Quote"noone else":You'd be very suprised. As I said, I tested this on other creator's chars and the ALL of them (100%) had the glitches.Have said creators mentioned them as issues? Or is it just you that spots them and believes them to be glitches.I wasn't speaking of characters. Noone coding an AI, which are usually tested pretty thoroughly to make sure the moves activate when they are supposed to even if they aren't even slightly balanced. Are checked for strange behaviour. I've never heard of anyone complaining that their AI isn't doing what it's supposed to. It either doesn't work, or does something too much.
Cyanide said, September 23, 2007, 07:07:43 amQuote"noone else":You'd be very suprised. As I said, I tested this on other creator's chars and the ALL of them (100%) had the glitches.Have said creators mentioned them as issues? Or is it just you that spots them and believes them to be glitches.I wasn't speaking of characters. Noone coding an AI, which are usually tested pretty thoroughly to make sure the moves activate when they are supposed to even if they aren't even slightly balanced. Are checked for strange behaviour. I've never heard of anyone complaining that their AI isn't doing what it's supposed to. It either doesn't work, or does something too much.If not slowed down, in the frame by frame method I mentioned (i.e. my method), then indeed that AI does appear to be doing everything you tell it to.At first I didn't even have this issue.I only managed to see it while creating a video of my own of a match between two CvS/CvS2 creators.Sending a PMTee Hee Hee said, September 22, 2007, 04:58:05 pmHere we are:NHK's - MakotoSenind's - AI coding:My (THH) - CNS, AIR (clsns), and CMD edits http://www.youtube.com/watch?v=TgH-mch1u48Notice at: 3:41The Last Hit of Makoto's Hyper Combo does not connect (Please read the two notes below)Note:I expanded the AIR Attack CLSNs "GREATLY" and I know they should have hit. (Between -43 and 58)The trigger coding I edited for her to use was:Quote[Statedef 3272]type = Smovetype= Aphysics = Svelset = 0,0ctrl = 0juggle = 1anim = 3272sprpriority = 2...[State 3000, 2]type = HitDeftrigger1 = animelem = 6...The problem... animelem lasts for only 1-tickYou would, and should, be tempted to say the distance factored into the equation, but it did not.I tested this many times on my own and found that the hit was not triggering appropriately EVERY time, but instead, again by rough estimate, only about 90% of the time did it trigger approriately...NO MATTER WHAT THE DISTANCE WAS OMG I was wrong about that video.I went back and checked the CLSNs Collisions for Kyosuke and saw that they did not encompass all of his legs, only most of them.Makoto's CLSN Attack and as it turns out, the CLSN Collisions, not the entire body, were actually outside Makoto's range.The issue with Alex, however, is still valid. @_@ I still haven't gotten the picture I needed yet, but I'm confident it will come.
I decided to make this a video:I found that making the video was far easier than trying to capture the single frame and a whole lot less time consuming http://www.youtube.com/watch?v=jsbFbYQcsYoVideo Info:Note: - The scenes that illustrate my point are AT THE END OF THE VIDEO.Scenes 1-4 are simply to illustrate that the video did consistently capture the CLSN Attacks and serve to verify that the CLSNs were legitimately activating (with no tampering with at all).Scenes 5 and 6 - Illustrate where the CLSN Attacks were not activated properly, with the same code and no tampering with at all.What went into the video:Recorded at 65 Frames Per SecondThe coding for the Hitdef activation was the same as what I posted in this MFG forum topic. The CLSN Attack is the same as what I said it was earlier.There are no edits to the specific scenes in the video that highlight my point.THERE IS NO BULL SH@T IN THIS VID!!!!I have other vids that support some of my other points as well but I am not sure I should publicly post them. I did get permission to publicly post one of the two vids, however.
The last one was very weird indeed. It never happened to me though And what I see is not HitDef problem, but the lack of CLSN1. If it were the HitDef problem, the attack wouldn't hit with CLSNs overlapping.Tee Hee Hee said, September 25, 2007, 10:58:58 pmI found that making the video was far easier than trying to capture the single frame and a whole lot less time consuming
the second to last one can be caused by juggle points, they sometimes even interfere on ground combos, and yeah, i also consider that a bug.
