'm not sure if it's possible or not, but I have ideas on how I'm gonna get there.

 

[M.K]

I did find something else interesting, though. The best I can tell, there is no actual acceleration due to gravity in brawl. When I tested it on a custom stage 4 blocks tall, and then dropped, it took the same number of frames for mario to travel the distance of the first block as it did to travel the distance of the last block (they were each 24 frames). The last block should've taken less time to travel if there was acceleration due to gravity in brawl. As for why it appears there is acceleration in brawl, I offer an explanation on how I think it works in this post, and have more about how I think it works in other posts in the last page or two of the gravity mod thread.

If acceleration due to gravity was present in the mechanics of Brawl, it would defeat the whole motion of fast-falling.
I have no idea why charaters like Fox and Falco fall faster than a character like Bowser.
If anything, all characters should fall at a constant acceleration. They fall at the same acceleration, 9.8 m/s because the gravitational pull is same for all objects irrespective of their mass. Terminal velocity is different concept .An object falling down tends to accelerate due to gravity,but the air will push it upwards. so both forces reach an equilibrium and the object stops accelerate. Then it will acquire a constant velocity called terminal velocity which turn sout to be it's maximum.
Thus, the heavier characters should theoretically fall faster in Brawl, which they do (for the most part)

 

[Adapt]

Ah, I didn't know the research had already been done, but yeah... we came to the same conclusion that knockback growth is linear. I had already assumed that stale moves would affect only the incremental portion of the kb, and not the base kb, seeing as that was how weight worked.

Do you have a link to your research thread? I searched the tactical forum and found this thread by Mr.Silver, but that's just the knockback formula, without the effect of stale moves on it, and I think his fall rate data is wrong.

The topic is not created yet, which is why you never knew. I have been after FadedImage to post his findings for months now....

Seeing as he did most of the stale moves research I think it is his topic to create.

I have been trying to pin down weight and fall speed. I have gotten some values that are similar to Mr. Silver's but different. They do however follow the same approximate trend.

I want to check the values I have with more moves, but I haven't has the time to work on it because I am in the exam period for the semester. I have enough time to do small things, but fall speed and weight are much more work.

 

[phi1ny3]

If acceleration due to gravity was present in the mechanics of Brawl, it would defeat the whole motion of fast-falling.
I have no idea why charaters like Fox and Falco fall faster than a character like Bowser.
If anything, all characters should fall at a constant acceleration. They fall at the same acceleration, 9.8 m/s because the gravitational pull is same for all objects irrespective of their mass. Terminal velocity is different concept .An object falling down tends to accelerate due to gravity,but the air will push it upwards. so both forces reach an equilibrium and the object stops accelerate. Then it will acquire a constant velocity called terminal velocity which turn sout to be it's maximum.
Thus, the heavier characters with smaller air resistance should theoretically fall faster in Brawl, which they do (for the most part)

Fix'd. That could explain a bit why the star chars fall fast, even though they aren't necessarily heavy. Terminal velocity is fun, it makes small, itty bitty animals survive big heights of falling (besides body/leg structure).

 

[Pr0phetic]

Ha, its funny that we should incorperate Earth physics into Brawl physics. I doubt the developers were thinking of Free Fall, acceleration due to gravity and constant velocities when they made their game. Lets do projectiles next, like how they should eventually fall.

But it sure would make a hell of a lot more sense...

 

[infomon]

I doubt the developers were thinking of Free Fall, acceleration due to gravity and constant velocities when they made their game.

I'm sure they were.

.... just sayin'.

 

[Browny]

in a game where snakes uptilt even exsits, what made you think devs had any care about real physics?

 

[PKNintendo]

in a game where snakes uptilt even exsits, what made you think devs had any care about real physics?

Or his forward tilt.

 

[Pr0phetic]

Or his forward tilt.

My point exactly, I love how his foot goes straight up and lets see... the force of his UTilt is projected out towards the opposing character and up.

Brawl's physics engine is a little more straightforward, you go up, you will eventually come down, you are hit, you will be sent in a direction. However, there's more influences such as DI and SA, so you throw those into the mix.

 

[leafgreen386]

It's amazing how much easier hacking makes this stuff (and you also get to see some funny things such as having an infinite float with every character when you walk offstage =D).

