• Welcome to Smashboards, the world's largest Super Smash Brothers community! Over 250,000 Smash Bros. fans from around the world have come to discuss these great games in over 19 million posts!

    You are currently viewing our boards as a visitor. Click here to sign up right now and start on your path in the Smash community!

Important Foxy's Guide to Testing and Labbing

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
Obligatory Intro.

Do you have questions about the game or want to know how it works more in depth? Do you feel that there's not much you can do because you don't have the equipment or the know-how, or you don't have someone else to help you? If you answered 'yes', then read on.

In this thread I'm going to explain in detail the tricks, methods, and knowledge I use to figure out all kinds of things with nothing more than a wii-u, the game, and two controllers. With just these bare essentials and a little bit of ingenuity, there isn't much that can't be figured out accurately.



Let's start out with some basic tools of the trade:

The 2 Frame Skip.

This will become a staple technique that you will use all the time.
Essentially, if you go to training mode and set the speed to '1/4 hold', you can accurately skip forward exactly two frames with perfect accuracy by hitting the 'L' button (on a gamecube controller; make sure it's not old or faulty) and immediately releasing it.

This works because 2 frames is the minimal amount of frames that can be moved forward, i.e. even if you were inhumanly quick at hitting and letting go of the L button, the game would still skip forward exactly two frames. If you get lazy it is of course perfectly possible to skip forward three frames, but it's not difficult to be 100% accurate with this every time if you're trying.

If e.g. you want to see the second frame of an action, you put it on 1/4 hold, for now let go of L, hold the button and/or joystick input that will activate the desired action (e.g. A to jab), then quickly tap the L button. Your character would have started the action and the game should be stopped on the second frame of that action. To see the fourth frame, of the action, just tap L again, etc.

To check if you've got this right, try going into training mode with Diddy, making him stand up close in front of the opponent then let go of L, and attempt to use a Jab to hit the opponent, except skip forward by two frames only. If you did it right, Diddy should not have hit the opponent because his Jab's hitbox comes out on frame 3. Not that difficult right?

Now let's say you don't believe me; let's say you actually think that you should be able to skip just 1 frame if you're quick enough. By all means, be my guest. Pick a character with a 2 frame move e.g. Fox's Jab, make sure Fox is in a neutral state standing in front of an opponent, use 1/4 hold, then hold A and tap L as fast as you like. (Don't accidentally use Jab 2 and not realise your mistake please...)
No matter how long or how hard you try, Fox will always hit with Jab 1.

Just so we're perfectly clear, when you have the game paused by not holding L, and you input (and hold) an action and then tap L, the input of that action doesn't actually register with the game until the game is moving, i.e. until you tap L, or another way to think of it would be until the following frame, i.e. the first frame of that action. Throughout this thread I'll be using the latter assumption even though technically the former is more correct, simply because I find it easier to talk about inputting actions on a specific frame rather than talking about inputting actions in between frames when the whole 'in between frames' theory only applies to the case of holding inputs during a paused game then moving forward with L. To say this again, because it's important that you understand, I'll be assuming that the first frame of the action or movement or animation or whatever, is the frame that the game registers the input for that action/movement/animation. The reason for this feat of mental gymnastics is because even though it feels like you're inputting the action earlier, i.e. on the frame you have the game paused, this would be a false assumption as the game does not behave as if this is true and instead will wait till you've hit L before registering the input, i.e. necessarily after the frame you were paused on. This is why you need to hold the input when tapping L, otherwise the input will not register at all. So again, the assumption is that while looking at the game paused on a frame, it will be the very next frame that will be the frame that the action is registered as an input, not on the frame it is paused on.

You may have noted that by using the 2 frame skip you can have access to every second frame, e.g. frames 2, 4, 6, 8 etc. But what about the frames in-between? Well, by using the two following skips you can have access to frames 1, 3, 5, 7 etc too.


The 1 Frame Skip.

The 1 Frame Skip used to be the primary way to access all the negative frames, but since the discovery of the 3 frame skip, it is mostly used for accessing frame 1 only now. In any case, if you so wish, you can still use a combination of 1 and 2 frame skips to access any frame if you don't want to fiddle around with the speeds [see the 3 frame skip below].

To skip forward by 1 frame, you need to once again be in training mode with 1/4 hold speed, hold the L button, then as you're releasing the L button, input the desired action (e.g. jump or an attack). If you do it correctly, you would have paused the game on the 1st frame of that action.

What you want to do is make adjustments by increments. At first aim to input the desired action way too late, as in, you've already let go of L, but then you try to input the action, and then keep trying while doing the action slightly sooner until you see your character move.

It's easy to mess this up at first until you get the hang of it. I would suggest first using the 2 frame skip to skip to the second frame of that action and study every inch of the character, what position the limbs are in etc down to the minute details, then when you know exactly what frame two looks like, you'll recognise it if you accidentally skip forward too far. Of course, after a while of using frame skipping techniques for labbing you'll get to the point where you can just see how many frames have skipped.

Once you're on frame 1 of that action, you can access frames 3, 5, 7 etc by using the 2 frame skip trick.
To test that you've got this, go to training mode with Mario, stand in front of the opponent, hold L and attempt to let go of L on the first frame of Mario's Jab. It will require some practice, but if you get it right, Mario will clearly begin his Jab 1 animation but will not actually hit the opponent; if instead you mess it up he will hit them because Mario's Jab comes out on frame 2.

There is one more advanced version of the 1 frame skip, and while I don't want to side-track from the basics, I'll mention it anyway. Essentially, there may come a time when you want to pause the game on frame 1 of an action in the middle of a frame skipping sequence. What I mean by that is, instead of pausing on the first frame of the action before you start skipping forward by 2 or 3 frames, you may already be on a specific frame, say frame 15, and want to have a move be used on frame 17 so that you are paused on the first frame of that action. To do this, you'll want to have the speed set to 1/4 hold, hit L to skip forward 2 frames but then time your input so that it is used on the second frame of the two frame skip. This will require a bit of practice and a thorough understanding of what exactly it is that you are seeing (which comes with experience), but I can personally do this consistently, no trouble. (See also the more advanced labbing scenarios at the very bottom of this post for more things like this.)


The 3 Frame Skip.

The 3 frame skip will become one of your most used tools when dealing with larger numbers of frames. Used in conjunction with the frame 2 skip, you can access any frame other than 1 (e.g. 3 + 2 + 2 = 7). What's more, unlike the 1 frame skip, the 3 frame skip can be used during frame skip counting to change the paused frame from e.g. a positive to a negative number, while the 1 frame skip can only be used at the very start (e.g. you skip forward to the fourth frame of an action, hold an input which will register on frame 5 but then you skip forward by 3 frames, you'll then paused on the 7th frame and be able to make your next input register on frame 8).

The 3 frame skip is pretty simple to use; you just have to switch the speed to 1/2 (hold instead of 1/4 (hold). In the same way that '1/4 (hold)' lets you tap L as fast as you can, and no matter how fast you tap L it will always only skip 2 frames because 2 frames is the minimal amount it can move forward, if instead you use '1/2 (hold)' and tap L as fast as you can, the minimum amount of frames it can skip forward is 3 frames. In short, the 3 frame skip offers the same leniency and reliability that the 2 frame skip does. Credit to Masonomace Masonomace .


Using Shield to Indicate the First Actionable Frame (if you're on the ground).

Two important things about the shield make it excellent for this purpose (though obviously other things can be used too e.g. if you're in the air). First of all, you don't have to buffer shield in order for it to come out on the first actionable frame, you just have to be holding it. Second of all, the first frame of the shield animation doesn't actually visibly show the shield, while the second frame does. (Note that the first frame will still shield things even though it doesn't look like it will.) This will allow you to easily know what frame you were first able to act on by using the 2 frame skip method and noting whether you can see the shield or not.




"And now for something completely different."

Ok so this is a bit unconventional, but at this stage I think it would be best to continue teaching by way of actual examples and throw some tips in along the way.

I'll provide a description of what I'll be testing at the start, so if you're not interested or you get the idea already, you can feel free to move on to the next lab scenario. It would be a shame if you didn't get to the end of this post just because reading parts of it became tedious.



Lab Scenario 1.

Resources such as Aero's frame data site http://kuroganehammer.com/Smash4 are very helpful for testing purposes as will be shown in 'lab scenario 2'. However it can't cover everything and indeed some characters have less information than others. For example, many little things like throw frame data and item toss frame data aren't covered, and not every character has 'air time' frame data for SH and FH. So we're going to quickly get the SH air time frame data for Ganon [which has since been added XD; oh well].

Using Aero's data, you can see that Ganon's jump-squat is 7 frames, so this will be subtracted from the total amount of frames it takes to begin the SH and enter the landing animation.

