A way to obtain frame data

[ibmutt]

CONTENTS
1.0 Introduction
2.0 Definitions

2.1 Frame
2.2 Rendered Images
2.3 Pose
2.4 Animation
2.5 Action​

3.0 The Basic Method

3.1 Example 1.a
3.2 Example 1.b
3.3 Summary
3.4 Animation Interruptions​

4.0 Problems with the Basic Method

4.1 Finding an Odd pose

4.1.1 Approach One
4.1.2 Approach Two
4.1.3 Approach Three​

4.2 Finding the End of an action

4.2.1 Example 1.d
4.2.2 Jumping
4.2.3 Buffering
4.2.4 Back to the Example 1.d​

4.3 IASA Frames
4.4 Approach Five
4.5 What if I can’t control the character?

4.5.1 Finding the start of an action
4.5.2 Finding the IASA​

4.6 Summary​

5.0 How to Obtain frame data yourself

5.1 What you’ll be working with
5.2 Where do you get the videos from?

5.2.1 I have a Capture Device
5.2.2 I don't have a capture device​

5.3 How to obtain frame data

5.3.1 Video Editors
5.3.2 Recommended Video Editors​

6.0 Downloads

7.0 Methods and Approaches

7.1 The Basic Method
7.2 Problems with Step 1 of the Basic Method

7.2.1 Approach One
7.2.2 Approach Two
7.2.3 Approach Three​

7.3 Problems with Step Two

7.3.1 Approach Five​

8.0 Nintendo Direct/E3 Footage

8.1 Frame Rates
8.2 Recordings at 60 fps
8.3 Recordings at 30 fps

8.3.1 Adjustments
8.3.2 Explanation​

8.4 Inputs​

9.0 Applications

10.0 Glossary

1.0 Introduction
The guide gives an approach to finding frame data without hacking which may or may not be useful when the 3DS version is released. Using this guide, you should be able to identify the start and “IASA” frames of a move, or an action as I’ll define later, to varying degrees of accuracy that depends on what you have at your disposal. I don't have a formalized approach for analyzing the middle of a move; yet. I think people can play it by ear once they're done with guide. Other frame data I haven’t looked deeply into.

The following is for if you're short on time. I recommend that you read '3.0 The basic method' to get the basics. You can practice using the link provided in '6.0 Downloads'. If you have problems or questions using the basic method, read section 4.0 Or sections 4.1 and 4.2, if you have problem finding the start and end of a move respectively. If you want to analyse E3/Direct footage/other peoples footage, read section 8.0.

If you don't understand any terms or a sentence, see "10.0 Glossary". If things are still unclear, see "2.0 Definitions". If are still unclear, I've got work to do. Just in case you don't know, you can use the finder [opened by ctrl + f] to find each section quickly.

I apologize in advance for any mistakes. I can only apologize to the people find the guide obvious to a degree. If most people think what I say needn't be said. I’ll consider pruning the guide further.Also I have done only some testing if my definitions or approaches don’t fit in some or many cases, tell me.

2.0 Definitions

I would suggest you read this even if you are familiar with some of the terms. So, we can definitely be on the same page. Of course if an established term more easily explains something, I will eventually change the guide. I suggest you read the definitions in order because one definition can lead to the next.

2.1 Frame

A frame is a length of time that lasts about a sixtieth of a second. The end of one frame is assumed to be the start of the next frame. A frame corresponds to a cycle of instructions [a.k.a a cyclic code] that a computer executes like the following.

loop start - [start of a frame]

get user input

simulate physics

update animation

update everything else

draw everything to screen

idle time

repeat/loop start - [end of a frame]

As shown, important processes of the game happen within frame intervals. Source: here and here. For the second source look at BlueRaja’s post.

2.2 Rendered Images

Rendered images or just images, as a shorthand, are the result of the “draw everything to screen” stage in the above code. Simply put, an image is what’s seen on screen when the game is paused - minus the help icons. Images are basically screenshots in this guide like image 1.

One image is generated every frame so a change in image implies a change in frame. Thus, we can measure frame succession using image succession. Although, if you have a video editor, you can move in units of frames.

2.3 Pose

A pose is a part of an image. More specifically, a pose is a position that a character takes in an image. In image one, Sheik and Lucas are in standing poses. For this guide, I will mainly refer to a pose rather than a whole image.