Anyway, when I set the gravity to zero, I found out that the "force degeneration" I thought I was seeing is actually solely due to gravity, which makes such perfect sense I don't know how I ever thought otherwise (it might have something to do with me actually being right about this as far as horizontal movement is concerned). The main problem I was running into when reasoning through this is that knockback is treated differently from other forces in this game. When I used fox's usmash against a stationary link in training mode, the link traveled upward from the usmash, reached a certain height and stopped his vertical movement while continuing moving to the left (he gradually came to a stop). The vertical knockback looked constant throughout the whole of it, before just abruptly disappearing (although the change in animation had me doing a double take for a moment there as it looked like they were boosted downward at the end of it going into the tumble). Something else interesting was that attacks had the same "max launch speed" in the end of game statistics as they do in regular brawl, which I guess confirms that gravity is not applied to the first frame of knockback (although you could've figured that out just by going into heavy/light brawl).

I'm not sure if knockback gradually decreases each frame due to gravity or not. It looks like it... barely, but looks can be deceiving. I mean, the fact that smoke clouds from high knockback attacks stop before knockback actually ends suggests this more than actually looking at the knockback in 1/4 speed does (edit: no it doesn't - I forgot to look at this when I had the code on before, but the smoke definitely disappears before the knockback is over in zero grav, too). It's possible that the knockback each frame is represented by x - g, where x is the knockback, and g is how much less distance you will fly due to gravity sending you downward that frame. Of course, the launch speed thing I showed earlier only shows that it doesn't affect velocity on the first frame of knockback; it could still be that knockback is gradually getting less each frame in the y direction. I don't think we're gonna get anything on this until it's looked at frame by frame.

Now, when I did a jump or used any other momentum altering move, such as diddy's upB, fox's fair, fox's upB in some direction besides straight right/left, or one of the various fast fall dairs, I just kept rising or falling forever, never stopping. Well, unless I had downward momentum and fast fell, which caused me to stop all vertical momentum completely. Something else I found interesting that came out of this was that some upBs actually cause you to have downward momentum afterward, such as mario's, ganon's, wario's, falcon's, and a few others (including samus' grounded upB; her aerial one has her going upward forever). All effects of moves that give you downward momentum (not knockback) seem to just completely replace gravity with some other value, which the effect can then be canceled by initiating a fast fall or giving yourself upward momentum. Also, lucas', ness', and yoshi's second jump don't provide any actual upward momentum after the jump is complete.

And as we already know, you're already at your maximum fallspeed (excluding fast falls, but even by looking in 1/4 speed you can see that they're also a constant falling rate) when you just walk offstage, so this is probably the same rate at which force is subtracted from knockback and other momentum.

That all said, it seems obvious enough that the game calculates your vertical momentum in two completely different components - knockback and other momentum altering effects. Force due to knockback will disappear entirely after a certain number of frames equal to the hitstun (up until when you go into the tumble), while force from other momentum altering effects must completely be canceled out by gravity before they disappear, at which point the character falls at a constant rate.

So I guess now you know why fast falling during knockback works, now

As for horizontal momentum, although it's obvious enough on its own in regular gravity, I guess I'll mention something there, anyway. When you're hit by an attack that gives you horizontal momentum, after the knockback ends, you will continue to have horizontal momentum that will gradually decrease according to what appears to be a square root curve. You can see this same behavior by simply pressing left or right and then releasing. You don't stop horizontal movement immediately, and instead continue moving for a short while as you slow down.

 

[infomon]

^^ Awesome. That pretty much validates a lot of my understanding of hitstun/knockback that I've been basing my momentum-cancelling stuff on.

 

[Pr0phetic]

Awesome stuff LeafGreen, I have questions for you later, good stuff once again.

 

[ph00tbag]

Leafgreen, I just have to quibble about terminology, because I'm just kind of confused by a few of the things you've said.

First of all, I don't think that knockback ends when histun ends. I imagine that the "x-g" function (or more specifically "sin(x)-g") is constantly happening until x=0. The thing is, hitstun is defined as the time during which no input can be made (save airdodging after a certain point), So the only thing counteracting knockback is gravity. But once hitstun ends, you can perform any input you want, so fast falling and air control begin to negate knockback as well.

I'm sure this is what you meant, but it's not what I was getting.