Go to training mode 1/2 hold with Ganon. Hold the jump button and tap L to skip three frames. Continue tapping L to skip three frames all while keeping a track of what frame you're looking at (e.g. '3', '6', '9 ', etc), and do not get lazy with the tapping; it is very easy to not take each frame skip seriously when you have to do heaps of them in a row. So take your time and get it right the first time.
When it looks like Ganon is getting relatively close to the ground, switch from 1/2 (hold) over to 1/4 (hold) speed and start doing 2 frame skips instead. You'll get to counting (e.g. if you stop 3 frame skipping at 30) '31 32', '33 34', '35 36', '37 38' and you'll note that when you tapped L and counted '37 38' Ganon landed at some point. If you were watching closely you would have seen that he landed on frame 38, but even so you're going to want to double check.
This time, you'll want to do one more or one less 3 frame skip in order to swap the negatives and positives. E.g. continue 3 frame skipping till you get to 33 this time, then switch to 2 frame skips and count '34 35', '36 37' '38 39' and you'll note that Ganon landed when you said '38 39'.
By process of elimination this means that Ganon definitely lands on frame 38, and you can see that he is still in the air on frame 37. If you wanted to be super sure, you could even have the game paused on frame 37 and just hold the shield button for your next frame skip, and then you can just see whether or not you experience airdodge landing lag or not, which in this case you won't because the shield input will register on the first grounded frame. At this point it's a simple matter of taking away the 7 frames from the jump-squat and we find that Ganon's SH air time is 30 frames exactly.

If you don't have the jumpsquat frames or you want to double check it manually, it's a simple matter of having the game paused on the frame you believe to be the last jumpsquat frame and inputting upwards and attack for the next frame skip. If you get an U-air, then that was the last jumpsquat frame. If you get an U-smash there was still at least one more jumpsquat frame.


Lab Scenario 2.

You want to find out what frame advantage your character has when using certain moves to hit the opponent's shield.

Let's say you want to test what the frame advantage of an aerial is on an opponent's shield, and you want to test it under the best possible circumstances, i.e. you hit the opponent's shield on the last frame you were airborne.
If you were testing for Link, you could look up Aero's frame data and see that Link has 7 jump-squat frames, 33 frames of air time in a SH, and e.g. Nair hits on frame 7.
Using the power of maths, you can work out that in order for the frame 7 aerial to come out on the last frame you are airborne, you will have to input it such that it begins on frame 34 (jump-squat plus airtime is 40 frames total, on frame 41 you will land, you want it to hit on frame 40, therefore frame 1 of the 7 frame Nair must happen on frame 34, meaning we need to get the game paused on frame 33, hold attack, then tap L.)
In order to get the game paused looking at frame 33, with Link standing doing nothing, go to 1/2 (hold), hold jump and use the 3 frame skip trick eleven times, ending on '30', '33'. Make sure the cpu is set to 'control' and have them hold shield with the second controller. Then coming back to the first controller, just hold attack and skip forward to get the Nair to hit on frame 40 (it doesn't matter at this point so much that you do this with any accuracy, just so long as you pause when the move actually hits).

Here's where things get a bit complicated. The frame advantage will be different depending on what you got the opponent to do. If they powershield the aerial there will be reduced shieldstun; if they use an OoS option they won't have to worry about how long it takes them to drop shield; if they didn't powershield, as in the above example we are working with currently, and they do drop their shield it will sometimes even matter how long they held shield up before it got hit against very weak moves, though this isn't an issue for us right now. (If they have been holding shield up for not that long and they only barely missed the powershield, they can be locked in their shield for slightly longer before they can shield drop compared to if they had been shielding for a little while before it got hit; again only on very weak attacks though).

I'm not about to go through every single scenario and we're just going to stick with what we're doing already, so no power-shield. Let's say for now that you wanted to find out what the frame advantage was with Link's Nair hitting an opponent's normal shield with the opponent then dropping their shield as soon as it is hit to try to attack you with e.g. a Jab.

Coming back to the above example with the game paused with Link hitting the opponent's shield with Nair on his last airborne frame; switch to 1/4 (hold) so you can skip forward by 2 frames. Feel free to hold shield on the first controller at any point, but it only really needs to be held once Link lands. As for the opponent, let go of shield until you see it begin to drop, then hold shield again. At this point, with both characters on the ground hold shield, both without their shield up yet, slowly skip forwards by two frames at a time. In this example you should find that Link will have a 5 frame advantage, i.e. that he will act i.e. begin the shield 5 frames earlier than the opponent. Remember that you can work out what frame someone first acts on by looking at whether you can visibly see their shield or whether they are merely in their neutral stance without the shield visible following a 2 frame skip.



Lab Scenario 3.

Just quickly, because it is similar to the above stuff, if you wanted to test the frame advantage gained by 'frame synching' you could use the above methods to consistently ensure that e.g. Ganon uses Dair on the last frame he is airborne. We found before that he has 30 frames of air time in a SH and that he has 7 jump-squat frames. Aero's frame data tells us that Dair hits on frame 16, so out of a total of 37 frames including the jumpsquat frames and the time he is airborne, we would want to have the game paused on frame 21 and input Dair such that the first frame of Dair happens on frame 22 and the hitbox comes out on frame 37.
Note that in order to do the Dair without fast falling, you'd probably have to only partially tilt down on the joystick to avoid the fast fall or simply have held down since before hitting down would fast fall.

Note however that this exact method will not work for every single aerial of every character that can do a frame synch, because frame synchs work by having the character so close to the ground when they hit the opponent's hurtbox with the aerial that they actually interact with the ground and land during hitlag, and not all character's last airborne frame out of a SH will be close enough to the ground, which is why sometimes fast falling is necessary to alter your relative airborne position in each frame. Take for example Link's Fair. It is impossible to set up a frame synch by using the above method because on one frame Link's Fair won't come out in time before he lands, and on the frame before that he will still be too high up, and so it won't frame synch. There are many possible ways to get around this, even just for testing purposes, but the one I favour at the moment is to do the following:
Just go into training mode with Link in 1/4 hold and:
Hold jump, tap L (to frame skip by 2 frames), let go of jump, tap L, hold jump again, tap L, let go of jump, tap L (you would have successfully buffered a DJ out of the jump-squat), hold attack, hold L and watch Link do a rising Nair out of the DJ, then use the 10 frame buffer window as Link is coming back down to do a Fair and it frame synchs every time. This is because the buffered DJ adjusts Link's position in mid-air such that the last airborne frame is closer to the ground.

There is one more thing to note, specifically about testing frame synchs. It is possible that given a very specific set of circumstances, the 1/4 (hold) speed could affect the results. Out of all the many and varied frame synchs I've tested, only one has ever given me trouble because of this, and that is Falco's Dair frame synch. For a reason that is too complicated to explain here in a short space (PM me if you're genuinely curious), using 1/4 (hold) two frame skips, it is actually impossible to make Falco's Dair get a frame synch. By using the same inputs however (SH, first frame of Dair on the 8th airborne frame then FF on the 17th airborne frame), and switching to 1/2 (hold) 3 frame skips to move forwards from there instead, Falco will get the frame synch as he should. This is the only known example of 1/4 (hold) affecting in-game phenomenon, and I can assure you that the difference is insignificant enough such that it would not affect anything else ever.

(In general though, many things can be figured out quite simply by testing them in '1/4 and 1/2 hold' where you can take your time to input the actions and by doing so work out what the real inputs are. See my write-up on bombsliding in the Link AT thread for a good example.
I see a lot of questions about simple things e.g. what can be done out of a dash/initial-dash/run, or how many frames you have to wavebounce, or just simple questions about the mechanics of actions that could all be very easily tested using frame skips in training mode by anyone regardless of skill level, and it is for this reason that I decided to make this thread to encourage and empower people to find these things out for themselves.)



Lab Scenario 4.

Time for a change of pace. Let's talk about the choice of stage to test things on. Some stages will naturally aid in your testing.

For example, let's say that you want to accurately figure out which move has more horizontal range.
A lot of the time hitboxes don't align with the visual representation of the move. To test it, it's a simple matter of going to Luigi's Mansion and using the pillars from the same distance using different moves to see which one hits first while slowly adjusting your position. Note that some moves will move the character from their original position, so use those last. You use the pillars because they have a perfectly vertical hurtbox (regardless of what it visually looks like) that doesn't move, which gives you far greater accuracy than trying to do the same on other characters, either because they will move or because their hurtboxes won't be perfectly flat and vertical, such that when comparing moves which hit slightly higher or lower you won't get as accurate a result as you would have by using the pillars.