Spoiler: Image 1

2.4 Animation

What’s seen on screen when playing SSB is an illusion of motion. An illusion created by the animation process. A process that shows each image consecutively where each image is changed to the next every sixtieth of a second approximately.

2.5 Action

Actions are what we observe a character does on screen. An action could be jumping, standing, dashing, walking, attacking and so on. Actions are caused by inputs or lack of inputs and they have a start and end. An action is a part of the illusion created by the animation process; you’re focusing on the character only. More importantly, an action is a result of a series of poses; each poses changes approximately every 60th of a second.

3.0 The Basic Method

3.1 Example 1.a

I think the basic method is best shown by example. So we’re going to confirm that Sheik is crouching for four frames when she jumps. The frame data is linked here.

Look at these five consecutive poses of Sheik. Do you notice any pattern?

Spoiler: Images 1-5

Look the next pose after the original five poses.

Spoiler: Image 6

The 6th pose looks out of place, right? Call the sixth pose an odd pose. Doesn’t it look odd compared to the sequence of previous poses? Thus the next definition arises. An odd pose is a pose that’s most likely out sequence when compared to previous poses.

Assume that the odd pose is the first pose of Sheik’s jump action so the next pose is the second pose of Sheik’s jump action and so on. Based on Sheik’s data, Sheik’s airborne on the fifth frame of her jump action. If Sheik is airborne on the fifth frame of here jump, Sheik's fifth pose should appear airborne. Count from image 6 as pose one till you think Sheik is airborne. The frame before is when Sheik stopped crouching.

Spoiler: Image 6-11

3.2 Example 1.b

Below is a video of Sheik’s jump action. Most of the following examples will use videos rather than images or both. Here’s a method you can use when you look through it.

• Observe the odd pose so you can identify the initial frame
  
• Starting from one on the odd pose, count till Sheik’s pose is in the air
  
• Take one away from the number to find the crouch frames

As an aside, I’m using poses and frames interchangeably. E.g the count of pose four is equivalent to a count of frame four. Poses and frames are different. One’s a window of time and the other is an image. However poses are dependant on frames [look up definition if you don’t understand]. So we can use them interchangeably. This is a reminder to myself as well.

3.3 Summary

We now have a simple method for analysing frame data. You find an odd pose to find the start of an action. You know a characteristic of the pose on the end frame so you can tell when an actions ends. [For example, “a characteristic of the pose on the end frame” is Sheik’s pose looking airborne. Specifically, the “characteristic” is “looking airborne”] You count the frames to see the length. As a step by step method it’s the following.

1. Find the start of an action by finding the odd pose
   
2. Count up the end of an action starting from one on the odd pose
   
3. ????
   
4. Frame Data

3.4 Animation Interruptions

An animation interruption occurs when one action is replaced with another. An action, that is defined in this guide, is caused by a series of consecutive poses. The interruption occurs when the first pose of a new action replaces the pose that should exist if the old action continued. The first pose of the new action rarely looks in sync with the poses of the old action; it looks odd by comparison. So a seemingly odd pose can indicate a start of another action. That’s the reason why my method works.

4.0 Problems with the Basic Method

4.1 Finding an Odd pose

Say I show you just image six and five. You should be able guess that image six has Sheik in a odd pose. I assume you can tell because of the distance she moves between the two frames and the sharp change in pose. So what if that wasn’t the case? This is first of two problems for step one. The difference between the odd pose and the pose before can be very subtle. Approaches one and two are solutions that worked for me.

Spoiler: Image 5-6

4.1.1 Approach One

Consider Sheik’s example. She’s standing for images one to five then she crouches for images six to nine. Based on images one to five, guess Sheik’s next pose and compare this with image six. The difference suggests that image six is an odd pose. For real accuracy, you could record every frame of Sheik’s standing action, match them to my images and see, rather guess, that they’re different.

Spoiler: Images 1-9

In a general sense, you find an odd pose candidate which is an pose which is seems discontinuous. Based on three or more previous poses, guess what the pose would look like. Compare your guess with the odd pose candidate. If there’s a notable difference, you likely have an odd pose. It might help to focus on different body parts.

I say three or more poses because I think you need that many to determine a sequence. Like you need three or more elements to determine a sequence of numbers IIRC. All that is assuming, the poses do show a sequence which I think they roughly and generally do.

Of course, you probably did the above unconsciously. If you’re not intuitively sure what the odd pose is, you can use approach one to analytically find it. At least, it helped me sometimes.