Now here's another thing I'm curious about. Are you suggesting that gravity doesn't act always downwards as I've described it, or does it act antiparallel to the knockback? If the former, then your analysis has nothing counteracting horizontal velocity during hitstun. That's interesting. Do you have any data to confirm or deny that?

 

[infomon]

Actually ph00tbag, even after you cancel hitstun with an aerial/airdodge/z-air/item-throw, but while you're travelling away from the stage in knockback still..... you don't have control of your aerial mobility yet; aside from fastfalling, directional input does nothing until after the point that hitstun naturally would have ended (and turned into a tumble).

.... at least, that's from my own experiments, in "regular Brawl", training mode.

........ not sure if that's at all relevant to the gist of your post lol, oh well.

 

[ph00tbag]

Actually ph00tbag, even after you cancel hitstun with an aerial/airdodge/z-air/item-throw, but while you're travelling away from the stage in knockback still..... you don't have control of your aerial mobility yet; aside from fastfalling, directional input does nothing until after the point that hitstun naturally would have ended (and turned into a tumble).

.... at least, that's from my own experiments, in "regular Brawl", training mode.

........ not sure if that's at all relevant to the gist of your post lol, oh well.

Really? I didn't utilize much rigor, but just now I achieved much more knockback reduction after using an aerial than without. It's more apparent with lower knockback, because the air control has a greater effect on lower knockback speeds. I don't think you'd get a very big difference in survivability once percents are high, and knockback sends one faster. Probably no more than a percent.

I do think it's interesting, though, that indeed, there is something reducing knockback horizontally. So either Leafgreen's "g" is a value anti-parallel to knockback, or there is an effect from gravity and from something else. M2K did some testing earlier on which may have indicated the latter, as he found that every character has a vertical and horizontal weight, although I don't remember the specifics of his findings.

 

[leafgreen386]

Leafgreen, I just have to quibble about terminology, because I'm just kind of confused by a few of the things you've said.

First of all, I don't think that knockback ends when histun ends. I imagine that the "x-g" function (or more specifically "sin(x)-g") is constantly happening until x=0. The thing is, hitstun is defined as the time during which no input can be made (save airdodging after a certain point), So the only thing counteracting knockback is gravity. But once hitstun ends, you can perform any input you want, so fast falling and air control begin to negate knockback as well.

I'm sure this is what you meant, but it's not what I was getting.

No. Knockback will literally disappear the moment hitstun naturally ends. I was in zero gravity when testing this. If it did not disappear after a set number of frames, then characters would keep flying forever and ever (at least in the y direction).

Now here's another thing I'm curious about. Are you suggesting that gravity doesn't act always downwards as I've described it, or does it act antiparallel to the knockback? If the former, then your analysis has nothing counteracting horizontal velocity during hitstun. That's interesting. Do you have any data to confirm or deny that?

Also no. I think there actually is a horizontal momentum degeneration, and that force is definitely not gravity. When floating around in zero g, after I released the control stick, it still took some time before the character slowed to a stop. The same thing happened when I used fox's usmash - they kept moving in the horizontal direction until slowing to a stop.

 

[Pr0phetic]

It sounds like they did take some keys from Physics, such as resultant forces.

 

[ph00tbag]

No. Knockback will literally disappear the moment hitstun naturally ends. I was in zero gravity when testing this. If it did not disappear after a set number of frames, then characters would keep flying forever and ever (at least in the y direction).

So they literally came to a stop after hitstun ended? That's not how I understood your first post.

 

[SCOTU]

you know, i'm no expert on brawl hacking at all, but how do you guys get your codes? is it one of those run checks for changes in memory sort of deals? I'm just curious, cause it seems like it could be very easy to mess something up and edit a wrong value that could look like the right value when testing, but actually turn out to fundamentally affect the physics in an undesirable way for testing. Again, i'm not criticizing the way ppl hack, i'm just questioning to further my understanding of it.

 

[phi1ny3]

It's amazing how much easier hacking makes this stuff (and you also get to see some funny things such as having an infinite float with every character when you walk offstage =D).