Pro tip: If you want to slowly move your character forwards with precision, pause the game with 1/2 (hold), slightly tilt the joystick forwards, tap L to skip forwards, let go of the joystick, then hold L to return to neutral. You will get the first 3 frames of the walk animation, which will move you forwards very slightly. Repeating this will make it so you always get to a point that the move you want to test will spark. Different characters will move different distances, and sometimes you'll find that it takes too long or perhaps you'll find that it's too quick, so you can adjust the amount of distance you move accordingly by either using 1/4 (hold) to move forwards 2 frames in the walking animation, or perhaps tap it twice for 4 frames, etc. Just know that by using this very precise movement method, the range that the move creates a spark will be a lot larger than you'd imagine, as it is possible to adjust your position multiple times and still get a spark before you finally move too close.

Or let's say you want to test something to do with comparing movement distance, e.g. slide distance. Well you might want to consider going to (omega) Wrecking Crew because the floor is separated into neat segments that you can use to visually measure the distance.

Yoshi's Island is great for testing slopes, while the upper middle of Temple is useful for testing ledges.

In order to properly test differences in vertical distance it is best to make a custom stage with a stack of thinly layered platforms which if done correctly can be accurately placed so that they are the exact same distance apart. In fact sometimes taking the time to make a custom stage for the specific purpose of running a test will be the only way to get accurate results, e.g. for accurately finding out the range and active area for tethers. I have a custom stage that simply has a long slightly tilted platform above a flat surface which has come in handy for testing moves hitting different areas of characters. Custom stages are amazing and will often be the only way to accurately test something, so familiarise yourself with it.



Lab Scenario 5.

So let's say you've used the pillars on Luigi's mansion to see how much range the hitbox has (regardless of what the move visually looks like), but now you want to know how disjointed that move is, e.g. how far away from your character's hurtbox the tip of the hitbox is during that move.

There are quite a few ways to test this. Some of them are bad though, but I'll write about them in the spoiler box below regardless; feel free to skip the chunk of text in the spoilers.
Bumper Method
The bumper method is where you set a bumper (the item) and slowly inch closer with your character till you get hit while doing the move, noting how close to the bumper you were able to get.
Here's some tips to make this method slightly easier and more consistent.

Go into training mode, make the cpu the character that you want to test and make player 1 Donkey Kong. Set the cpu to control. Get DK to spawn a bumper. Because he's so big you can pick up the bumper without needing to move (though even if you used a character who can't, you can get around this by simply spawning two bumpers). From now on, don't move player 1, because we want to be able to consistently place the bumper in the exact same spot so that when it disappears you're able to replace it perfectly.

Depending on what move you're testing you will need to be able to consistently place the bumper in different positions.
For more chest level attacks, you can simply throw the bumper forwards. (Note that when testing things like f-tilt it would be best to let go of L on player 1, input the action with player two (lightly tilting the joystick), then hit L again to avoid moving forwards slightly before doing the move.)
For attacks that are slightly lower, make DK stand on the top platform of BF and move him to the edge of it so that he starts teetering. Spawn a bumper, catch it, then throw the bumper down at the ground. It will set itself slightly lower than normal.
For attacks that are more ground-level, make DK stand on one of the side platforms and move him to the edge furthest away from the middle of the stage. Spawn a bumper, catch it, throw it down. It will set itself on the ground.
Testing aerials is a bit more complicated. It is easiest to simply stand on a platform then throw it out, and differently placed platforms will give different heights (so perhaps consider using a custom stage with stacked sloping platforms such that you will be able to adjust the height of DK and therefore the bumper with minute differences and still always throw it to roughly the same spot horizontally speaking). In order to accurately test things like U-air, you will need to use the frame skip trick to accurately use Uair on the exact same frame of your jump every time.


If you want an actual indicator to show where exactly the disjoint is however, you'll want to test using an actual hitbox that can cause a spark to appear on your hurtbox. This will be more difficult but also much more accurate.

What we'll want to avoid is having the hitboxes clash. There are two ways to do this. You can either time it with the frame skip trick so that the hitboxes aren't active at the same time, or you can simply use a move with transcendent priority.

Mirror Method
So for example, with the first method, if you were wanting to test the disjoint of e.g. Marth's D-tilt, you could look up Aero's data and find out that the hitboxes are active on frames 16 and 17, i.e. for two frames. Then it's a simple matter of going into training mode with two Marth's, have them stand in front of each other (but surprisingly far apart), get them both to crouch, input a D-tilt with one of them and tap L to frame skip ahead by two frames, then make the other do a D-tilt. Slowly inch closer to the other Marth while repeating this process until you see the spark.

The problem with this method however, while exceedingly simple, is that you're only showing where the hurtbox is on the frame after the last active hitbox, which isn't exactly what you're after.

Transcendent Method
You'll need a move that has transcendent priority. For example, you've got Little Mac's angled down F-smash, Fox's blaster, Pikachu's Jab, and Palutena's U-smash, each of which has their own advantages (though there are plenty of other moves which transcendent hitboxes too).

Little Mac's F-smash is great because it has super armor right up until the same frame that the hitbox comes out on so you can hit Mac on the same frame his hitbox is out without him going anywhere. You just need to calculate on what frame you should use each move.
Fox's blaster is also great because even if you miss-time/space it the opponent won't be going anywhere. If you set up a custom stage with platforms at thinly spaced intervals and/or on a slight tilt, you can get the blaster to hit anywhere.
Pikachu's Jab is great because it's so fast which means it's not much of a hassle to set up with frame skips to line it up with the move you want to test, and when you combine this with giving Pikachu a Super Star item which grants invincibility, this can be great because the opponent will experience hitlag and Pikachu won't go anywhere when hit. Try setting it up on a custom stage so that the Power Star always falls on Pikachu.
But these all pale in comparison to Palutena's U-smash, which is what you'll be using.

Select the character you want to test then select Palutena as the CPU in training mode. Go to a stage such as battlefield that allows the two characters to be on different levels with Palutena's head just beneath the hitbox you want to test. In this case, make Palutena go to the side platform and stand on the edge toward the middle facing the middle of the stage, then get your testing character on the top platform. Feel free to put the camera on zoom and get rid of the info bar. Now it's just a matter of positioning your test character so that they are at the perfect distance where they can e.g. Jab in the direction of Palutena's U-smash light and cause a spark to appear on their closest hurtbox. What you want to do is have the game paused on the frame the spark appears, which will typically be the first hitbox frame of that move. The hitbox on Palutena's U-smash lasts a long time (see Aero's website http://kuroganehammer.com/Smash4/Palutena) and it is a stable vertical hitbox for the duration; the point is, you don't need to be particularly well-timed to line Palutena's U-smash hitbox up with the testing character's hurtbox. Feel free to get a rough idea of where you need to stand with the help of the zoomed camera by making a few mistakes first and getting hit, then when you have a rough idea of where you need to be, inch forward by tilting the joystick and tapping L between tries (see the above 'pro tip' in lab scenario 4 for more detail).
When it comes to testing horizontal disjoint, this is the best way to do it. If you want to test vertical disjoint you should use Fox's blaster and a custom stage.



Lab Scenario 6.

You want to find out what frames of an action have invulnerability/invincibility/super-armor etc.

Fairly basic. Using the techniques of frame skipping outlined above to access any frame, use Little Mac's 1 frame Jab that is also only active for 1 frame to see if it stops them in their tracks on the following frame as you tap L, and then test for each frame.

If you're testing invulnerability frames and you want to get fancy/complicated, but also much much quicker, you can actually use a move with many active frames. In fact the more solid active frames, the better. Let's say that you want to test how many invulnerability frames someone's airdodge has. Select that character along with e.g. Toon. Have them both jump and have the opponent use airdodge while Toon uses Uair and position Toon so that the Uair will hit as soon as the invulnerability runs out. Make sure you count the amount of frames that the opponent has been airdodging. Then what you want to do is slowly skip forward by 2 frames until you see the Uair hit. This next part is important. With the game paused on that frame, being the first 2 frame skip you saw where the Uair actually connected and did damage, you will see one of two very distinct things, depending on the move used. This is because the first frame that a move hits looks very different to the frame after the move hits. The frame after will always have all kinds of visual effects with streaks/splashes of light etc, while the very first frame the move hits will always have little to no visual effects at all. Depending on your testing move of choice you will want to simply familiarise yourself with what both frames look like, and then you'll be good to go. You can use this visual confirmation in the same way that you can use shield as a way of showing which frame you can first act upon when grounded, so that by skipping forward by two frames, you'll be able to tell just by looking at it which frame the move hit on.

A general tip for testing this sort of thing: when testing e.g. to find out what frames a character is invulnerable, it would be wise to test it on a 2D stage like Duck Hunt (omega version also works) because that eliminates the possibility of confusing your results with missing the opponent simply because they were in the Z-axis.



