A better 'Super Armor List' Thread (4/30, 18:07 EST)

NoVaLombardia

Smash Journeyman
Joined
Dec 10, 2007
Messages
400
Location
Your Face
#1
After much complaining on the original thread about lack of updates, and me taking initiative to make a new thread, i asked for support before i would do this... So, here it is.

I will try to look at this thread every day, at most 2 days to see any new findings. Usually I will test specific frames to see when exactly the Super armor is active, but until my finals are over, it ain't happening for now.



TABLE OF CONTENTS:

1. Terms Used
2. What is Super Armor?
3. NoVa's Super Armor Theory TRASHED!!!
4. Confirmed Findings
5. Unconfirmed/Untested Findings
6. Super Armor Specifics
7. Trash / Myths


PLEASE READ ALL SA INFORMATION BEFORE POSTING ANY (what you think to be) NEW INFO.




1. Terms Used:
SA = Super Armor
Flinch = An act of your attack being cancelled by another source of damage that causes
knockback


2. What is Super Armor?
Super armor is a state where during an attack, any attack inflicted upon the super armor holder will not cause them any knockback, allowing the attack to continue. Full damage is still inflicted even though there is no knockback.


3. NoVa's Super Armor Theory


BAM!!! NoVa just got PWNT and this lil theory just got trashed! PROVEN FALSE by Doval


Super Armor List




4. CONFIRMED:

Grabs

-All grabs, including Ganondorf's Side-B, Wario's Neutral-B, and Bowser's Side-B, do NOT have SA frames. Simply put, grabs trade hits with regular attacks. If you miss a grab and get hit on the same frame it would initially grab on, you will get knocked back.
-Video proof by Magus420; http://www.youtube.com/watch?v=GeuXNw5P3G4

-"I don't think I'd really call it true super armor, but when a grab connects, it has the same effect as super armor for that one frame. In other words, I can grab someone and get hit from behind (the person I grabbed was not even attacking), but my character does not flinch if the hit connects on the same frame as the grab connects. I believe this mechanic was implemented to prevent freezing glitches or other strange effects." ---by 3GOD

Video proof: http://uk.youtube.com/watch?v=3BAcZLy353o ---by 3GOD


Charizard
  • Charizard's up+B, the initial frames of the move.
  • Charizard's Forward+B, during startup, contributed by crazymasterhand

Donkey Kong
  • Donkey Kong's fully charged Giant Punch, not sure when it has super armor, contributed by Kyle_Wattula
  • UP-B (Grounded) on start-up, by Kyle_Wattula
  • Cargo Carry has massive SA frames, by Kyle_Wattula

Ganondorf
  • Side-B, after you grab someone, until they are released.

Ike
  • Ike's neutral+B, when executing the move but not charging, contributed by Miracle Matter
  • Ike's up+B, I believe from when he throws his sword up until he grabs it he jumps to grab it, by Ankoku

King DeDeDe
  • UP-B, once in motion and ends slightly before peak of jump. contributed by Doval


Olimar
  • Olimar's down+B, initial frames of the move, contributed by takun

Pit
  • Pit's Down+B, the Shield. Confirmed in the video by Ryanarius

Snake
  • Snake's Up+B, during the initial frames of the move, contributed by Miracle Matter
  • Up+B is not standard SA

Squirtle
  • F-smash, by Serpit

Wario
  • Wario's Forward Smash, contributed by Raph_Stryker
  • Wario's down+b

Yoshi
  • Yoshi's second jump, during the time he's jumping, contributed by Raph_Stryker
  • 2nd jump is not standard SA



5. UNCONFIRMED / UNTESTED:

Donkey Kong
  • F-air

Ice Climbers
  • Side-B (Squall Hammer), in the first few frames, contributed by slacker!

King DeDeDe
  • Neutral-B, right before he sucks you in, by Mew2King





Squirtle
  • F-smash has exactly 1 frame of SA, right before the launch, but unsure at what exact frame, by NoVa





6. SPECIFICS:


Charizard
, contributed by NoVa
UP-B (Spiral Jump): from start to disabled (blinking in air), lasts 1.08sec.