Anyway, when I set the gravity to zero, I found out that the "force degeneration" I thought I was seeing is actually solely due to gravity, which makes such perfect sense I don't know how I ever thought otherwise (it might have something to do with me actually being right about this as far as horizontal movement is concerned). The main problem I was running into when reasoning through this is that knockback is treated differently from other forces in this game. When I used fox's usmash against a stationary link in training mode, the link traveled upward from the usmash, reached a certain height and stopped his vertical movement while continuing moving to the left (he gradually came to a stop). The vertical knockback looked constant throughout the whole of it, before just abruptly disappearing (although the change in animation had me doing a double take for a moment there as it looked like they were boosted downward at the end of it going into the tumble). Something else interesting was that attacks had the same "max launch speed" in the end of game statistics as they do in regular brawl, which I guess confirms that gravity is not applied to the first frame of knockback (although you could've figured that out just by going into heavy/light brawl).

I'm not sure if knockback gradually decreases each frame due to gravity or not. It looks like it... barely, but looks can be deceiving. I mean, the fact that smoke clouds from high knockback attacks stop before knockback actually ends suggests this more than actually looking at the knockback in 1/4 speed does (edit: no it doesn't - I forgot to look at this when I had the code on before, but the smoke definitely disappears before the knockback is over in zero grav, too). It's possible that the knockback each frame is represented by x - g, where x is the knockback, and g is how much less distance you will fly due to gravity sending you downward that frame. Of course, the launch speed thing I showed earlier only shows that it doesn't affect velocity on the first frame of knockback; it could still be that knockback is gradually getting less each frame in the y direction. I don't think we're gonna get anything on this until it's looked at frame by frame.

Now, when I did a jump or used any other momentum altering move, such as diddy's upB, fox's fair, fox's upB in some direction besides straight right/left, or one of the various fast fall dairs, I just kept rising or falling forever, never stopping. Well, unless I had downward momentum and fast fell, which caused me to stop all vertical momentum completely. Something else I found interesting that came out of this was that some upBs actually cause you to have downward momentum afterward, such as mario's, ganon's, wario's, falcon's, and a few others (including samus' grounded upB; her aerial one has her going upward forever). All effects of moves that give you downward momentum (not knockback) seem to just completely replace gravity with some other value, which the effect can then be canceled by initiating a fast fall or giving yourself upward momentum. Also, lucas', ness', and yoshi's second jump don't provide any actual upward momentum after the jump is complete.

And as we already know, you're already at your maximum fallspeed (excluding fast falls, but even by looking in 1/4 speed you can see that they're also a constant falling rate) when you just walk offstage, so this is probably the same rate at which force is subtracted from knockback and other momentum.

That all said, it seems obvious enough that the game calculates your vertical momentum in two completely different components - knockback and other momentum altering effects. Force due to knockback will disappear entirely after a certain number of frames equal to the hitstun (up until when you go into the tumble), while force from other momentum altering effects must completely be canceled out by gravity before they disappear, at which point the character falls at a constant rate.

So I guess now you know why fast falling during knockback works, now

As for horizontal momentum, although it's obvious enough on its own in regular gravity, I guess I'll mention something there, anyway. When you're hit by an attack that gives you horizontal momentum, after the knockback ends, you will continue to have horizontal momentum that will gradually decrease according to what appears to be a square root curve. You can see this same behavior by simply pressing left or right and then releasing. You don't stop horizontal movement immediately, and instead continue moving for a short while as you slow down.

So obviously, we know that Brawl gravity is much different from our gravity, seeing how moves will do a host of different things, knockback affects, and there's no natural acceleration (except from fastfall, etc.). Also, if gravity is involved in any horizontal knockback, then there is no way it is like real gravity.
Question: Would this explain some things like snake grenade manipulation? All this time, I thought that was the result of bad programming (giving events to the same variable, like maybe gren_1, as opposed to 2 different ones). But would altered physics affect the sudden change from that? Judging by this, I would say it wouldn't much, but I wouldn't know. Also, Snakes recovery has "inertia" when starting after falling motion. If there is no brawl gravity... I can see what would happen, nevermind.

 

[Red Arremer]

Of course the physics in Brawl are different to reality. The game has princesses firing magic fireballs and small yellow rats shooting electricity out of their cheeks at space bounty hunters and Italian plumbers.

I think the fast fall speed on the space animals has been included to give the cast more variety. We have quick fastfallers but also quick floaty characters. Best example for a Brawl addition of new type is Wario, who's a floaty and fast heavyweight. Or, as mentioned, the space animals, who are lighter, but fall insanely fast (specially Fox, who has, as far as I remember, the highest fall speed).