Lab Scenario 7.

So you want to test if a move has a transcendent hitbox.

In training mode, make the cpu a Toon Link set to 'control', move the Toon a certain distance away such that a smash or tilt thrown boomerang will turn around just in front of the character you want to test, and use the move you want to test on the boomerang. If the move cancels the boomerang, it's not transcendent.



Lab Scenario 8.

Testing DI.

DI has to be input either before being hit or during the hitlag frames. It's easiest to simply make the second character taunt and then hold a direction for them (especially if you're testing outside of training mode), and then use the move on them. This is especially true for downwards DI so as to avoid crouch cancelling.



Lab Scenario 9.

You want to test the effects of rage or stale moves in general.

You'll need to enter a real match as training mode does not account for either of these. Just set it to an infinite time match, and for things like rage change the settings so that 'handicap' is set to 'on' so you can adjust the percents much easier.

You'll need to suicide each time with the character using the attacks if you want to refresh their stale moves completely and reset the test.

Keep in mind that moves are often not whole percents, even when fresh, and moves often have different hitboxes that do different amounts of damage depending on where and/or when you hit with the move. Check Aero's data to make sure you don't mix this up.

Especially keep in mind that in a real match there is a 1.05 x damage multiplier. So for example, if you have a move that does 20% in training mode (and is listed in Aero's data as doing 20%), it will do 21% in a real match when completely fresh. If a move does 10% it will do 10.5% in a real match, and two of them (assuming you killed yourself in between to refresh the move) will do 21%. 5% will do 5.25%, and 1% will do 1.05%. These decimal percents matter, or in other words, even though the percent counter only shows whole percents, the character will e.g die earlier if they have e.g. an extra 0.1%, even though the percent shown is exactly the same.

Note that some moves (e.g. Link's bombs) aren't affected by staleness or rage and aren't affected by the 1.05 damage multiplier, so they are useful for adjusting percents accurately in a real match.



Lab Scenario 10.

So you want to test things accurately in a real game just in general.

First of all, read through the lab scenario above for some tips. Obviously in a real game you won't have access to frame skips which is the real issue here. This can be helped to some degree by using the pause button in special smash set to slow.

If you pause the game in a real match and hold an input using the controler that paused the game, you won't be able to unpause. Fortunately there is a small window of time between hitting unpause and the game actually starting again for you to buffer held inputs such that they are inputted on the frame the game starts again. For more complicated inputs you might want to consider using a controler that didn't pause the game, as they are allowed to hold their inputs before the game is unpaused.
Note that unlike in training mode, the smash-stick does not work and will only give neutral A inputs if used out of a pause screen, while the attack-stick and special-stick work perfectly. (Normally it's the other way around.)

Perhaps the most important thing to understand about this whole scheme (which is why I'm giving it a separate paragraph), is that unlike in training mode, your actions will not start on the frame the game resumes. In training mode, if you pause the game in 1/4 hold or 1/2 hold, hold an action, then hit L to resume the game, the first frame of that action will occur on the first frame that the game resumes again. This is all turned on its head in a real match using the pause strat. In a real match when you pause the game, hold an action on another controler (using another controler just to make this point clear), then unpause the game, the first frame of that action will occur on the second frame that the game resumes again. The game needs a frame to register inputs when using the pause strat. This affects everything. So yes, it is perfectly possible to re-pause on the dead frame that the inputs were simply taken, which can look like you ended up just re-pausing without going anywhere to the untrained eye, and then if you unpause it again your character will do the action without any further need for an input, this time with the first frame of that action occurring on the first frame that the game resumes. Technically you can make separate inputs on every consecutive frame, but it would require you to be able to consistently re-pause the game with perfect timing, and there is no trick that I know of to make this easier, except for the following.

I mentioned going to special smash in slow mode. If you slow a character down, normally you can access frames that don't otherwise show up, which is useful in one sense, but for the purposes of labbing it typically isn't useful at all because you want to be seeing the frames of the animation that you actually deal with and you don't want to be confused by others. Slow mode isn't like this though; it will only let you pause on frames that we are familiar with, and being slower, it means that you are better able to unpause then re-pause on specific frames once you know the timing. It's certainly not perfect, but if you know what the frames look like (you can look at them beforehand in training mode) and you keep in mind that any further inputs will require a frame to register them before performing them on the frame after, with patience, practice and experience it is possible to perform complex frame perfect labbing tests in a real match. It will never be as guaranteed/consistent as frame skipping, but it is an option. I believe this could easily replace lab scenario 13 for testing simple frame perfect combos with rage/staleness.



Lab Scenario 11.

So you want to find out if something combos into something else.

Let me begin by saying that the combo counter in training mode is inaccurate and cannot be relied upon to determine whether something is a combo. It doesn't take into account things like landing lag and it is often wrong about whether the opponent can airdodge in time. You should always double check yourself by trying to buffer e.g. an airdodge or shield with the second controller. Obviously you'll be wanting to use 1/4 hold to slow things down and make duel-wielding controllers easier.
Under no circumstances whatsoever should the combo counter be used to determine whether something is a true combo, because it does not take into account DI either. This is also something you'll have to do manually.

I'll continue to use Link as an example because that's what I'm familiar with.
Let's say we're finding out between what percents Link can combo D-tilt into FH Fair. Obviously it will depend on the opponent's weight / fall speed / gravity, so always specify which character you tested on. The traditional default testing character is usually Mario for 'reasons'.

Note that you'll want to make sure that the move you're using doesn't have multiple hitboxes with different properties, otherwise you'd need to specify which hitbox you're hitting with as there could easily be different results for different hitbxes.

First up you'll want to become familiar with what point of the D-tilt you can act first upon, so use a D-tilt and skip forwards two or three frames at a time while holding shield to get an idea of what point in the animation you can act on.
Once you know this, use D-tilt on Mario, for now we won't be bothered with DI as we're only interested in what percent it combos, not true combos. Frame skip forward till you're almost at the point where you can act, then hold Jump while you continue to skip forwards. At the point where D-tilt can be interrupted, Link will automatically Jump on the first actionable frame assuming you didn't input it too soon. At any point during the jump-squat animation, hold the joystick forwards and hold attack then continue to frame skip forwards; as soon as Link leaves the ground he will automatically use Fair on the first actionable frame. Note that if you wanted a FH and not a SH, you would need to have held the jump button during the entire jump squat animation. Just before the Fair is about to hit, have Mario input e.g. a DJ or an airdodge. Did the Fair connect regardless? If so then that's a combo. Adjust the opponent's percent accordingly and find the range that it works between.

To test whether something is a true combo, you'd have to also try to DI the D-tilt in various different directions (using the taunt method described above if necessary). You will usually want to be DI-ing up and away to see if the opponent is hit out of reach of the next hit (Fair). If DI doesn't help, then that's a true combo.

Note that rage and staleness will affect the percent ranges. See below.



Lab Scenario 12.

Testing to see whether a hitbox has set-knockback.

Super easy.
In training mode, hit the opponent at 0% and note how far they get hit, then adjust their percentage by moving it left once so that it is at 999% and hit them with the hitbox again. If there is no difference between how far the opponent got hit, then the hitbox has set-knockback.



Lab Scenario 13.

So you want to test if something combos with rage or staleness, or related to that, how many hitstun frames a move induces if it has rage or staleness affecting it.

[see lab scenario 10 first]
Many moves are too weak or have set knockback, so their ability to combo will often be tied to whether or not the character using the move has rage because of the added knockback/hitstun. Or perhaps the move has too much knockback and you think that if staled it might open up new combos or simply extend pre-existing ones. But as seen above, the methods for testing whether something combos and/or finding out how many hitstun frames there are require you to be in training mode, and as you know, staleness and rage are nonexistent in training mode.

The way around this problem is to use equivalent knockback. Because hitstun is directly tied to knockback, all you need to do is find the exact amount of knockback that a rage-affected or staled move does in a real game and then simply recreate it inside training mode where you have access to frame skips. If the move you want to test has set knockback or you're testing staleness, you'll actually have to make use of a different move entirely; so long as it has the exact same knockback launch angle (see Aero's website for this information) and doesn't hit too far, nothing else matters. If the move hits the opponent more horizontally than vertically, it's a simple matter of setting them in a specific spot and measuring horizontal distance induced in a real match then adjusting the percents in training mode until you reach that exact same distance, but typically the moves you'll want to test will hit opponents more vertically, so here's how you want to test these ones:

Create a large custom stage, starting with a long completely flat and thin base for the stage itself, then add a few ever so slightly slanted platforms (the lower the slant, the more accurate and time-consuming your testing will be) that stretch across the width of the stage stacked upon each other such that one side of them is roughly the same horizontal height as the other side of another platform above or below it. Have these platforms at around the height that the move you want to test will knock the opponent up to. Then along the base of the main stage, just beneath the surface, add equally spaced angled platforms that slant the same way (the more the merrier, though you'll probably run out of available pieces) that you can see the tip of just beneath you when standing on the main base of the stage; they will act as a measuring device of sorts. I like to add a very small platform that crosses through every 5th platform to make it easier to know where I am exactly.