  • Starts at Frame 4 (0.0666...sec into the attack)
  • Ends at Frame 13 (0.21666...sec into the attack).
  • Consists of 9 SA Frames, or 0.15sec worth of SA Frames, which consists of 13% of the UP-B.

SIDE-B (Rock Breaker): from start to finish (recovered and can do anything after), lasts 1.00sec, or 60 frames.

SIDE-B attack (1st frame to initial hit frame), takes 0.36sec, or 21 frames.
  • SIDE-B attack's SA starts at Frame 1
  • SIDE-B attack's SA ends at Frame 19 (0.31666...sec).
  • SIDE-B attack consists of 88.888...% SA frames.
  • SIDE-B consists of 31.666...% SA frames

SIDE-B attack's initial hit frame = no SA frame. It is a regular attack, and can be negated with another strike (also seen as two swords clashing), therefore a grab will trade hits at the initial hit frame 21.

Donkey Kong, contributed by 3GOD
Neutral B (Giant Punch // FULL CHARGE): Initial hit starts on frame ___, attack lasts ____ frames.
  • Starts on Frame 17 (0.28333...sec)
  • Ends on Frame 20 (0.333...sec)
  • Consists of ___% SA

UP-B (Spinning Punch // GROUNDED):
Initial hit starts on frame ___, attack lasts ____ frames.
  • Starts on Frame 10 (0.1666...sec)
  • Ends on Frame 16 (0.2666...sec)
  • Consists of ___% SA

Ike, contributed by NoVa
Neutral B (Eruption // UNCHARGED [tapped]): From initial start up frame to initial hit frame, takes 0.5 sec or 30 frames.
  • Starts at frame 21 (0.36sec)
  • Ends at frame 30 (0.50sec)
  • Consists of 30% SA frames (0.14sec)

Up-B (Aether): On level ground -> initial frame to initial landing frame = 1.2sec or 72 frames

I haven't tested lag time on the initial hit frame when an opponent is standing on top of you.

DISCLAIMER: I am still unsure of the ending SA frame. IMO, its pretty difficult to test
the ending SA frame of aether by yourself, and keep an eye on the timer at the same time.
Although unsure, i feel I got pretty **** close to the real thing.

  • Starts at Frame 18 (0.30sec)
  • Ends at Frame 48(?) (0.80sec)
  • Consists of 41.66...%(?) SA (0.50sec)

Wario, contributed by 3GOD

F-Smash: Initial hit starts on frame 9, attack lasts ___ frames
  • Starts at frame 8 (0.1333...sec)
  • Lasts through frame 10
  • Ends at frame ___
  • Consists of ___% SA

Down-B Initial hit starts on frame ___, attack lasts ____ frames
  • Starts at frame 5 (0.08333...sec)
  • Lasts through frame 10
  • Ends at frame ___
  • Consists of ___% SA



7. TRASH:

Bowser
  • Down+B, contributed by Raph_Stryker, PROVEN FALSE by NoVa

G&W
  • U-Smash, after the initial start up, contributed by Raph_Stryker , PROVEN FALSE by NoVa

Ike
  • AAA combo, has super armor after initial jab, contributed by Miracle Matter PROVEN FALSE by NoVa
  • D-tilt, after the start up frames, contributed by Miracle Matter PROVEN FALSE by NoVa

Jigglypuff
  • Down-B, contributed by Corner-Trap, PROVEN FALSE by NoVa
  • These are invincibility frames, not SA frames.

Lucas ***UPDATED***
  • U-smash, during the initial frames of the attack, contributed by Krlos F. PROVEN FALSE by Doval.