Of course, figuring out what is causing the movement in knockback is highly interesting, but I wouldn't say that gravity in the game is really comparable to the physics in our real world.

 

[leafgreen386]

I was in a bit of a rush this morning when I posted, so a few more things now:

1) Thanks guys =)

2) In my big post earlier, I said that I think horizontal momentum degeneration is done according to a square root curve. It would probably be more accurate to model it as an inverse square root curve. Of course this is just from eyeballing it, and there's a strong likelihood that the game does it differently. I think I'll have a way to accurately determine this sometime in the near future.

3) I'm still far from done. I think I've figured out how to crack the game's units from this, which I'll elaborate on after answering a few other posts.

So they literally came to a stop after hitstun ended? That's not how I understood your first post.

Yeah. Sorry if it wasn't clear. What would you suggest I change to improve the clarity?

you know, i'm no expert on brawl hacking at all, but how do you guys get your codes? is it one of those run checks for changes in memory sort of deals? I'm just curious, cause it seems like it could be very easy to mess something up and edit a wrong value that could look like the right value when testing, but actually turn out to fundamentally affect the physics in an undesirable way for testing. Again, i'm not criticizing the way ppl hack, i'm just questioning to further my understanding of it.

I'm not a hacker, myself, so I'm a little fuzzy on the details, but from my understanding... pretty much, yeah. Experienced hackers probably have other methods, but something simple like the gravity hack I was using could probably be done just by going into heavy and light brawls and comparing what changed from regular brawl.

So obviously, we know that Brawl gravity is much different from our gravity, seeing how moves will do a host of different things, knockback affects, and there's no natural acceleration (except from fastfall, etc.). Also, if gravity is involved in any horizontal knockback, then there is no way it is like real gravity.
Question: Would this explain some things like snake grenade manipulation? All this time, I thought that was the result of bad programming (giving events to the same variable, like maybe gren_1, as opposed to 2 different ones). But would altered physics affect the sudden change from that? Judging by this, I would say it wouldn't much, but I wouldn't know. Also, Snakes recovery has "inertia" when starting after falling motion. If there is no brawl gravity... I can see what would happen, nevermind.

You didn't have to quote my whole post lol

Anyway, I have no idea what the snake grenade manipulation thing you're talking about is, so I can't really answer that. Your fast fall comment is actually wrong - I'm pretty sure that your character starts to fall at the higher rate immediately, with no actual acceleration involved. Furthermore, gravity is not involved in horizontal knockback. If I gave this impression, then I'm sorry, but what I'm saying is that horizontal momentum (maybe including knockback and maybe not) does have some method of degeneration that is not related to gravity. When horizontal knockback (or any other force, such as you pressing left/right in the air) wears off, the character continues drifting to the side until their momentum eventually tapers off.

--------

Now that that's out of the way, I have good news. I just saved a bunch of money on my car insurance by switching to geiko.

No, not really, but I think I did figure out how to crack the game's units. I'm gonna need a move that sends characters perfectly vertical (not zelda's usmash since it lifts you before launching you, but snake's utilt might work; edit: lugi's upB vs 0% jiggs works perfectly!), and also a character and percent to use it against that sends that character at a launch speed of exactly 4000.

See, after finding out that knockback disappears the moment hitstun ends, and also that knockback sends you the same number of units vertically each frame in the absence of gravity, I can use the knowledge I collected from finding the hitstun calculation in order to figure out exactly how many units an attack will launch a character vertically. This requires me to make one (somewhat big) assumption, however, that being that the launch speed the game reads is the number of units that the game sends you on the first frame of knockback. I think this is a fairly valid assumption, considering that the launch speed varies appropriately depending on the game speed; on regular brawl snake's utilt sends a 0% pit at 2794 upf (units per frame - which is a whole lot more accurate than saying mph), and at half speed it sends pit at 1397 upf.