Now go into a real match on this custom stage with the characters you want to test and with the percents set properly if you want to test rage, otherwise you'll also want to first stale the move you want to use. Place the character that will get hit on the base of the stage in such a position such that using the measuring device below you, you can tell later with accuracy where exactly he was standing. Now hit that character with the rage-affected or staled move you want to test (making sure to connect with the correct hitbox with the launch angle you want to test). Did they get hit up through a platform and then land on it? If so, do the test again (making sure to reset the damage and the freshness/staleness back to where it was where necessary by suiciding) and this time move the character (that will be hit) further over to the higher end of the above platform that they landed on to make it more difficult for them to do so again; if they didn't land on the platform, move them further over to the lower end. Continue repeating this until you have worked out the exact spot where the character needs to stand in order to land on the platform if hit, but if moved even slightly over to the higher end they will not land on it (then take a note somewhere of where this was exactly using the measuring device as an indication e.g. it might be 5.2 platforms from the left).
With this knowledge, go to training mode and into that same custom stage. If you're testing rage and the move you tested doesn't have set knockback, then you'll want to place the character (that will be hit) in the exact spot that they were able to land on the above platform (e.g. 5.2), then adjust the percent until the same move is able to make them land on the platform in that same place, but the percent below will not be enough. If you're testing rage and using a move with set knockback, you will want to use a move with the exact same launch angle whether that means using the same character or not, then do the same; simply adjust the percent till that move makes them land on the platform but the percent below will not (again, being careful to hit with the correct hitbox that induces the correct launch angle).
With this knowledge, you can use this exact same move and exact percent in training mode to induce what will be equivalent knockback and hitstun to the rage-affected or staled version that was used in a real match. You can then use frame skips to test anything you want about the rage-affected or staled move.

This exact same test can be done in reverse of course. Let's say that you already know how much hitstun is required to combo. Go into training mode and first find the percent where the move will inflict the required amount of hitstun but the percent below will not. Then still in training mode, go to the custom stage with the exact percent and find the exact spot where that move at that percent will hit the opponent far enough to land on the platform but if you moved over slightly it wouldn't. Then go to a real match and find what staleness or rage requirements are needed to get that exact height standing in that exact spot. The point is, while rage and staleness do not exist in training mode, they can be recreated.



Lab Scenario 14.

Testing something that requires a third input to be made that is on the second frame of the 2 frame skip.

Now we're getting difficult. So let's say that you want to do an aerial on a certain frame. We'll say that we want to pause the game on frame 11 so that the first frame of the aerial comes out on frame 12. Well that's no big deal because you can just use the 3 and 2 frame skips to pause the game on frame 11. Very good. So now you're all set to use the aerial starting on frame 12, but then, what happens if you want to fast fall on frame 15? Also simple; just input the aerial and use a 3 frame skip to pause on frame 14. Ok, but what if you wanted to fast fall on frame 13 instead, i.e. on frame 2 of the aerial? If you just held down at this stage and frame skipped, the fast fall would occur on frame 12 i.e. frame 1 of the aerial, and if you skip forwards by 2 frames first you'll only be able to input the fast fall on frame 14.

It is actually perfectly possible to deliberately time your input so that it is done on the second frame of a two frame skip. Now this takes some practice, and it takes experience to actually see that you've done it correctly. You need to be able to actually see the two frames as separate and distinct frames, so that when you e.g. fast fall on the second frame of the 2 frame skip, you will know that you got it right, also you'll see a flash of light but won't visually move anywhere.



Lab Scenario 15.

Testing something that requires an input to be held for only 1 frame.

A good example of this is perfect pivoting. Perfect pivoting requires that you input the dash then input backwards on the joystick for one frame such that the joystick is returned to neutral on the following frame. Now obviously if you're skipping forwards by 2 frames at a time you cannot simply hold backwards and tap L because you would have held backwards for 2 frames.

There are two ways to do this. Either you can time your input so that you hit backwards on the second frame of a 2 frame skip, i.e. my preferred method as it allows you to do any input you desire on the following frame, or, you can hold backwards before tapping L to do a 2 frame skip and then simply time it so that you let go of the joystick before the second frame. Both are perfectly possible with practice and you'll know the difference is you mess it up.



Lab Scenario 16.

Testing something that requires 3 or more separate inputs back to back on separate consecutive frames, or just two inputs back to back within the same 2 frame skip.

Similarly difficult to the lab scenarios above. I had to do this when testing some of the bombslides for Link, so I'll use that as the example.
The input required e.g. for the 'bombslide soft throw' is to do a dash throw by dashing then hitting and holding the attack button, then on the next frame (i.e. frame 1 of the dash throw) lightly tilt the joystick backwards and hit the special button (with attack still held), then on the next frame hit up on the joystick.

In order to do this, I used the 1 frame skip to get the first frame of the dash throw, then I held the joystick in place plus special (plus attack still), and as I tapped L to frame skip 2 frames I moved the joystick upwards so that I inputted the joystick upwards on the second frame of the frame skip.
In this case it was super obvious whether I got it correct or not because it meant the difference between getting completely different kinds of bombslides.

This won't come up a lot, but it is certainly possible to do two separate actions on each frame of the frame skip with a bit of practice.






Much of what has been said can be adapted to almost anything you want to test with a bit of ingenuity, and that's something I can't teach.

I'll add more as I think of it (It's kind of late for me atm), but at the same time I'm trying to avoid overlapping on pointers and ideas. Having said that, I would welcome input on any testing and labbing tips and tricks that I've failed to mention and I'll update the OP with them when I come back online.
 
Last edited:

Dr. Tuen

Smash Lord
Joined
Apr 26, 2009
Messages
1,396
3DS FC
0559-7294-8323
This is awesome! I was just trying to figure out how to run the test I outlined in the Mechanics and Techniques discussion, since I am by myself. I didn't realize that the frame skip was so accurate in that mode. I'll have to run some of your scenarios to get used to it. Thanks for the guide!
 

Browny

Smash Hater
Joined
Mar 22, 2008
Messages
10,416
Location
Video Games
I think its important to add a section on investigating the size of disjoints as well, since this is something commonly brought up in matchup discussions, and also commonly very wrong.

When trying to find the disjoint on an attack, the first thing to do is find the extent of your characters hurtbox. Do this by having another character with a disjointed attack (ROBs fsmash is probably the best in the game, any disjointed attack works though. Sword attacks are the best for testing aerials) hit the character being tested at the same frame so they spark. This will show the maximum hurtbox extension, the beginning of the disjoint. This will only work if the attack being tested has less of a disjoint than the move hitting them, thats why robs fsmash is one of the best. For testing low-hitting moves though you need something that hits low and marths fsmash is the best for this.

http://i.imgur.com/9pHb1bm.png

You will have to move the characters forward tiny amounts with each test until the moves spark by coming out on almost the same frame at the perfect distance. Its very tricky, but its the only way to do it without getting into the game code.

From there, its very easy to find the maximum range of the attack by looking for the spark.

Maximum range - maximum hurtbox extension = total disjoint

http://i.imgur.com/MDdsZqc.jpg

I did this for marths dtilt too for another example
http://i.imgur.com/uIFQtkV.png
http://i.imgur.com/5jJogJT.jpg
 
Last edited:

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
I think its important to add a section on investigating the size of disjoints as well, since this is something commonly brought up in matchup discussions, and also commonly very wrong.

When trying to find the disjoint on an attack, the first thing to do is find the extent of your characters hurtbox. Do this by having another character with a disjointed attack (ROBs fsmash is probably the best in the game, any disjointed attack works though. Sword attacks are the best for testing aerials) hit the character being tested at the same frame so they spark. This will show the maximum hurtbox extension, the beginning of the disjoint. This will only work if the attack being tested has less of a disjoint than the move hitting them, thats why robs fsmash is one of the best. For testing low-hitting moves though you need something that hits low and marths fsmash is the best for this.

http://i.imgur.com/9pHb1bm.png

You will have to move the characters forward tiny amounts with each test until the moves spark by coming out on almost the same frame at the perfect distance. Its very tricky, but its the only way to do it without getting into the game code.

From there, its very easy to find the maximum range of the attack by looking for the spark.