4.1.2 Approach Two

Basically you make the odd pose obvious compared to previous poses. In other words, make the first action visually distinct from the second action which you want to investigate. For example, if you want to see the odd pose for an fsmash. Smash the analogue stick in the opposite direction the character is facing. For example, compare Marth’s Fsmash facing forwards and backwards.

Image thirteen is first pose of Marth’s fsmash in the direction he’s facing and image twelve is a pose before that. Image fifteen is the first pose of Marth’s fsmash that is opposite to the direction he’s facing and image fourteen is a pose before that.

Spoiler: image 12-13

Spoiler: image 14-15

The odd pose is much easier to see for Marth’s fsmash in the opposite direction he’s facing. The same can be true for other combinations
animation interruptions.

The first video features Marth fsmash in the direction he is facing. Video 3 has Marth fsmash in the opposite he is facing.

You need control of the footage to use this approach. If you don’t, you can try to find a more obvious initial action. So if I wanted analyse some jump, I would look for poses where the character is running, walking ducking or crawling beforehand.

Approach Three
I used approach three and four when a number of sequential poses could have been considered a start of an action.

Approach three is the following:

• Find an action that's visually distinct from the interrupting action so the odd pose is easy to see.

Example 1.c

Check the following video of Bowser jumping.

What was the first pose of Bowser jump? I guess you chose one of three odd poses.

Spoiler: images 16-18

I suppose the reasons are:

• the first image is definitely odd compared to the previous frames in the video.
  
• The image seventeen and image eighteen have a more notable change in positioning compared with the image sixteen and image seventeen respectively.
  
• Bowser’s starts to crouch in the third image.

So three odd poses candidates exist. A simple way of checking is changing the interrupted action. In the video, Bowser's standing action is interrupted. So I'll interrupt Bowser's crouching and standing animations. Watch the following two videos.

The pose in image nineteen and image twenty are the first similar poses between Bowser crouching and sheilding respectively. This suggests that the first image is the first frame of Bowser’s jump. This method is generally enough if you know what inputs are made. If you don’t know the inputs, it’s somewhat more uncertain. For more certainty you could use approach four which is basically proof by contradiction using knowledge of inputs mostly. [It’s not done and too long].

Spoiler: images 19-20

4.2 Finding the End of an action

4.2.1 Example 1.d

Back to the Bowser video. Say, we wanted to find how many frames Bowser remains crouching during jumping. We know the first pose of Bowser’s jump so try counting till you think he’s off the ground.

Without the frame data, the pose where Bowser’s jump ends is unclear - at least to me. This brings us to another weakness of my method. It can only accurately identify the start of actions - assuming the odd pose is correct. We can get around this. An odd pose is a start of a new action so the pose, before the odd pose, must be the end or last pose of the old action.

For the Bowser example, we’re interrupting Bowser’s crouching action with a jump. The result is finding the start of the jump. Likewise, now we interrupt the jump action with another action.

4.2.2 Jumping

A bit about jumping. A jump lags before the character is considered airborne by the game. The crouching portion of a jump is the lag. If a crouch ends on some frame, the character is considered airborne on the next frame. Bowser, like any character, can start an aerial attack action on the first frame he’s airborne. Assuming Bowser did this, we should see an odd pose on the first frame he’s airborne. So the odd pose for his aerial attack would indicate the end of Bowser’s crouching stage.

4.2.3 Buffering

How do we know Bowser uses his aerial on the first frame airborne? An answer is buffering. For purposes of this guide, buffering is a tool used to execute subsequent actions as soon as possible. If you want to know why and how it works, I suggest you read the linked guide here. Bowser can’t start another action during crouching. He can only execute a command as soon as he’s airborne. In other words, it’s only possible for Bowser to attack on his first airborne frame.

4.2.4 Back to the Example 1.d

In the video below, I buffered a neutral aerial that occurs right after Bowser’s crouching portion. Count from the first pose of Bowser’s jump till you see another odd pose nearer the end. The other odd pose is the start of a new action so the ducking segment ended a frame ago.

He crouches for eight frames, right?

4.3 IASA Frames

IASA frames is a term often used for frame data. In this guide, an IASA frame is the earliest frame in which an action may be interrupted by an input. An IASA frame is equivalent to the frame in which the first pose of a new action occurs, because buffering by definition allows for inputs to be executed early. For Bowser’s example, the first pose of his Nair occurs on the IASA frame of his jump. Exceptions exist but that’s generally the case.