Elix said, September 26, 2007, 12:09:19 amThe last one was very weird indeed. It never happened to me though And what I see is not HitDef problem, but the lack of CLSN1. If it were the HitDef problem, the attack wouldn't hit with CLSNs overlapping.Yay, someone agrees I didn't make the title to this post actually... LOL.PotS moved it because it was getting sidetracked, Cyanide moved it back for being in a different maybe wrong section, and moved it to the same section that PotS had disconnected it from. Then I think Cyanide saw that it had the same title as PotS's and changed the title, maybe based on what I suggested in the "technically" introductory post(s) @_@CONFUSING!!! - I couldn't explain that situation well sorry Elix said, September 26, 2007, 12:09:19 amTee Hee Hee said, September 25, 2007, 10:58:58 pmI found that making the video was far easier than trying to capture the single frame and a whole lot less time consuming I feel as though it's just as effective.But that's just my opinion. R[E]ika said, September 26, 2007, 12:19:20 amthe second to last one can be caused by juggle points, they sometimes even interfere on ground combos, and yeah, i also consider that a bug.Score. Mugen indeed does have its bugs.There is another video I have, that I want you all to see, but I feel as though it would be appropriate to re-request permission for viewing it.There is also a third (made for a different purpose), but in which, other Mugen bugs are displayed... alongside its true meaning.
That last punch. It's obvious why it misses, there's no red box. I would query WHY there's no red box. Even when an attack has used up all it's juggle points the red box still displays. Or if there's no hitdef it still displays. Or even when the state you're in isn't classed as A but I or H you'll still get the red box. Why it's not showing on that one frame is rather confusing. Not saying you're wrong here but i think you have the wrong end of the stick. There's no reason hitboxes shouldn't display unless you haven't put them in, or are using a helper for the attack and it doesn't show up. Non activating hitboxes != non activating hitdef.A lot of your misses on that uppercut are more p2 than p1, person has gone for a too accurate idea of them. Not the general lines. Thats not so much mugen as it is whoever's making the character.The standing heavy step forward kick. THAT one confuses me, i could attribute it to the inherent nothitby on getup except for the fact his clsn2 doesn't change to pale blue. Not sure, would need both characters. Oh and he's not getting up.
Cyanide said, September 26, 2007, 01:29:40 amThat last punch. It's obvious why it misses, there's no red box. I would query WHY there's no red box. Even when an attack has used up all it's juggle points the red box still displays. Or if there's no hitdef it still displays. Or even when the state you're in isn't classed as A but I or H you'll still get the red box. Why it's not showing on that one frame is rather confusing. Not saying you're wrong here but i think you have the wrong end of the stick. There's no reason hitboxes shouldn't display unless you haven't put them in, or are using a helper for the attack and it doesn't show up. Non activating hitboxes != non activating hitdef.A lot of your misses on that uppercut are more p2 than p1, person has gone for a too accurate idea of them. Not the general lines. Thats not so much mugen as it is whoever's making the character.I'm not too sure what you meant by the lines that I bolded from your post in the quote.. I don't know if you did, but I'm not sure you read my full posts, BUT, as said before, I don't quite understand what you're saying.I am tempted to respond with these two posts, based on my, most likely incorrect interpretation.Tee Hee Hee said, September 22, 2007, 04:58:05 pmNote:I expanded the AIR Attack CLSNs "GREATLY" and I know they should have hit. (Between -43 and 58)The trigger coding I edited for her to use was:Quote[Statedef 3272]type = Smovetype= Aphysics = Svelset = 0,0ctrl = 0juggle = 1anim = 3272sprpriority = 2...[State 3000, 2]type = HitDeftrigger1 = animelem = 6...The problem... animelem lasts for only 1-tick[glow=red,2,300]The full hitdef was is this:[/glow][mcode];---------------------------------------------------------------------------[State 3000, 2]type = HitDeftrigger1 = animelem = 6attr = S, HAanimtype = hardair.animtype = backdamage = 85guardflag = Mhitflag = MAFpausetime = 12,12sparkno = s66guard.sparkno = s49sparkxy = 20,-90hitsound = s5,9guardsound = s6,1getpower = 0,0ground.type = highground.slidetime = 12ground.hittime = 30ground.velocity = -.7,-12air.velocity = -.7,-12;fall = 1;fall.recover = 0p2stateno = 8000yaccel = 0.53[/mcode]Tee Hee Hee said, September 25, 2007, 10:58:58 pmVideo Info:Scenes 1-4 are simply to illustrate that the video did consistently capture the CLSN Attacks and serve to verify that the CLSNs were legitimately activating (with no tampering with at all).Scenes 5 and 6 - Illustrate where the CLSN Attacks were not activated properly, with the same code and no tampering with at all.