Maximum range - maximum hurtbox extension = total disjoint

http://i.imgur.com/MDdsZqc.jpg

I did this for marths dtilt too for another example
http://i.imgur.com/uIFQtkV.png
http://i.imgur.com/5jJogJT.jpg
Thanks for reminding me. There are a few ways that you can test disjoints. I've added them to the OP. I felt it was better coming straight after lab scenario 4 so I've made it lab scenario 5 and then moved everything else down one place.
 

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
Lab Scenario #4 using Luigi's Mansion's pillars to test horizontal range is brilliant. I used a custom stage & spawned a Crate from a distance away & back rolled to the very end of my custom stage's ledge. That's how I tested it since Crates for the most part are perfectly square & flat sided, so it really helped me find out just which of Shulk's moves had better range than others.

Anyway I'll definitely try out Luigi's Mansion's pillars.:shades:
 
Last edited:

Locke 06

Sayonara, bye bye~
Joined
Oct 2, 2014
Messages
2,725
Location
Grad School
NNID
tl.206
So, this should be in the academy. 100%. Incredibly useful and it shouldn't just get buried.
 

Yikarur

Smash Master
Joined
May 29, 2007
Messages
4,595
Location
Germany
This is amazing. I've used most methods myself already but this is a really amazing guide. I'm impressed.
 

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
So that lab scenario involving the Luigi's Mansion pillars is radical. I dig that. One question though, is the structure of either pillar completely straight from the base to the ceiling? It mostly looks convincing except when you look at the center of the pillar where it compacts near the white ring.

Anywho, may I get some confirmations or de-confirmations on these values I was labbing using the frame skip method?:

Shulk's frame values
Inputs & the method used to find these frame values

Dash to Neutral
(Input a dash, tap L, reset joystick to neutral, tap L a couple more times, then hold shield to see when you can act)
Note: Even if you reset the joystick to neutral after the first L tap, holding shield before tapping L the second time will still input a forward roll. That's why you tap L a couple more times before holding shield.


Dash to Shield (Ready a dash input by holding the joystick, tap L a few times, then hold shield while frame skipping to see when you can act)
Note: Holding shield too early during the dash will input a forward roll. That's why you tap L a few times.


Skid
(Hold L, input a dash, & hold the joystick to run. After you see your running animation, release L & reset joystick to neutral. Tap L once to frame skip & then ready holding shield to see when you can act after the skid)
Note: You have to reset the joystick only after your Dashing to Shield frame data. So if Shulk's Dash to Shield's Total Frames is 10 aka FAF being 11, then you have to frame skip until you see frame 11 or a frame beyond that. So for this you can use the 1-frame or 2-frame skip method & you still have to tap L six times regardless.


Dash Turn to Neutral (Different ways to do it since you can turn during the dash or during running. But hold L, input a dash, & hold the joystick to run. After you see your running animation, only release L. Then hold the joystick to the opposite direction to input the dash turn & tap L. Reset the joystick to neutral & tap L one more time. Hold shield & frame skip to see when you can act)
Note: There's no frame difference between doing a Dash Turn during the dash or run, so it works either way. For Shulk you can use the 2-frame skip method & tap L 3 times to end on the 6th frame. This is the earliest time you're able to Dash Turn during the initial dash sequence. If you input a turn earlier than frame 7, then Shulk will dash the other direction resulting in a dash dance. Doing the 1-frame skip method & tapping L two more times to end on the 5th frame & then holding the opposite direction to try turning on frame 6 will result with dash dancing.


Dash Turn to Shield (Different ways to do it since you can turn during the dash or during running. But hold L, input a dash, & hold the joystick to run. After you see your running animation, only release L. The hold the joystick to the opposite direction to ready the dash turn input & tap L. Keep holding the joystick & tap L once more. Then hold shield & frame skip to see when you can act)
Note: See the Note for Dash Turn to Neutral.


Tip #1: To know the FAF, just add +1 frame to the amount of Total Frames.
Tip #2: Jump art, Decisive Jump art, & Hyper Jump art have the same movement frame values as Vanilla.



Numbers
Dash to Neutral (Every Monado, Decisive, & Hyper art and Vanilla) | Total Frames=33
Dash to Shield (Every Monado, Decisive, & Hyper art and Vanilla) | Total Frames=10

Vanilla skid | Total Frames=24
Jump art skid | Total Frames=24
Speed art skid | Total Frames=28
Shield art skid | Total Frames=14
Decisive Speed art skid | Total Frames=28
Decisive Shield art skid | Total Frames=14
Hyper Speed art skid | Total Frames=27
Hyper Shield art skid | Total Frames=11


Vanilla Dash Turn to Neutral | Total Frames=35
Jump art Dash Turn to Neutral | Total Frames=35
Speed art Dash Turn to Neutral | Total Frames=39
Shield art Dash Turn to Neutral | Total Frames=25
Decisive Speed art Dash Turn to Neutral | Total Frames=39
Decisive Shield art Dash Turn to Neutral | Total Frames=25
Hyper Speed art Dash Turn to Neutral | Total Frames=39
Hyper Shield art Dash Turn to Neutral | Total Frames=21


Vanilla Dash Turn to Shield | Total Frames=34
Jump art Dash Turn to Shield | Total Frames=34
Speed art Dash Turn to Shield | Total Frames=35
Shield art Dash Turn to Shield | Total Frames=35
Decisive Speed art Dash Turn to Shield | Total Frames=35
Decisive Shield art Dash Turn to Shield | Total Frames=35
Hyper Speed art Dash Turn to Shield | Total Frames=35
Hyper Shield art Dash Turn to Shield | Total Frames=34
═════════════════════════════════════════════════════════════════════════════════════
If this is all correct, then I'd be fine with finding this kind of frame data for every character. But only if these numbers are correct.
 
Last edited:

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
So that lab scenario involving the Luigi's Mansion pillars is radical. I dig that. One question though, is the structure of either pillar completely straight from the base to the ceiling? It mostly looks convincing except when you look at the center of the pillar where it compacts near the white ring.
Yeah the pillars look like they change their hurtboxes, but in my experience they are actually completely flat and vertical. For example, if I move Toon to the right spot on the ground such that his Nair will barely miss and cause the 'spark' (which I believe is something close to if not actually pixel perfect) if done out of a standard vertical jump, I can use Nair on any airborne frame of a SH/FH and always get that same spark. So yes, I believe that regardless of the shape of the pillars, they are in fact perfectly vertical and flat hurtboxes.

Anywho, may I get some confirmations or de-confirmations on these values I was labbing using the frame skip method?:

Shulk's frame values
Inputs & the method used to find these frame values

Dash to Neutral
(Input a dash, tap L, reset joystick to neutral, tap L a couple more times, then hold shield to see when you can act)
Note: Even if you reset the joystick to neutral after the first L tap, holding shield before tapping L the second time will still input a forward roll. That's why you tap L a couple more times before holding shield.


Dash to Shield (Ready a dash input by holding the joystick, tap L a few times, then hold shield while frame skipping to see when you can act)
Note: Holding shield too early during the dash will input a forward roll. That's why you tap L a few times.


Skid
(Hold L, input a dash, & hold the joystick to run. After you see your running animation, release L & reset joystick to neutral. Tap L once to frame skip & then ready holding shield to see when you can act after the skid)
Note: You have to reset the joystick only after your Dashing to Shield frame data. So if Shulk's Dash to Shield's Total Frames is 10 aka FAF being 11, then you have to frame skip until you see frame 11 or a frame beyond that. So for this you can use the 1-frame or 2-frame skip method & you still have to tap L six times regardless.


Dash Turn to Neutral (Different ways to do it since you can turn during the dash or during running. But hold L, input a dash, & hold the joystick to run. After you see your running animation, only release L. Then hold the joystick to the opposite direction to input the dash turn & tap L. Reset the joystick to neutral & tap L one more time. Hold shield & frame skip to see when you can act)
Note: There's no frame difference between doing a Dash Turn during the dash or run, so it works either way. For Shulk you can use the 2-frame skip method & tap L 3 times to end on the 6th frame. This is the earliest time you're able to Dash Turn during the initial dash sequence. If you input a turn earlier than frame 7, then Shulk will dash the other direction resulting in a dash dance. Doing the 1-frame skip method & tapping L two more times to end on the 5th frame & then holding the opposite direction to try turning on frame 6 will result with dash dancing.


Dash Turn to Shield (Different ways to do it since you can turn during the dash or during running. But hold L, input a dash, & hold the joystick to run. After you see your running animation, only release L. The hold the joystick to the opposite direction to ready the dash turn input & tap L. Keep holding the joystick & tap L once more. Then hold shield & frame skip to see when you can act)
Note: See the Note for Dash Turn to Neutral.