Snake
  • Snake's Down tilt, as the hit ends. PROVEN FALSE by NoVa

Sonic **UPDATED**
  • Up+B, by TheCatInTheHat, PROVEN FALSE by Doval
  • UP+B instead has invincibility frames, from start-up to the frame before when he can first act. contributed by Doval

Squirtle
  • Side-B, this does not have SA, its just some type of gay invincibility
  • IF ONE MORE PERSON POSTS SOMETHING ABOUT THIS IM GONNA KILL EM >_<

Wolf **UPDATED**
  • Down-B, the moment the reflector comes out, contributed by shadowmm151, PROVEN FALSE by Doval
  • Down-B instead has invincibility frames
 

slacker!

Smash Journeyman
Joined
Oct 23, 2007
Messages
430
Location
Carrollton, TX
#2
+++thread. A little bit unorganized (colorsss), but better than last one.


ICs get SA in their Squall Hammer (side+b) in the first frames, not sure EXACTLY at what frames, but they do get SA.
 

omegablackmage

Certified Lion Rider
Joined
Oct 16, 2004
Messages
1,902
Location
Spencer, MA
#3
thank god, its about time. also thank you for removing the game and watch up smash, i tested that quite a bit and found no SA frames, while the OP in the other thread seemed to ignore my testings.

also yeah, if you cross out previously thought to be SA frames (which is a good idea, prevents people from posting what we've already tested) make sure they go at the end of the list, bunched together, its much nicer on the eyes that way. Also work on some organization. Other than that thanks a lot for the initiative.
 

Brahma

Smash Journeyman
Joined
Dec 13, 2002
Messages
455
Location
South Bend, Indiana
#4
Not sure why DK's punch is on the unconfirmed/untested list. It gains SA shortly after release, when DK's fist is moving forward.

DK also has SA frames on his Cargo carry (Fthrow). Not sure on the frames, and not sure how it works. For example, it you grab Snake out of a grenade and it explodes on you, he recieves knockback while you don't. If you try the same with a landmine, you both get knocked back.

Also, on DK's UpB, the SA is basically the whole time he has his arm raised behind him on the startup animation. This is tur in the air and on the ground.
 

NoVaLombardia

Smash Journeyman
Joined
Dec 10, 2007
Messages
400
Location
Your Face
#5
Are you sure the UP-B isn't just the hits negating attacks? Or has someone come up from behind or below and you didn't flinch?

DK's punch is controversial, because a few people have PMed me different results, and not only that, but i keep getting negative tests. I wish i could test it some time soon but my work load is way overboard right now in school. If you could do frame-by-frame tests that would be much appreciated.
 

scrubadub

Smash Rookie
Joined
Mar 21, 2008
Messages
24
#6
question... does getting hit while in super armor frame (SAF) slow down your attack? For example, on the frame you get hit, your animation freezes as to absorb the hit but not flinch then your attack completes...
 
Joined
May 22, 2006
Messages
6,933
#10
Actually, the day you made this I was going to ask someone in the thread to create a new one, so thanks for taking initiative. Whenever I get on the boards (when I'm not on my phone), I'll update the sticky with this thread.

By the way, someone seriously needs to figure out an easy method for testing for SA frames. It would make this thread progress smoother, faster, and easier.
 

Magus420

Smash Master
Joined
Dec 13, 2003
Messages
4,541
Location
Close to Trenton, NJ Posts: 4,071
#11
Think people. If the move is a grab and it's right when it's connecting, it is very very likely that it is NOT superarmor you are seeing. Read the part about grabs near the beginning. If you can't get it to work when the grab based move misses them then there is your answer.

You should clarify that part actually. Grabs don't outprioritize attacks; they trade hits with attacks. If they outprioritized them the grabber wouldn't get hit at all, which is what they did in Melee.
 

FzeroX

Smash Journeyman
Joined
Mar 20, 2007
Messages
305
Location
In your Shadow
#13
I know for a fact that Olimar's down B has SA frames.

I have a replay of olimar being falcon punched several times in a row by two falcons doing his down b and goes absolutely nowhere .