Red clsns = clsn1's = what hits in an attack.There is no reason for the clsn1's not to appear. Nothing you do can cause that. Setting juggle points too high, an impossible trigger. If the clsn boxes have been placed ON the sprite, that's where they stay. They don't just vanish is my point. I'm not saying it's not wrong. I'm just wondering why it's occuring. I've had my share of wierd shit from mugen but that's not one of them.Even if an attack has no hitdef it doesn't make clsn boxes vanish. If they aren't there there must be a reason for it, even if said reason is mugen's broken.Actually, all i can think of is he's set it up so a 1 tick frame has those hitboxes and your video capture hasn't captured them so it makes them appear missing.Just to give you a better idea of what i'm talking about. Take any character, load it up in fighter factory. Then draw a clsn1 (red) box above the characters head on his standing animation. Set it up as a default box. Then test it, you'll see the box is always there even though stand is a non attack state and has no hitdef associated with it.
Cyanide said, September 26, 2007, 02:29:04 amEven if an attack has no hitdef it doesn't make clsn boxes vanish. If they aren't there there must be a reason for it, even if said reason is mugen's broken.YESSSS!!!. I know I wasn't crazy.Hmmm... if a creator had given me permission to make a video public but that was way in the past, but for a different reason (or should I say, one of the reasons that involved one of the glitches I was talking about was brought up in the vid) do you think it would be appropriate of me to re-open that video for viewing if that creator hasn't been on in a while (or at least I haven't caught up to him for a few days).
Is it a public character? I mean if it's private and he doesn't want people to know about it, keep it that way so he doesn't get annoyed. If it's online and readily available then you're being useful providing feedback.I'm personally of the opinion it's a 1 tick hitbox. I suggest you open the animation and check how long the frame is that has the red box on it.
Cyanide said, September 26, 2007, 03:31:26 amIs it a public character? I mean if it's private and he doesn't want people to know about it, keep it that way so he doesn't get annoyed. If it's online and readily available then you're being useful providing feedback.Public so... I'll put up the vid.Cyanide said, September 26, 2007, 03:31:26 amI'm personally of the opinion it's a 1 tick hitbox. I suggest you open the animation and check how long the frame is that has the red box on it.Tee Hee Hee said, September 22, 2007, 04:58:05 pmThe problem... animelem lasts for only 1-tickThat's actually why this whole thread was started. or should I say, why PotS seperated it.We were originally on opposing ends involving this (thread reference at teh bottom):QuoteRenzo [a.k.a. Sachavaca Salvaje] said, September 22, 2007, 03:41:22 pmFor all this time, my general rule was assigning higher priority values for weaker attacks; the logic behind that was exactly:QuoteThe hitting portion of the animation comes out quick enough (many attacks actually only hit a tick or two after the char has reached the animation frame that appears to hit)Finally Someone Understands My Pain.: Tee Hee Hee (myself) in "Your Problems with Mugen Creation" said, September 22, 2007, 03:41:22 pmTriggersTriggers are largely reliable when you are generally coding, but they can encounter problems when you are coding specifically for CLSN Attack activation for the hitdef. Juggling in particular holds this problem up front and center.There are many times where I have realized, that a ton of the things I was absolutely coding for in the hitdef, weren't activating properly (I.E. Time = 1; Time = 0; or an animelem that lasted for less than 2 ticks). If you place the CLSN Attack in an animation, where the attacking frame only lasts for 1 tick, or else directly code the hitdef to activate within 1 tick (i.e. time = 6), then only sometimes, you will notice, that the hit will not make contact . If you're not going for accuracy, or whatever, code it for something that activates for a period of 4 ticks, thereby increasing your chances of having the attack properly activate and connect with the opponent.Movehit activation within specific animelems can also be a pain in certain cases. If that animelem lasts for less than 2 or 3 ticks, quite frequently you'll notice that the char won't activate the hitdef that you defined, meaning, again, that it will not make contact. Make sure the animelem lasts for maybe 3 or more ticks, or else code for more than one animelem.http://mugenguild.com/forumx/index.php?topic=65633.msg541809#msg541809
Scratch previous gibberish.Made a 3 frame, 3 time attack. Quite simple nothing wierd. Tried triggering on both animelem = 2 and time = 1 and time = 0. Punched p2 about 80 times in all situations. Didn't miss once. Well aside from when cornerpush sent me out of range.How many variables are you adding. Like is this standing vs standing? Or part of a combo.I'm happy enough to believe you but i'm going to need to experience it first.