Tip #1: To know the FAF, just add +1 frame to the amount of Total Frames.
Tip #2: Jump art, Decisive Jump art, & Hyper Jump art have the same movement frame values as Vanilla.



Numbers
Dash to Neutral (Every Monado, Decisive, & Hyper art and Vanilla) | Total Frames=33
Dash to Shield (Every Monado, Decisive, & Hyper art and Vanilla) | Total Frames=10

Vanilla skid | Total Frames=24
Jump art skid | Total Frames=24
Speed art skid | Total Frames=28
Shield art skid | Total Frames=14
Decisive Speed art skid | Total Frames=28
Decisive Shield art skid | Total Frames=14
Hyper Speed art skid | Total Frames=27
Hyper Shield art skid | Total Frames=11


Vanilla Dash Turn to Neutral | Total Frames=35
Jump art Dash Turn to Neutral | Total Frames=35
Speed art Dash Turn to Neutral | Total Frames=39
Shield art Dash Turn to Neutral | Total Frames=25
Decisive Speed art Dash Turn to Neutral | Total Frames=39
Decisive Shield art Dash Turn to Neutral | Total Frames=25
Hyper Speed art Dash Turn to Neutral | Total Frames=39
Hyper Shield art Dash Turn to Neutral | Total Frames=21


Vanilla Dash Turn to Shield | Total Frames=34
Jump art Dash Turn to Shield | Total Frames=34
Speed art Dash Turn to Shield | Total Frames=35
Shield art Dash Turn to Shield | Total Frames=35
Decisive Speed art Dash Turn to Shield | Total Frames=35
Decisive Shield art Dash Turn to Shield | Total Frames=35
Hyper Speed art Dash Turn to Shield | Total Frames=35
Hyper Shield art Dash Turn to Shield | Total Frames=34
═════════════════════════════════════════════════════════════════════════════════════
If this is all correct, then I'd be fine with finding this kind of frame data for every character. But only if these numbers are correct.
Let's see here.
So far so good.

There's no need to over-complicate the testing of things such as 'skid' by starting on the exact right frame, potentially having to use 1 frames skips etc. You can just make the character enter a run while holding L, then when you're more than sure that the run has started, continue holding forwards and let go of L, then the rest is as you say. In any case, you're not wrong.

I'm seeing a problem with your method for getting Dash Turn to Neutral and Shield. The method you mention involves immediately returning the joystick to neutral for the remainder or continuing to hold the joystick after the dash turn input respectively. The problem with both of these methods is that they are the worst possible way to do it quickly. The numbers you got are in fact correct, but because there is a better way to do it, the numbers you got are also redundant. What you want to do is continue to hold the joystick during the dash turn at first as it will reduce the amount of time it takes to stop, and then as soon as you have stopped you should return the joystick to neutral instead of starting out in the other direction. The tricky part for you then will be finding the best possible amount of frames to hold the joystick during the dash turn, which would probably vary depending on the MArt/character, and getting 1-frame skips for the dash turn may require some practice. The good news is, you'll be able to combine 'dash turn to neutral' and 'dash turn to shield' into simply 'dash turn', as they will now be the exact same thing. To give you an idea though, using Vanilla Shulk, if you enter a run, then hold backwards for exactly 20 frames, then return the joystick to neutral, you can shield or jab starting on the 21st frame. Yes, it's potentially that quick.

So yeah, you'll want to review your Dash Turn numbers. But I'll check what you have for dash to neutral/shield and skid now.

The numbers you got for dash to neutral/shield were correct for vanilla, so to save time I'll assume you got the same for everything else as you obviously know what you're doing.
And the skid frames are also correct.
 

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
Yeah the pillars look like they change their hurtboxes, but in my experience they are actually completely flat and vertical. For example, if I move Toon to the right spot on the ground such that his Nair will barely miss and cause the 'spark' (which I believe is something close to if not actually pixel perfect) if done out of a standard vertical jump, I can use Nair on any airborne frame of a SH/FH and always get that same spark. So yes, I believe that regardless of the shape of the pillars, they are in fact perfectly vertical and flat hurtboxes.
Nice, nice. I may convert to using the Luigi's Mansion stage for hitbox range testing entirely then.:shades:
Let's see here.
So far so good.

There's no need to over-complicate the testing of things such as 'skid' by starting on the exact right frame, potentially having to use 1 frames skips etc. You can just make the character enter a run while holding L, then when you're more than sure that the run has started, continue holding forwards and let go of L, then the rest is as you say. In any case, you're not wrong.

I'm seeing a problem with your method for getting Dash Turn to Neutral and Shield. The method you mention involves immediately returning the joystick to neutral for the remainder or continuing to hold the joystick after the dash turn input respectively. The problem with both of these methods is that they are the worst possible way to do it quickly. The numbers you got are in fact correct, but because there is a better way to do it, the numbers you got are also redundant. What you want to do is continue to hold the joystick during the dash turn at first as it will reduce the amount of time it takes to stop, and then as soon as you have stopped you should return the joystick to neutral instead of starting out in the other direction. The tricky part for you then will be finding the best possible amount of frames to hold the joystick during the dash turn, which would probably vary depending on the MArt/character, and getting 1-frame skips for the dash turn may require some practice. The good news is, you'll be able to combine 'dash turn to neutral' and 'dash turn to shield' into simply 'dash turn', as they will now be the exact same thing. To give you an idea though, using Vanilla Shulk, if you enter a run, then hold backwards for exactly 20 frames, then return the joystick to neutral, you can shield or jab starting on the 21st frame. Yes, it's potentially that quick.

So yeah, you'll want to review your Dash Turn numbers. But I'll check what you have for dash to neutral/shield and skid now.

The numbers you got for dash to neutral/shield were correct for vanilla, so to save time I'll assume you got the same for everything else as you obviously know what you're doing.
And the skid frames are also correct.
The notes I made for the following inputs do make it complicated, especially for skid. My intentions was to explain the whys & hows, especially with skid in that you can only skid when you're already running. I brought information into it, such as the Dash to Shield data for a reference as to why you could skid on so-and-so frame, but not on this-or-that frame. In the end, a simple, "just start running & let go of the joystick" would of been better, but I felt like adding in the trivial information so that one would know when a character's running actually starts frame-wise, or something like that.

The Dash Turn to Neutral & Shield inputs were performed that way simply because I wanted to do inputs identical to Dashing to Neutral & Shield, but with adding in Dash Turn to the mix of course. I didn't really have a clear objective with performing those inputs other than to figure out more about my character's frame data with certain movement inputs such as these, since Speed & Shield affecting Shulk's movement definitely made me curious of just what the differences would of been. However, in my case with testing the Dash Turn to Shield input for Shulk, the frames for Vanilla Jump & Hyper Shield were identical & yet everything else is 1 frame later. I found that really odd & uninspiring despite factors such as the Speed arts traveling more distance. Although, Dash Turn to Neutral was plausible to test for me in this case because there's a clear frame difference between the arts & Vanilla Shulk, which made it more satisfying to lab. A huge frame difference between Vanilla & Shield for example was a mighty reduction in comparison.

Thank you for the suggestion of finding just the Dash Turn's data itself. I can see that being more helpful to know than Dash Turn to whatever. And thanks for confirmation on the Dash to Neutral/Shield & skid frames.:)

Oh, & here's some numbers for Shulk's Dash Turn using the suggestion idea you gave:

Vanilla | Total Frames=20
Jump art | Total Frames=20
Speed art | Total Frames=21
Shield art | Total Frames=21
Decisive Speed art | Total Frames=21
Decisive Shield art | Total Frames=21
Hyper Speed art | Total Frames=21
Hyper Shield art | Total Frames=20

I liked the Dash Turn to Neutral numbers better, 'cus the frame data for the Shield arts is real.:shades:
 
Last edited:

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
Oh, & here's some numbers for Shulk's Dash Turn using the suggestion idea you gave:

Vanilla | Total Frames=20
Jump art | Total Frames=20
Speed art | Total Frames=21
Shield art | Total Frames=21
Decisive Speed art | Total Frames=21
Decisive Shield art | Total Frames=21
Hyper Speed art | Total Frames=21
Hyper Shield art | Total Frames=20
Yep that's all correct. But you already knew that XD.
 

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
Yep that's all correct. But you already knew that XD.
Awesome then. I'll probably do frame values for the rest of the cast after Sunday since that's a tournament day in case that I'm attending it.

And honestly, I really don't know how your thread isn't in the WiiU / 3DS Smash Academy yet. I feel like I have to tag a Moderator or suggest it via PM because it deserves to be there.
 

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
Because we've discussed about this, I'm gonna help add in an additional frame skip method. (You can edit this information the way you see fit, Detective.)

The 3 Frame Skip.