I dont have my wii right now so i cant record but it is really easy to test. I believe that the SA frames start when he pull out the whistle to when he stops blowing it.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
#14
By the way, someone seriously needs to figure out an easy method for testing for SA frames. It would make this thread progress smoother, faster, and easier.
Yeah, I've been playing around with various ways to test super armor, but we have to be careful because sometimes attacks simply "clank" rather than it actually being super armor.

However, as near as I can tell, ROB's laser never clanks any attacks, so it should be the standard for testing super armor. I go to training mode, set the speed to 1/2, and do my testing while capturing the video. You have to do multiple tests, so it's not quick by any means, but at least we can get consistent results this way.

By the way, I know Wario's Fsmash super armor starts on Frame 8 while the hitbox comes out on Frame 9. I've yet to find out how long the super armor actually lasts, but it's kind of irrelevant since that attack will clank with most other attacks on frames 9-11.
 
Joined
Jul 18, 2007
Messages
1,525
Location
San Antonio, TX
3DS FC
0404-6991-4531
#15
Thanks for making this. I don't actually have the game. I just posted it because I thought it was useful.

Sorry if everyone complained. I thought the guy from shoryuken would take over, because he PMed me asking about it. I welcomed it, because I don't really care much for Brawl.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
#16
Ok, after a lot of testing, I can verify that Wario's Fsmash has super armor on frames 8-10. His fully charged Waft has super armor on frames 5-10. That waft was ridiculously tedious to test since I could only test it once every 2 minutes or so.

Getting back to the Grab SA stuff...

I don't think I'd really call it true super armor, but when a grab connects, it has the same effect as super armor for that one frame. In other words, I can grab someone and get hit from behind (the person I grabbed was not even attacking), but my character does not flinch if the hit connects on the same frame as the grab connects. I believe this mechanic was implemented to prevent freezing glitches or other strange effects.

Video proof: http://uk.youtube.com/watch?v=3BAcZLy353o

By the way, this "grab super armor" works for Wario's Bite as well (he won't flinch if he Bites someone and is hit from behind on the same frame). I haven't tested other grab moves, but I assume they also will work the same.

Obviously, if the grab does not connect, the character will still flinch from attacks.
 

Magus420

Smash Master
Joined
Dec 13, 2003
Messages
4,541
Location
Close to Trenton, NJ Posts: 4,071
#17
Glad someone finally tried this situation. I haven't had access to the game to do it since I did the original test however many weeks ago.

If you could, try it with Samus or someone with a long grapple grab to see if it occurs on the connect/pulling in frames when you land grabs on people. There should probably be at least 2 frames in those cases instead of the 1 normally. That would make sense I think if it's during that time when you aren't yet holding them but extremely quickly pulling them towards you, and probably keeping them from being launched off the screen if the grab were cancelled at that moment.
 

NoVaLombardia

Smash Journeyman
Joined
Dec 10, 2007
Messages
400
Location
Your Face
#21
ill just do 24-hour time then from now on :p

also 3GOD im curious to know how you do your tests, cuz if it's any easier, i wanna know cuz im a lazy bastage :D

also, can you somehow test that with 3 controllers and team damage on? I don't have 3 controllers >.>
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
#22
3GOD, are you testing from behind as well to make sure it is true super armor?

oh and very nice find on that grab video
Yeah, for the DK punch at least, I was testing with DK facing away from ROB. Basically ROB's laser never clanks with anything, so the only possibility of someone not flinching is super armor.

ill just do 24-hour time then from now on :p

also 3GOD im curious to know how you do your tests, cuz if it's any easier, i wanna know cuz im a lazy bastage :D

also, can you somehow test that with 3 controllers and team damage on? I don't have 3 controllers >.>
I go to training mode and set the speed at 1/2 (this allows you to capture and view 60 frames per second). For the DK punch, I'd charge the punch, and then have ROB laser him and release the punch at the same time. I did this multiple times (took close to an hour probably). While doing these tests, I was capturing the video with my computer, and then I analyze the video frame by frame.