I think IASA frames should be considered the end of moves, input wise, because players generally want to act as soon as possible. So, I don’t record the frame when an action ends. Although you could guess it for yourself.

Generally IASA frame is independent of the buffered input. That means for Bowser’s example, IASA for his Nair is the same as his Fair, Bair, forward special and so on. For some jab combos, the IASA depends on if you hold the attack button, press it again or use another input. Other exceptions exist as well.

4.4 Approach 5

• Buffer an input during the action of interest to find the IASA frame
  
• Count from the start of the action to the odd pose of the next action
  
• The number you count to is the IASA frame

“to buffer” means to input early or to input within the non-interruptible duration (e.g. start up lag and ending lag) of an action. The game only remembers buffered inputs for ten frames so buffering is done near the end of action; the game forgets otherwise. Or buffering is done quickly if an action can be interrupted earlier than 10 frames e.g. jumping and jabbing

4.5 What if I can’t control the character?

4.5.1 Finding the start of an action

I used approach three. For the direct video, I analysed screenshots of characters doing jumps from standing, walking and so on. The consistent odd pose candidate was considered the start of the jumping action.

4.5.2 Finding the IASA

You don’t know that a move is buffered so you can’t confirm if you’ve found IASA frame. Instead, an animation interruption on say frame ten means the interrupted action can definitely be interrupted on frame ten or greater. The interrupted action may be interrupted earlier for all we know.

4.6 Summary

I guess that’s the basics. The essence of method is identifying the start and “end” of an action by looking at poses. Identifying the odd pose(s) reliably is where it can get more complicated. You just frame count from there mainly. I have used the method to analyse parts of an action such as hitstun, hitlag and hit on frames but I haven’t tried finding shield hitlag or shield stun yet.

5.0 How to Obtain frame data yourself

5.1 What you’ll be working with

Before, screenshots and doctored videos were analyzed to obtain frame data. The basic material I would work with are videos which are recordings of the game. These videos is either directly or indirectly recorded. Direct recordings are by game capture devices and indirect recordings use camcorders or video cameras.

I think of direct recordings as the following. They record images, that the game sends to the TV, at set intervals. On the other hand, indirect recordings record what you see on TV which can cause problems. I recommend direct recordings because of that.

Although direct and indirect recordings are fairly important, a more important property of a recorded video is frame rate which is measured in fps (a.k.a frames per second). I check the frame rate using VLC or KMPlayer for example. Not Windows Media Player to my knowledge. A video recorded at 60 fps means an image was recorded every sixtieth of a second. A video at 30 fps means an image was recorded every thirtieth of a second.

5.2 Where do you get the videos from?

5.2.1 I have a Capture Device

If you have a Game Capture device, this isn’t a problem. The videos will be saved to your Hard drive or an SD card of sorts.

Recommended Game Capture Devices

WiiU

• Elgato Game Capture HD - I personally have one and I'm pretty happy with it. It has minimal lag but it needs to be connected to a computer to record.
  
• Avimedieux - You can record to an SD card so you don’t need you computer to be in the same room. I’ve read the device causes lag so I say buy a HDMI splitter
• Elgato Game Capture HD60 - it does 1080p 60 fps. The HD can only handle 1080p 30 fps. I would recommend this since SSBWiiU will run at 60 fps 1080p. The HD can handle 720p 60fsp but you will add some display lag due to down scaling.

3DS

• 3dsvideocapture I brought a 3ds from here I'll update later. They sell modded 3DS/3DS XL or they can upgrade your own. Long predicted delivery times; 1-2 weeks before an order is dispatched according to an email 10/09/2014.
• 3dscapture 3DS Mods only as of 12/09/2014.
• Merki 3DS Mods only. 3DS XL Mods coming end of October 2014. German based company. It's linked to the 3dscapture guy.

I will look into game capture devices for the Wii and Gamecube.

5.2.2 I don't have a capture device

If you don’t have a capture device, you can download game recordings uploaded to the video sharing websites. I will upload to MediaFire some Brawl videos so you can practice. A number of video sharing websites such as Vimeo or Youtube only support up to 30 fps videos. I use Clipgrab to download YouTube videos.

5.3 How to obtain frame data

5.3.1 Video Editors

I recommend using a video editor to obtain frame data. They easily allow you to progress the video frame by frame and they can easily set and record times accurate to the frame in seconds and tens of milliseconds. All of that is compared to media players such as VLC and KMPlayer which are relatively hard.