Newly Re-Made Public Video:http://www.youtube.com/watch?v=QWTPVkemiXQThe Specific Flaws:1:33 - roll break flaw2:36 - roll break flaw3:26 - roll break flawNote there are times in the video where Yuri COMPLETES the roll.Note: I didn't check teh coding on the current Yuri but it is still the current one. I doubt though, that PotS would have coded this specifc flaw into Yuri and I do believe that the roll would have required ctrl.
You mean yuri breaking OUT of it into a parry correct?If no human can do that it's an AI flaw. Or pots wanted his AI to do that, for some reason. It won't be a bug with mugen code, it'll be bad triggering.
QuoteHowever in the hitdef (dunno if you've tried this) but using time = 0 and having teh hitframe on animelem 7 will still functionSpent so much time trying to recover this. What you said here was and is right. At the time I had said that a "time = 0" set wouldn't necessarily work, I was referring to a char that had a 1-tick hit box after it also.I was mistaken when I typed that, back then.Cyanide said, September 26, 2007, 03:58:58 amYou mean yuri breaking OUT of it into a parry correct?If no human can do that it's an AI flaw. Or pots wanted his AI to do that, for some reason.I mean Yuri breaking out of it into the parry and Vica Versa (apparently it's supposed to be Vica Versa. 'Latin Class' is a lie if that is not so.)I also noticed similar instances in Waru's Ken (which I have a vid of but never got permission from Waru to show... it was originally meant to highlight some AI advantages so I doubt it would be allowed)Then I put my own and qwer's codes to the test and found that even ours had a few attack breaking into attack, flaws, even though we specifcied "ctrl" for each of the commands that registered some flaws. We also saw from our (PotS, Waru's, and ours) chars, that there were some unregistered vulnerabilities and higher priority overrides that occured very rarely.
There's no Clsn1 on that last one because it only lasts one tick and you clearly didn't record at 60fps, so it can't be captured by the video 100% of the time. Next case.
P.o.t.S. said, September 26, 2007, 11:19:49 amThere's no Clsn1 on that last one because it only lasts one tick and you clearly didn't record at 60fps, so it can't be captured by the video 100% of the time. Next case.Tee Hee Hee said, September 25, 2007, 10:58:58 pmWhat went into the video:Recorded at 65 Frames Per SecondAlso, that didn't answer the question though.How did the attack miss at all? The CLSN did not activate properly and no hit was registered, even when it was in range. This is what is confusing me.R[E]ika answered my other question though about the scene before the last:R[E]ika said, September 26, 2007, 12:19:20 amthe second to last one can be caused by juggle points, they sometimes even interfere on ground combos, and yeah, i also consider that a bug.What's your take PotS?
Tee Hee Hee said, September 26, 2007, 11:28:54 amTee Hee Hee said, September 25, 2007, 10:58:58 pmWhat went into the video:Recorded at 65 Frames Per SecondAlso, that didn't answer the question though.Recording software easily skips frames PLUS there are clear frame gaps between each frame you paused.Stop blaming Mugen/Elecbyte for it, only happens because NHK didn't use large enough Clsn1, when auto-combos should use the kind that never misses (in KOF games that means boxes covering half the screen).QuoteR[E]ika answered my other question though about the scene before the last:R[E]ika said, September 26, 2007, 12:19:20 amthe second to last one can be caused by juggle points, they sometimes even interfere on ground combos, and yeah, i also consider that a bug.What's your take PotS?Assuming you mean the kick, yeah it's caused by a custom juggle system that only resets when P2 stands up. Since the attack started before that happened, the HitDef never activated.
P.o.t.S. said, September 26, 2007, 11:59:13 amTee Hee Hee said, September 26, 2007, 11:28:54 amTee Hee Hee said, September 25, 2007, 10:58:58 pmWhat went into the video:Recorded at 65 Frames Per SecondAlso, that didn't answer the question though.Recording software easily skips frames PLUS there are clear frame gaps between each frame you paused.Stop blaming Mugen/Elecbyte for it, only happens because NHK didn't use large enough Clsn1, when auto-combos should use the kind that never misses (in KOF games that means boxes covering half the screen).Sorry if I'm annoying you PotS. I really would never intend to.So by my interpretation I guess you are saying that the frame was skipped anyway, and the attack missed because the CLSN box still wasn't large enough. I'll try a faster FPS and running Mugen at its slowest speed as well as widening teh CLSN even more.Thanks for teh info PotS.