DK gets hit out of his punch on frame 16 and frame 21, but he does not flinch on frames 17-20. The tedious part is getting the tests to line up perfectly to get the exact frame data. For instance, I knew after about 10 minutes that the DK punch had SA on frames 17-19, and I knew he did not have it on frame 16 or 21. It took about 30 minutes more to get a test that finally confirmed frame 20 as having SA as well....

Testing Wario Waft was a nightmare since I had to wait about 2 minutes between each test.

Test what with 3 controllers and team attack on? The grab mechanic?
 

NoVaLombardia

Smash Journeyman
Joined
Dec 10, 2007
Messages
400
Location
Your Face
#23
yeah the grab mechanic. I'm just curious what would happen with that.

And shucks. I don't have any software for video capturing equipment for my computer, i have no laptop, and i have no means to hook anything up from my TV to my computer, lol. So i have to stick to the tedious VS. mode timer.

BTW i don't know whats the fastest projectile, but falco's laser works too. I might want to try ROB from now on though.

Also, as a favor, do you think you can retest the frames on Ike's Aether? I had a difficult time testing that.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
#25
yeah the grab mechanic. I'm just curious what would happen with that.

And shucks. I don't have any software for video capturing equipment for my computer, i have no laptop, and i have no means to hook anything up from my TV to my computer, lol. So i have to stick to the tedious VS. mode timer.

BTW i don't know whats the fastest projectile, but falco's laser works too. I might want to try ROB from now on though.

Also, as a favor, do you think you can retest the frames on Ike's Aether? I had a difficult time testing that.
I'll see about testing Aether later - seems like it would be troublesome to test.

By the team attack thing - I'm not sure what you're wanting...one player to hit his team mate while his team mate grabs the opposing team player? Grab your own team mate while getting attacked? I don't see what situation could cause any different effect than what's already been displayed.

I don't use ROB for speed so much as I do for priority since no attacks will stop the laser (other than reflectors obviously). Actually, for the way I test, I like the fact that his laser has a slight delay (in 1/2 speed) since it allows me to press the button and then somewhat time the other players attack to test all frames.

By the way, I tested DK's Up-B (spinning Kong)...it has SA on frames 10-16 when grounded. I'll update when I get the aerial Up-B data.

EDIT: I figured out you want to know what happens if the person who gets grabbed takes damage - do they have a frame of super armor also? I'll try to test this out....

EDIT 2: I confirm that a person being grabbed also gets one frame of apparent "super armor."

Also, Magus420, I tested Samus' grab and it seems she only gets SA on the frame where the grapple beam connects. She is vulnerable after that frame (the frames where they are being pulled toward Samus by the grapple beam).
 

Magus420

Smash Master
Joined
Dec 13, 2003
Messages
4,541
Location
Close to Trenton, NJ Posts: 4,071
#26
I go to training mode and set the speed at 1/2 (this allows you to capture and view 60 frames per second).
Just wondering, are you able to consistently get 6 frames for grabs doing it this way on many separate recordings at different times? I'm guessing Luigi's would be among the 'standard' grab speed and I know for a fact his is 6 through my normal in-game speed 59.94fps recording (after deinterlacing the source video properly).

I was originally using a slower speed too but simply to make timing things easier and not for the recorded video itself since I get the 60fps fine normally, but I stopped and went back to normal speed when I realized the game seems to do speed changes differently. Instead of just multiplying existing frames, it actually animates and creates new frames to keep the video playback smooth.

When capturing with Composite or S-Video (in normal speed or otherwise) you get 29.97 frames per second, but you are actually getting 59.94 fields per second with 2 fields interlaced in each frame. Under normal game speed each of those fields contains 1 unique in-game 'frame', which gives you (and your TV when using those cables) the ~60fps of gameplay.

The way you are capturing/deinterlacing (and pretty much anyone that doesn't do the deinterlacing themselves) you are discarding one of these fields per frame (probably the bottom/2nd field), which contains half the in-game frames, leaving you with only ~30 frames per second of captured video at half the height (half the video data per frame was removed), and usually resizes it vertically to fill in the gaps left from removing the 2nd field or it resizes horizontally to match the height. When actually done right though, you would instead split these fields apart into their own separate frames giving you a ~60 frames per second video at half the height (you could either scale this up vertically or down horizontally the same way).