When using video editors, you need know how to import media and open a viewer. The other features are unnecessary for analysis or intuitive. Anyway, once you’ve opened the media in viewer, you can start applying the methods above.

5.3.2 Recommended Video Editors

Lightworks

• The interface may be complex but the company has tutorials
  
• Lightworks can be free and it's open source.
  
• I don’t recall any advertisements.
  
• The catch for when your free trial ends is that you can only upload videos to YouTube. For the purposes of analysis, this isn’t a problem.

Cyberlink PowerDirector

• The software isn’t free but you get a 30 day free trial.
  
• Screenshots are integrated into the program unlike Lightworks.
  
• I think you can export media with the free version
  
• Easiest editor to work with in general that I’ve used
  
• The offers to buy the paid edition may be annoying to some

Windows Movie Maker

• The software is free with no paid version so you can make an export videos with no restrictions.
  
• I haven’t found a way to play and edit 60 fps videos on the program so I would only use it for 30 fps videos.
  
• No advertisements
  
• Screenshots can be taken; I haven’t figured out a way to use them easily

6.0 Downloads

Some videos which are linked here for practice. Below I’ve added some descriptions of what I did or thought I did to get those actions.

• Fox Jumps
  
• Link Jumps
  
• Mario Jumps
  
• Sheik Dash to Sheild - smash control stick then immediately held the Sheild button
  
• Lucario Jab to Aura Sphere - press attack button then buffer a neutral special attack
  
• Lucario Jab walk Grab - press attack button then immediately hold control stick forward then press the Grab button
  
• Sheik Nair - Press the jump button the press the attack button before airborne
  
• Sheik Nair 2 - Same input as above. This is meant to show hitstun.

• Link to E3 footage at 60 fps

7.0 Methods and Approaches

This is just a summary of method and approaches discussed so far.

7.1 The Basic Method

• Find the start of an action by finding the odd pose
  
• Count up the “end” of an action starting from one on the odd pose

7.2 Problem with Step One?
7.2.1 Approach One

• find an odd pose candidate which is an pose which is seems discontinuous.
  
• Based on three or more previous poses, guess what the pose would look like.
  
• Compare your guess with the actual pose. If there’s a notable difference, you likely have an odd pose.

7.2.2 Approach Two

• Make the interrupted action visually distinct from the interrupting action so you can see the odd pose easily

7.2.3 Approach Three

• Find a number of new actions that interrupt different initial actions.
  
• When the similarities start and end so does the new action

7.3 Problem with Step Two?

7.3.1 Approach Five

• You need to buffer during the action of investigation if you want to know the end of an action input wise.
  
• If the odd pose of the new action is difficult to see, you could use approaches one to three.

8.0 Nintendo Direct/E3 Footage

8.1 Frame Rates

Almost all of the Nintendo direct was directly recorded at approximately 30 fps. E3 footage varies between whether it was recorded via camcorder or directly. I would recommend you check frame rate of the footage using video editors or video players.

Personally, I would only use recordings at approximately 30 fps or 60+ fps. For example some camcorders or video recorders record at 24 fps. So an image is recorded every two and a half times an image is generated on screen. You may get meaningful frame data from that but it’s much easier to use 30 fps or 60 fps. It gets even harder to accurately obtain frame data if frame rate doesn't divide neatly into 60 such as 25 fps recordings.

8.2 Recordings at 60 fps

Here's one recording of E3 footage linked here. Unfortunately, the TV screen was recorded which caused “afterimages” and its annoying friends in the footage. You can analyze it reliably but it’s more of a hassle in few words. Thus why I prefer direct recordings over indirect ones. The methods and approaches in section seven are applied as described.

8.3 Recordings at 30 fps

I’d say the majority of SSB4 footage is recorded in a frame rate of thirty or less. Recordings at 30 fps are records of every other image the game generates. This results in the following:

8.3.1 Adjustments

• Count from one point to another. Double what you counted to to get the maximum frame length. Take two from the value to find the minimum frame length.

8.3.2 Explanation


The above is a timeline. Each circle represents a time where a pose is generated.The curve represents how the video records images; a curve touching a circle means that image was recorded.

If an image is recorded every other frame, the two curves represent the only two distinct ways images are recorded. Say, the first pose of an action starts at point A. The red curve represents the possibility that it's captured. The blue curve represents the possibility where the the earliest pose recorded is the second pose shown by point B. We don't know which of the two curves was used so we don't know if the second or first pose was recorded. In other words, an odd pose found, for 30 fps videos, can be the first or second pose of an action.