To perform this frame skip method, go to training mode & set your speed to '1/2 (Hold L)'. Once you've done that, your screen will pause until which you can skip forward exactly three frames by hitting the 'L' button & quickly letting go of it.

Keep in mind that you want to be literally hitting the shoulder button in a manner that you'll be tapping & releasing the button almost instantly. This way, you're being as precise with the execution as intended to be moving the minimal three frames forward & like the 2 frame skip, it's not difficult to be performing this skip method consistently once you get the gist of it.

So with this method, you'll see frames that are more easily accessible that are also in the mix whereas before you'd need the 1 frame skip method to see (example: you can see frames like 3, 9, 15, & 21 more quickly just from using this method by itself).
 

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
Because we've discussed about this, I'm gonna help add in an additional frame skip method. (You can edit this information the way you see fit, Detective.)

The 3 Frame Skip.

To perform this frame skip method, go to training mode & set your speed to '1/2 (Hold L)'. Once you've done that, your screen will pause until which you can skip forward exactly three frames by hitting the 'L' button & quickly letting go of it.

Keep in mind that you want to be literally hitting the shoulder button in a manner that you'll be tapping & releasing the button almost instantly. This way, you're being as precise with the execution as intended to be moving the minimal three frames forward & like the 2 frame skip, it's not difficult to be performing this skip method consistently once you get the gist of it.

So with this method, you'll see frames that are more easily accessible that are also in the mix whereas before you'd need the 1 frame skip method to see (example: you can see frames like 3, 9, 15, & 21 more quickly just from using this method by itself).
Oh thanks. Yeah I already had it written out while I was waiting for you to get back so I'd be ready to update the OP XD, so I just went with what I had. Thanks anyway.
 

LancerStaff

Smash Hero
Joined
Jan 28, 2014
Messages
8,118
Location
Buried under 990+ weapons
3DS FC
1504-5709-4054
The 3 frame skip isn't supposed to be consistent, is it? On 3DS I'm advancing two frames.

Found a way to advance one frame just now, not that it's very useful... On 3DS at least hitting reset has the game advance one frame for a moment, before reseting obviously.
 

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
Oh thanks. Yeah I already had it written out while I was waiting for you to get back so I'd be ready to update the OP XD, so I just went with what I had. Thanks anyway.
I was thinking of adding in the fun fact in there, which was that when you release the L button as you input an action in 1/2 (Hold L) speed, you'll skip 2 frames forward. It's similar to the 1 frame skip but we have a much easier & accessible way of skipping two frames.:shades:
 
Last edited:

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
The 3 frame skip isn't supposed to be consistent, is it? On 3DS I'm advancing two frames.
Yeah, the 3 frame skip (using 1/2 hold) is consistent, at least on wii-u. I've never been able to skip just 2 frames using it. It's to the point where I can see the animation continue even though I've already completely released L and then it always stops on the third frame. I don't have a 3ds to see if it's just that, but this is sounding more like something we should figure out in PM's.

Found a way to advance one frame just now, not that it's very useful... On 3DS at least hitting reset has the game advance one frame for a moment, before reseting obviously.
There's already a proper 1 frame skip in the OP though.

I was thinking of adding in the fun fact in there, which was that when you release the L button as you input an action in 1/2 (Hold L) speed, you'll skip 2 frames forward. It's similar to the 1 frame skip but we have a much easier & accessible way of skipping two frames.:shades:
Actually this isn't true. I'm able to 1 frame-skip consistently using 1/2 hold.
 

Yikarur

Smash Master
Joined
May 29, 2007
Messages
4,595
Location
Germany
You don't really need the 1 skip method anymore except if you need the very first frame because with alternating between 2 frame skip (1/4) and 3 frame skipp (1/2) you can get any frame you want.
 

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
You don't really need the 1 skip method anymore except if you need the very first frame because with alternating between 2 frame skip (1/4) and 3 frame skipp (1/2) you can get any frame you want.
True. I'll have to remember to go through the whole OP later and make edits where necessary to reflect this.


Edit: OP fully updated. I added quite a bit of content and re-wrote quite a bit because since writing this my labbing techniques have continued to develop, and that's not to mention how the 3 frame skip has improved the frame skip method just generally, so I've had to re-write everything to involve it being properly utilised.
 
Last edited:

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
So I don't know how helpful this is, but using any speed (preferably 1.5x speed if you want to be quick about it) & having the pause menu up while falling on springs causes characters to freeze in their jumpsquat animation indefinitely. It's until you unpause or set it towards 1/4x (Hold L) or 1/2x (Hold L) & tap L that they start to move again. Like I said, dunno if this is helpful but it's something to do with springs I guess. Helps sync up jumps & it allows you to clearly see the difference in their air times or fall speed due to gravity taking place early n stuff like that.
 
Last edited:

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
Could another lab scenario go into the OP, Detective? A lab scenario like finding the amount of shield damage to lead into counting shield hitlag and or shieldstun frames.

EDIT: Ignore this. Forgot that we already have formulas to find these things.
 
Last edited:

Big-Cat

Challenge accepted.
Joined
Jul 24, 2007
Messages
16,176
Location
Lousiana
NNID
KumaOso
3DS FC
1590-4853-0104
Nice one, Foxy.

I still wish this game had a recording feature. Maybe one can be modded in one day.
 

Sonicninja115

Experiment. Innovate. Improve.
Joined
Jul 8, 2015
Messages
2,429
How would one go about testing forwards and Backwards DI by themselves? Up and down are pretty simple, but the others not so much. Is there a custom stage configuration? Another method?
 
Last edited:

Stryker95

Smash Journeyman
Joined
Aug 11, 2015
Messages
252
Location
Texas
How would one go about testing forwards and Backwards DI by themselves? Up and down are pretty simple, but the others not so much. Is there a custom stage configuration? Another method?
Lab Scenario 8.

Testing DI.

DI has to be input either before being hit or during the hitlag frames. It's easiest to simply make the second character taunt and then hold a direction for them (especially if you're testing outside of training mode), and then use the move on them. This is especially true for downwards DI so as to avoid crouch cancelling.
Same way.
 

Druncle_Titus

Smash Rookie
Joined
May 14, 2016
Messages
12
Location
Simsbury, CT
NNID
Sean.Dunn17
I want to test the kill percents of different kill moves with rage, so how would i get the characters to the right percents to test?
 

Stryker95

Smash Journeyman
Joined
Aug 11, 2015
Messages
252
Location
Texas
I want to test the kill percents of different kill moves with rage, so how would i get the characters to the right percents to test?
I know I'm not Foxy but maybe I can help. To test killing with rage you obviously need to go to Smash mode (go to slow mode if it helps, I usually do). (Also, there is a way to do it in training mode but I am not comfortable with it yet so it's best if Foxy explains it when he adds it to the guide.)

Anyhow, Link's bombs always deal 5% damage on a direct hit (except on Bowser Jr. due to his cart, but there is a pattern) and doesn't have the 1.05 freshness issue so that can get you close. Otherwise there may be ways to get exact, like Foxy's training mode method and I have heard of another method.

Use Falco's Burst Blaster and hit the opponent 9 times (then kill the opponent). The blaster now deals 1.05% up close and 0.52% from far. Thanks to @GetShulked for that method.

Hope this helps.
 

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
I want to test the kill percents of different kill moves with rage, so how would i get the characters to the right percents to test?
Lab Scenario 9 mentions putting the handicap setting on in a real match to help adjust the percents and it also goes into the 1.05 freshness multiplier. Then if you want to adjust the percents to something more specific beyond what handicap adjustments permit, as Styker said, you can use different moves to accurately adjust it.
 

Masonomace

Yeah Shulk, get stuck in!
Joined
Apr 10, 2010
Messages
4,622
Location
Independence, MO
NNID
Masonomace
Stryker95 Stryker95 Fox Is Openly Deceptive Fox Is Openly Deceptive Druncle_Titus Druncle_Titus I too used that Falco Burst Blaster strategy back in early 2015 to test Shulk's relative kill percents on No DI Mario from center stage & at the ledge when Shulk players collaborated data. 1.05% is pretty handy & still accurate for the most part. If anything it was a marginal difference of 1% at times, but it's pretty good at the time. Others used weak pummels like Lucario which is still good.
 
Last edited:

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
I probably put this off too long, but I've added a new lab scenario, number 12, with everything below it moving up a number. It talks about getting equivalent rage/staleness in training mode.
 

Fox Is Openly Deceptive

Smash Detective
Administrator
BRoomer
Joined
Sep 22, 2008
Messages
3,969
Location
WinMelee, Australia
It's been over a year since the last update and here we are with a new lab scenario, slotted in at number 10, this one covering labbing things accurately in a real match (because I was forced to understand it better myself recently).
 
Top Bottom