You are attempting to work around this by slowing down the in-game speed by half, so after you inevitably lose half of that information through your program's deinterlacing method and are left with ~30fps from the signal, only half a second has passed of 'in-game time' anyway. However, when recording in 1/2 speed, you're actually receiving ~120 in-game images per 'in-game second' in the video signal. The game is filling in the parts inbetween the normal frames so that it looks prettier.

Now, when slowing down the game speed and the game is adding in these 'halfway frames' to make the animation smoother... what would happen on in-game frame 5.5 on a move that hits on in-game frame 6? It shouldn't hit right? Let's say it doesn't.

Now what if the video signal syncs up on a particular recording so that when the program you are using goes to deinterlace the video (let's say it uses the top field and deletes the bottom field) it is scrapping the bottom field that contains the 'whole'/'real' in-game frames, and using the top field that contains those 'fake' halfway frames? You would be left with a video containing frames 0.5 (the first frame where the animation has begun), 1.5, 2.5, 3.5, 4.5, 5.5, and 6.5 (the first frame where it hits). Unless you specifically already knew these were halfway frames, this would appear as though the move hits on frame 7.

However, let's say in a different recording session the signal syncs up so that the preserved top field contains the 'whole' frames and it scraps the bottom field's 'half' frames, it would leave you with frames 1 (1st frame of move), 2, 3, 4, 5, 6 (1st frame it hits). This would instead show it to be hitting on frame 6 which is correct.
 

NoVaLombardia

Smash Journeyman
Joined
Dec 10, 2007
Messages
400
Location
Your Face
#27
man, i don't know why they programmed that.

counting frames through video capture would be easier in melee, because at 1/2 speed, animations were shot in 2's instead of 1's
 

Zankoku

Never Knows Best
Administrator
Premium
GRimer
Joined
Nov 8, 2006
Messages
22,990
Location
Milpitas, CA
NNID
SSBM_PLAYER
#28
Ike's Aether SA frames last from the beginning up until when he actually jumps up to catch the blade - it's possible to hit Ike away while he's rising.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
#31
I was originally using a slower speed too but simply to make timing things easier and not for the recorded video itself since I get the 60fps fine normally, but I stopped and went back to normal speed when I realized the game seems to do speed changes differently. Instead of just multiplying existing frames, it actually animates and creates new frames to keep the video playback smooth.
How did you determine that the game was creating new frames? I feel like you are right because I've noticed (particularly on 1/4 speed) that the game is sometimes unresponsive to commands when slowed down, but I'm just curious how you know this for sure.

Just wondering, are you able to consistently get 6 frames for grabs doing it this way on many separate recordings at different times? I'm guessing Luigi's would be among the 'standard' grab speed and I know for a fact his is 6 through my normal in-game speed 59.94fps recording (after deinterlacing the source video properly).
What do you mean by separate recordings...in other words, I realize that my computer is deinterlacing the source, but I'm not sure when the signal syncs. In other words, I personally don't know if the computer is syncing with the top field or bottom field, and I'm not sure when/what resets this sync in order to perform this test. When I restart my computer? When I restart my capture program? When I start a new recording in the program?

The way you are capturing/deinterlacing (and pretty much anyone that doesn't do the deinterlacing themselves) you are discarding one of these fields per frame (probably the bottom/2nd field), which contains half the in-game frames, leaving you with only ~30 frames per second of captured video at half the height (half the video data per frame was removed), and usually resizes it vertically to fill in the gaps left from removing the 2nd field or it resizes horizontally to match the height. When actually done right though, you would instead split these fields apart into their own separate frames giving you a ~60 frames per second video at half the height (you could either scale this up vertically or down horizontally the same way).
Anyway to record at normal (1x) speed and deinterlace myself so that I don't lose data? I wouldn't mind even if I could just look at fields for information rather than frames. I have some Roxio program for capturing, but I can also use VirtualDub. Maybe VirtualDub has some options to preserve all the fields by simply filling in the empty lines with gray/black/white lines or something...I'll play around with it until I hear back from you on the matter.
 