If you combine each curve together, isn't every frame recorded of an action? If you find recordings starting from first and second poses, you can find the exact frame length.

8.4 Inputs

I would use the ideas of this section if for some reason I couldn’t apply the approaches or I was feeling picky. Like approach four, I might explain it but only if I must.

9.0 Applications
Coming Soon-ish
10.0 Glossary

• action - an input dependent movement a character does on screen for example standing, walking, attacking, running and so on. Every action is composed of a series of poses.
  
• candidate odd pose - a pose which may be an odd pose.
  
• frame - a window of time lasting a sixtieth of a second
  
• image - an image that's shown on screen which changes every sixtieth of a second
  
• odd pose - a pose that is likely to be out of sequence compared with previous poses. An odd pose usually is the a first recorded pose of an action.
  
• pose - a character's position in an image

 

[Tristan_win]

5.2.1 I have a Capture Device

• 3DS
  
  Nintendo 3DS video capture board - I’ve read it’s the only and unofficial capture device for the 3DS. The link is here.

~For original 3DS only. 3DS XL/LL not supported (yet).

=( Well until it's supported I guess I wont be able to collect Sheik frame data. ~sighs~ I should of bought a normal 3ds.

On another note I don't think it would be wise to collect data on the tournament as the game is still in development and from the characters we saw frame data could be the one major thing they could fix/change.

 

[ibmutt]

=( Well until it's supported I guess I wont be able to collect Sheik frame data. ~sighs~ I should of bought a normal 3ds.

On another note I don't think it would be wise to collect data on the tournament as the game is still in development and from the characters we saw frame data could be the one major thing they could fix/change.

Firstly thanks for reading, I don't know if you read all of it, but the guide seems pretty long and you still read some. Thanks again.

I want to know as much as I can about the game. I think the best contribution I could make is an analysis and some more objective thoughts. I find that fun. A reason why I made this guide is so anything I say in the game play and analysis section could be repeated and tested by someone else if they knew my method - maybe later though.

A character can be used to check normal landing lag, fast fall landing lag, shielding and so on which tends to be universal. Landing lag and fast fall landing lag seem to be similar Brawl's which suggests they stay the same. Even if a character's frame data changes, I would be interested to see the differences between the demo and the real thing.

I think I see why you would call this unwise. If the frame data is true or false, it must be repeated once the game is released via hacking or no hacking. I'd only recommend my guide before release, if someone thinks they will enjoy using it or they're interested in the potential differences between the demo and the actual game.

 

[Tristan_win]

Firstly thanks for reading, I don't know if you read all of it, but the guide seems pretty long and you still read some. Thanks again.

I want to know as much as I can about the game. I think the best contribution I could make is an analysis and some more objective thoughts. I find that fun. A reason why I made this guide is so anything I say in the game play and analysis section could be repeated and tested by someone else if they knew my method - maybe later though.

A character can be used to check normal landing lag, fast fall landing lag, shielding and so on which tends to be universal. Landing lag and fast fall landing lag seem to be similar Brawl's which suggests they stay the same. Even if a character's frame data changes, I would be interested to see the differences between the demo and the real thing.

I think I see why you would call this unwise. If the frame data is true or false, it must be repeated once the game is released via hacking or no hacking. I'd only recommend my guide for if someone thinks they will enjoy using it or if they're interested in the potential differences between the demo and the actual game.

I think I read over 50% of it. I've done some frame data collecting in the past so I skimmed over what I did know. There was a few things I wish I thought about when I did my testing though, like the whole 'odd pose'. I don't know how much exactly that would help but its a excellent way to make it so you have absolute certainty. Overall a fantastic thread. I would of killed to have something like this back when Brawl first came out, or back in Melee when I was trying to grasp the concepts of frame data for the first time.

A reason why I made this guide is so anything I say in the game play and analysis section could be repeated and tested by someone else if they knew my method - maybe later though.

THIS, things like this is what our community really needs. I don't know if you were lurking back then or not but at one point in Brawl history there was a title called 'Scientist' which was rewarded to people who actively tested things like frame data. I went to them and stated I wanted to join and help, "just tell me how to test and I'll do it." is basically what I said and they brush me aside saying figure out the method for testing was part of the job... Needless to say I never share any of my own collection of data (when I did finally figure out a method) on other characters or publicly release information that would hurt Sheik.