So, if my assumption is correct, then the number of units the game sends you from any attack can be represented by:

u = l * math.ceiling ( l / 165 )

u is of course the number of units you travel due to knockback
l is the launch speed
math.ceiling, for those not familiar with the notation, is simply saying to round the value in the parenthesis up to the nearest whole number

Now, for why I need the launch speed to be exactly 4000:

u = 4000 * math.ceiling ( 4000 / 165 )
u = 4000 * math.ceiling ( 24.2 )
u = 4000 * 25
u = 100,000

Using the stage builder blocks as a reference, I'll be able to easily see what the game defines as 100,000 units in a zero gravity environment (I'm really hoping that a stage builder block is 25,000 or 10,000 units, here). After defining what 100 ku (kilounits (and yes this is now metric)) looks like, I'll be able to generate exact values for the effect of gravity on characters when they fall in a regular environment, by seeing how many frames it takes them to travel some 500-1000 ku distance - whatever I need to give me an accurate measure.

After defining the unit and figuring out the fall rate of characters, it will become possible to find the exact effect of gravity on knockback, and even find the force applied to characters when they jump.

Yeah. I wasn't kidding when I said I wanted to crack the game's units.

 

[infomon]

I didn't read all ur post... but Snake's Utilt is not perfectly vertical, it goes behind him a bit. Use Luigi's up-B instead, that's what I used for all my vertical tests.

 

[leafgreen386]

Ah, good to know. In fact, I just tested luigi's upB against jigglypuff, and it gave me a launch speed of exactly 4000 upf. So I'm all set to do my tests! =D

edit: ok, I tested it. Assuming I'm right about the way knockback and launch speed works, then this should all be correct.

I made a custom stage as tall is it would go, then went there with jiggs and luigi in zero gravity. I had jiggs float off the edge, then back onto the stage. In her freefall animation, her hands are level with the ground, so this made a really good point of reference. I upB'd her with luigi, counted the number of blocks she went, and was disappointed to find out that it wasn't a perfectly whole number of blocks. Jiggs was launched 6 whole blocks, and stopped approximately 4/5 up the way of the 7th (her hands were level with the set of bricks closest to the top of the block).

After this test, I think it would be accurate to say that the stage builder blocks are 14,700 units tall.

(6 * 14700) + (4/5 * 14700) = 99600

Which is very very close to 100,000, certainly close enough to say that any variation is due to jiggs not being perfectly at 4/5 of the way up the block.

And with that, I'm gonna be done with testing for a bit. In order to figure out stuff like fall rates I'm gonna need to view the game frame by frame, and I don't have access to recording equipment.

 

[ph00tbag]

Yeah. Sorry if it wasn't clear. What would you suggest I change to improve the clarity?

I think I may just have gotten confused while reading it. I also somewhat resisted the notion in general, because it leaves me with more questions than I started with.

Namely, why do we keep moving after hitstun with normal gravity, but not with 0 gravity?

 

[M.K]

I think I may just have gotten confused while reading it. I also somewhat resisted the notion in general, because it leaves me with more questions than I started with.

Namely, why do we keep moving after hitstun with normal gravity, but not with 0 gravity?

Maybe because there is no gravitational force acting upon the objects?
Newton's Laws of Motion?

 

[leafgreen386]

I think I may just have gotten confused while reading it. I also somewhat resisted the notion in general, because it leaves me with more questions than I started with.

Namely, why do we keep moving after hitstun with normal gravity, but not with 0 gravity?

I think the animation change is fooling you, by making it look like they're moving up and then start to fall. This is especially noticeable with mk, and had me confused for a moment until I realized what I was seeing.

Also, note that when I say "hitstun," I mean the full length of hitstun, and not the point you can aerial/airdodge out of it at.

 

[infomon]

The transition from hitstun to tumble happens a bit before the end of knockback, yes?

OR, does the transition happen when knockback ends, but there's additional momentum (along knockback direction) that's "under the user's control" and not really knockback?

Since I've noticed that one's aerial mobility seems to come into effect before knockback actually ends; hence, leaning towards the stage can be enough to survive some would-be-KOs.

Equivalently, afaict you have full aerial mobility during tumble.

 

[leafgreen386]

The transition from hitstun to tumble happens a bit before the end of knockback, yes?

OR, does the transition happen when knockback ends, but there's additional momentum (along knockback direction) that's "under the user's control" and not really knockback?

Since I've noticed that one's aerial mobility seems to come into effect before knockback actually ends; hence, leaning towards the stage can be enough to survive some would-be-KOs.

Equivalently, afaict you have full aerial mobility during tumble.