Magus420

Smash Master
Joined
Dec 13, 2003
Messages
4,541
Location
Close to Trenton, NJ Posts: 4,071
#33
How did you determine that the game was creating new frames? I feel like you are right because I've noticed (particularly on 1/4 speed) that the game is sometimes unresponsive to commands when slowed down, but I'm just curious how you know this for sure.
While I didn't actually analyze the frames in a slowed down recording, it seemed pretty obvious that when slowed to 1/2-1/4 speed it was displaying just as fluid as normal game speed, and definitely not like a choppy 30/15fps animation. This means there must be images inbetween the normal frames to make the movements display just as smooth at that reduced speed. Compare it to Melee's 1/4 speed. It's VERY choppy, because there's only 15 different images displayed per second, since it's showing the same frames 4x in a row so that it takes 4 times as long and is 1/4 speed.

What do you mean by separate recordings...in other words, I realize that my computer is deinterlacing the source, but I'm not sure when the signal syncs. In other words, I personally don't know if the computer is syncing with the top field or bottom field, and I'm not sure when/what resets this sync in order to perform this test. When I restart my computer? When I restart my capture program? When I start a new recording in the program?
I'm not sure, but I would think it's just when you start the recording. However, I'm also fairly sure it can shift mid-capture to maintain the video timing by occasionally dropping/duplicating a field. When I do a recording in 1/2 speed in Melee, at one point in the recording it can be matched so that it's (each in-game image will appear twice due to 1/2 speed):

Frame 1: Field A=In-game frame 1; Field B=In-game frame 2
Frame 2: Field A=In-game frame 2; Field B=In-game frame 3
Frame 3: Field A=In-game frame 3; Field B=In-game frame 4

...etc where the duplicate in-game images are captured in different frames and the fields in each frame are different.

It looks like this:



Now, starting on the next captured frame of the video the field/frame sync shifted, so that the duplicate in-game images appeared in the same captured frame as the 2 fields. This appears like it isn't even interlaced at all. I think it occured here because a field was dropped at this moment (video captures aren't always perfect at maintaing exact timing over extended periods of time and in this case a field was dropped to do it), and this shifted the captured fields around and put them together in the same capture frame:

Frame 1: Field A=In-game frame 2; Field B=In-game frame 2
Frame 2: Field A=In-game frame 3; Field B=In-game frame 3
Frame 3: Field A=In-game frame 4; Field B=In-game frame 4

The resulting captured frame was this:



This is why it's always good to have a timer on and do more than one trial of something just in case it happened to drop/add a field at that particular moment.

Anyway to record at normal (1x) speed and deinterlace myself so that I don't lose data? I wouldn't mind even if I could just look at fields for information rather than frames. I have some Roxio program for capturing, but I can also use VirtualDub. Maybe VirtualDub has some options to preserve all the fields by simply filling in the empty lines with gray/black/white lines or something...I'll play around with it until I hear back from you on the matter.
Yes. I use VirtualDub to do the recording, and use a simple AviSynth script (an extra program you need to install to use it) that I drop into VDub to separate the fields into their own frames and view it there. You don't even need to do any re-incoding since you can just view the source video in VDub using the script and displaying with the fields separated and analyze it right there.

Download AviSynth if you don't already have it installed:
http://sourceforge.net/project/showfiles.php?group_id=57023

Once installed, open Notepad and paste this into it:

AVISource("YourVideo.avi")
AssumeTFF()
separatefields()

'YourVideo' is obviously just whatever your source capture video is called. Go to Save As, and change the file extension from .txt to .avs (avs is AviSynth's script filetype), and save it in the same location as the captured video.