Unfortunately Ibmutt I doubt your thread will ever get much attention. Although most decent players have some understand of frame data there is a overwhelming number of them that would refuse to ever gather it, even among professional players the number of people is few and far between. No a thread like this is aimed at a very small pool of niche smashers in a already small pool so while some people like myself will read, like and bookmark for future reference most will just hit the back button as soon as they see the index.

HOPEFULLY THOUGH, this very nice thread for introducing people to frame data gathering will be linked in the future 'Useful Link Thread' thread so it will be continued to get used for the years to come. If it's not there I'll make sure to suggest it.

 

[Amazing Ampharos]

Needless to say I never share any of my own collect of data (when I did finally figure out a method) on other characters or public release information that would hurt Sheik.

This thread wouldn't have been otherwise notable, but seriously dude? This is like the worst possible attitude. I did so much research in Brawl's heyday, and I can say with total sincerity that I never held anything back. People were convinced my character sucked, and I never worked with any kind of an agenda to somehow "make G&W better"; I put out everything I learned as I learned it to help the community. Hiding knowledge is basically the one and only sin of having knowledge, and while it sucks some of the Smash Lab people may have been jerks to you (I wasn't all that active in the smash lab, but I would have helped you out if I had come across it), that's really not an excuse for doing stuff like that.

The worst part is that a lot of the things you think are secret are probably known by others who have the same lousy attitude about it, and you're really just enabling them to exploit the hapless masses by sitting on knowledge. If you find a tech for another charcter and make it public, anyone else who was sitting on it and exploiting it now has to deal with everyone knowing how it works. If you find something bad for Sheik, share it so now all the Sheiks can know how to work against the unfortunate reality instead of being hit hard by opponents who know when they don't (and please, please no "character backrooms" designed to limit knowledge about characters to only notable mains of that character) and just plain to keep the game honest. I promise if I happen upon something good for Sheik, even though honestly all else being equal I'd kinda prefer Sheik to lose more often, I'd make it public. If we all work together with this kind of attitude, we all win. I don't mean to just go off on you for this, but really, very seriously rethink what you just said here. When the new game actually comes out and we all begin work on truly breaking down how it works, we need to do it right and make sure all the knowledge we can get is available to everyone.

 

[ibmutt]

I think I read over 50% of it. I've done some frame data collecting in the past so I skimmed over what I did know. There was a few things I wish I thought about when I did my testing though, like the whole 'odd pose'. I don't know how much exactly that would help but its a excellent way to make it so you have absolute certainty.

'Odd pose' sounds kind of weird to me. If you have any suggestions, I'm happy to listen. The main use of defining odd poses is as you said - for certainty. It's possible for some actions to be interrupted on the next frame. In some cases, a series of seemingly odd poses can be caused by a series of different inputs and so different actions arise. Since you're interested in the start of a particular action, that could be a problem. I'd thought of number of ways to find when a particular action in that case starts, but the easiest way was approach three.

I would of killed to have something like this back when Brawl first came out, or back in Melee when I was trying to grasp the concepts of frame data for the first time.

HOPEFULLY THOUGH, this very nice thread for introducing people to frame data gathering will be linked in the future 'Useful Link Thread' thread so it will be continued to get used for the years to come. If it's not there I'll make sure to suggest it.

What you said is encouraging. That's exactly what I wanted from this guide. I wanted this guide to be useful for relatively inexperienced people who wanted to find frame data. I wasn't too sure how other people collected frame data. I remember some used hacking so I thought it may be useful for them if they didn't know a way without. I was worried much of what I said was obvious to experienced frame collectors so the apology in the introduction was for you and others like you.

Unfortunately Ibmutt I doubt your thread will ever get much attention. Although most decent players have some understand of frame data there is a overwhelming number of them that would refuse to ever gather it, even among professional players the number of people is few and far between. No a thread like this is aimed at a very small pool of niche smashers in a already small pool so while some people like myself will read, like and bookmark for future reference most will just hit the back button as soon as they see the index.

I never really thought of how popular the thread would be. I was more worried I had made some great error somewhere. Plus, I want anything I say to be easily repeatable and so objective to a degree. I don't mind about the numbers but I hope interested individuals do find this guide even if it's not popular.