There is definitely a transition between "I'VE BEEN HIT!" and tumble (in fact, two transitions). The first transition has you still in stun, but knockback has stopped. The second transition has you coming out of hitstun and going into your true tumble animation, during which you have air control beginning at the transition. I don't know if this acts any different in regular brawl, seeing as I'm playing using mods that prevent you from acting while in hitstun.

I tested by just going into training mode with snake and using utilt against a pit set to control at 1/4 speed. First, I just observed what animation pit went through, then I redid the test, this time holding toward snake as soon as hitlag ended.

This is gonna make me have to reexamine my unit tests, now, as this could prove to be a fatal flaw, although I am glad this came up, instead of it just having gone by ignored. The problem is I'm not the one who was examining the video when figuring out the hitstun formula (kupo15 was, since he was the one with recording equipment). I just used the numbers I was given. In light of this, I may have to revise not only my unit findings, but also the hitstun formula. It's fine, though, since it's always better to be wrong and correct it than to be wrong and never know the truth. edit: Ok, hitstun formula is fine, but I'm going to need to reevaluate how I calculated what would be 100,000 units, seeing as my assumption of the number of hitstun frames being equal to the number of knockback frames was obviously wrong.

edit: Honestly, the more I test, the more I'm getting confused by what I'm seeing. It seems that with some moves, you have knockback right into the tumble, while on others the knockback is gone by what I was labeling before as the second transitional phase.

 

[phi1ny3]

Sorry, I meant that Acceleration that is part of natural gravity is nonexistent, and only happens when u fast fall.

 

[leafgreen386]

Ok. I think I’m starting to get what I’m seeing here.

First off, I’d like to apologize for posting some erroneous “findings.” My first large post (link) is still at least partly correct, although it is now obvious I was wrong on a few accounts. The primary error I made was in only testing one level of knockback and assuming it acted the same at higher levels. I’ve made a couple assumptions here, some with decent reasoning behind them (such as what launch speed means to the game) and others (such as this one) where I was just not thinking things all the way through.

Anyway, what I think is happening here is that characters are propelled in the direction of the knockback at a constant rate (it might not be quite so constant once you factor in gravity) until “transition 1,” at which point a force is applied to the character that is counter parallel to the force of the knockback (but the character is still in stun until “transition 2,” when they begin going into the tumble animation and regain air control). This force is applied until the force due to knockback is either equal to or less than 0, at which point it stops being applied. This force acts similar to gravity in that it is applied each frame and subtracts from the current velocity to be used for the next frame; if it didn’t exist then the force of the knockback would probably never subside, similar to the way jump force never subsides without the existence of gravity, except this one isn’t in effect at all times. I think this is why sometimes it looks like attacks just instantly end their knockback or even cause the character to be propelled downward at the end of it – the launch speed was low to begin with. Having a constant counter parallel force to the knockback would also explain why stronger knockback continues past transition 1 into transition 2 and even into the tumble, as the knockback would not have fully dissipated by then.

The most important thing to note, though, is that this force is not applied until transition 1, which I believe happens halfway through hitstun. I noticed the existence of a slowing effect when I repeated the luigi/jiggs test again – jiggs traveled at what appeared to be a constant rate until the end, when she started visibly accelerating to a stop (I did this in slow brawl). I came to the conclusion that it starts halfway through hitstun when I was viewing a video of marth fsmashing a 0% pit frame by frame, and noticed that pit first started changing animations 11 or 12 frames in (I couldn’t really tell which marked the change), while I know that the hitstun lasts for 23 frames.

If this ends up being true, then this could help explain why KOs are affected by stale moves so badly, as not only is knockback getting weakened due to the lower launch speed, but hitstun is also decreasing, which then causes knockback to get weaker sooner in the trajectory. Of course, although unlikely, it’s also possible that the counter parallel force is in effect from the very beginning of knockback. I’m gonna have to look at this frame by frame later today.

I’d also like to note that what I described earlier as fox usmashing link and the link halting on the y-axis but continuing moving slowly to a stop on the x-axis is pretty much well... wrong. It was simply the result of me forgetting to change the computer to “control” and so it DIed/used air control, which screwed up my results. It is true that momentum gained from air control gradually degenerates if you let go of the control stick, but force due to knockback is a completely different deal.

Sorry, I meant that Acceleration that is part of natural gravity is nonexistent, and only happens when u fast fall.

Even still, I'm pretty sure fast falling just causes an immediate change in the gravity, so you're going from one constant to another constant.