Once saved, you drop the script into VDub and the video you listed in the script should appear at half the height, and if you look at the frame count it will be 2x the normal amount. It's now displaying the fields as separate frames and the frame rate is automatically adjusted to 59.94. You can use it as is for just checking stuff in it, or you can select some VDub filters so it's a bit easier to view and/or if you want to save it as a normal looking 59.94fps video.

OPTIONAL: If you scroll frame by frame the image should 'bob' up and down from frame to frame. This is because fields aren't normally positioned to lay directly ontop of another. To correct this I then add the 'Field Bob' filter in VDub, selecting 'Quarter scanline Up' for the Even field and 'Quarter scanline down' for the Odd field. If it makes it 'bob' more then select the opposite settings in the filter. This should align the frames correctly so they no longer have that 'bob' effect between them.

OPTIONAL (but it makes it much easier to view): Add in the 'Resize' filter in VDub. I usually do 320x240 with the 'Lanczos3' filter mode (since the vertical is already at 240 you're scaling the horizontal to match it). You could also do 640x480, but it will blur a bit since you're stretching it vertically.


With a script sitting around ready to edit in the name of a new source file as needed, the whole process takes no additional time at all from other methods. You can go from stopping the capture to viewing it in 59.94fps with the full amount of in-game frames in about 30 seconds.


Here's an example of a video from Melee captured in normal game time, bob deinterlaced (that's what this method is called btw), and resized to 320x240 to get a 59.94fps video:

http://www.mediafire.com/?t8nguxrsxmi
 

Zankoku

Never Knows Best
Administrator
Premium
GRimer
Joined
Nov 8, 2006
Messages
22,990
Location
Milpitas, CA
NNID
SSBM_PLAYER
#34
The creation of "new frames" is likely the rendering engine interpolating animation between keyframes, and it's probably what they did to create the effect for the Timer item, as well. SDI looks very awkward in 1/4 speed.
 

3GOD

Smash Ace
Joined
Jul 23, 2006
Messages
745
Location
Athens, GA
#38
Lots of great information about video capture...
Thanks a lot - I've been able to retest a few things now with these new techniques, and the difference is small (obviously +/- one frame), but there are some differences.

One quick question...if I'm capturing and deinterlacing as you described above, Wario's Fsmash has super armor on frames (well fields really I guess) 8-10 for sure. However, it does rarely seem to have super armor on frame 7 (maybe 1 out of 30 times that he gets hit on frame 7, he appears to have SA). And probably 90% of the time or more, he seems to have SA on frame 11.

I take it that these discrepancies could be explained by dropping some frames while capturing?

It would make sense that an occasional dropped frame would give the appearance of SA on frame 7 from time to time when actually the game is displaying frame 8 of the attack. And for those rare times when it seems like frame 11 doesn't have SA, the game would be displaying frame 12 really.

If what I've stated here is correct, then Wario's Fsmash actually has super armor on frames 8-11.
 

Magus420

Smash Master
Joined
Dec 13, 2003
Messages
4,541
Location
Close to Trenton, NJ Posts: 4,071
#40
I take it that these discrepancies could be explained by dropping some frames while capturing?

It would make sense that an occasional dropped frame would give the appearance of SA on frame 7 from time to time when actually the game is displaying frame 8 of the attack. And for those rare times when it seems like frame 11 doesn't have SA, the game would be displaying frame 12 really.

If what I've stated here is correct, then Wario's Fsmash actually has super armor on frames 8-11.
That sounds very likely. Are you doing it in vs mode with the timer running? If it's as consistent as it was in Melee you should always be able to tell when one is dropped/added once you familiarize yourself with the timer's counting pattern.

In Melee the clock was always perfectly in tune with the gameplay, but the debug menu's text display (where it states the current frame of what you're doing) can be a little off in normal game speed in that it occasionally will display the same frame number twice (though the gameplay advances one frame), and then skip the next frame on the display to catch up to the gameplay.

As long as the Brawl timer isn't occasionally lazy like the Melee debug display was and always changes with each in-game frame passed it should remove any possible doubt of exactly which frame is being displayed of the move you're looking at.
 
Top