 

[RODO]

For real tho...tl:dr

 

[Tristan_win]

This thread wouldn't have been otherwise notable, but seriously dude? This is like the worst possible attitude. I did so much research in Brawl's heyday, and I can say with total sincerity that I never held anything back. People were convinced my character sucked, and I never worked with any kind of an agenda to somehow "make G&W better"; I put out everything I learned as I learned it to help the community. Hiding knowledge is basically the one and only sin of having knowledge, and while it sucks some of the Smash Lab people may have been jerks to you (I wasn't all that active in the smash lab, but I would have helped you out if I had come across it), that's really not an excuse for doing stuff like that.

The worst part is that a lot of the things you think are secret are probably known by others who have the same lousy attitude about it, and you're really just enabling them to exploit the hapless masses by sitting on knowledge. If you find a tech for another charcter and make it public, anyone else who was sitting on it and exploiting it now has to deal with everyone knowing how it works. If you find something bad for Sheik, share it so now all the Sheiks can know how to work against the unfortunate reality instead of being hit hard by opponents who know when they don't (and please, please no "character backrooms" designed to limit knowledge about characters to only notable mains of that character) and just plain to keep the game honest. I promise if I happen upon something good for Sheik, even though honestly all else being equal I'd kinda prefer Sheik to lose more often, I'd make it public. If we all work together with this kind of attitude, we all win. I don't mean to just go off on you for this, but really, very seriously rethink what you just said here. When the new game actually comes out and we all begin work on truly breaking down how it works, we need to do it right and make sure all the knowledge we can get is available to everyone.

~sighs~ Yeah... I know. I'm not going to sit here and argue what I did was right but that was my mindset back then. Honestly though if I did find anything revolutionary for another character I would of said something. It was all in one way or another Sheik related because that's all I cared about... Now I will make the point against sharing weakness as not every character flaw can be worked around. A good example for this would be how you can smash DI down into the 2nd hit of Sheik usmash. There is no true counter play for Sheik to prevent this and it destroys Sheik best killing option. This simple fact hurts the moral of all Sheik mains and ultimately makes her a worst character in every match up that ftilt into usmash is a real possibility. Which is why I never mention it on the boards until someone posted about it.

For this next smash though I do plan to be more open as I don't have anymore illusions that would prevent me from making the right decision.
.
Oh also I think there's two sins, hiding knowledge and lying.

 

[Amazing Ampharos]

~sighs~ Yeah... I know. I'm not going to sit here and argue what I did was right but that was my mindset back then. Honestly though if I did find anything revolutionary for another character I would of said something. It was all in one way or another Sheik related because that's all I cared about... Now I will make the point against sharing weakness as not every character flaw can be worked around. A good example for this would be how you can smash DI down into the 2nd hit of Sheik usmash. There is no true counter play for Sheik to prevent this and it destroys Sheik best killing option. This simple fact hurts the moral of all Sheik mains and ultimately makes her a worst character in every match up that ftilt into usmash is a real possibility. Which is why I never mention it on the boards until someone posted about it.

For this next smash though I do plan to be more open as I don't have anymore illusions that would prevent me from making the right decision.
.
Oh also I think there's two sins, hiding knowledge and lying.

I actually felt really bad for making that post since, while I don't retract anything I said in terms of how being open is important, I was way more harsh than I really needed to be. Sorry for being a jerk; it's not what I want to be on these boards. You are also right that in the long run discoveries can just prove a character worse with little upside (though if there is an upside, making it public makes finding that upside easier!), but you just have to figure that if you found it someone else did too so you might as well just post it so everyone can face the music. Even if you're super character loyal, you have to feel kinda bad for the guy who would have dropped Sheik if he knew that, didn't know that, and ends up feeling like he wasted a ton of time on her down the road when he does find out.

I'm glad you've come around though; I'll be doing a ton of research when the game drops (and I have recording equipment and will probably gather some frame data myself), and it's always good to see as many people as possible ready to do what they can do.

 

[ibmutt]

For real tho...tl:dr

I recommend that you read '3.0 The basic method' to get the basics. You can practice using the link provided in '6.0 Downloads'. If you have problems or questions using the basic method, read section 4.0 Or sections 4.1 and 4.2, if you have problem with the start and end of a move respectively. If you want to analyse E3/Direct footage, read section 8.0.

If you don't understand any terms or a sentence, see "10.0 Glossary". If things are still unclear, see "2.0 Definitions". If they're still unclear, I've got work to do. Just in case you don't know, you can use the finder [opened by ctrl + f] to find each section pretty quickly.