General Kong - Bullets and Bananas
^ I could attempt to get back on it. It was bad enough doing the one I completed and you obliterated it. 
Got it.
On to Narcis, I suppose.
On to Narcis, I suppose.
SDA Forum
B+Left, Left, Up+B, ★
I can't believe it's come down to this... Narcis is the only guy left that doesn't match the TAS; we got 3 frame perfect circuits now. I'll send out a mass PM to a list of folks shortly and we'll figure out how to do this.
If anyone else is even remotely interested in this, let me know and I'll add you to the thread too.
If anyone else is even remotely interested in this, let me know and I'll add you to the thread too.
B+Left, Left, Up+B, ★
Alright so, this Saturday at 8PM PDT, I'm gonna stream the Narcis Prince tutorial. If you're interested in doing Narcis Prince IL attempts, or are just curious and want to watch, come and check it out!
Also...
this still stands.
Also...
Quote from zallard1:
If anyone else is even remotely interested in [doing] this, let me know and I'll add you to the thread too.
B+Left, Left, Up+B, ★
Oh yeah, that's a good point presjpolk! The game is technically free if you have a Wii this month and have some Club Nintendo coins lying around (or Nintendo games with inserts in the box that haven't been redeemed yet). They also seem to have Donkey Kong '94 and Star Fox 64 too; both of which I also highly recommend.
https://club.nintendo.com/rewards.do
Also, I've added you to the thread, presjpolk.
https://club.nintendo.com/rewards.do
Also, I've added you to the thread, presjpolk.
HELLO!
Thanks.
Dangit, just realized I'm going to have to beat the Special Circuit to participate in this properly.
Well, I needed an intro to the basic game mechanics of this anyway. Wow, I already knew intellectually this game is a lot faster paced than MTPO, but wow.
Well I'm 4-0 on the minor circult, 4-1 on major (not consistent yet on Sandman), and am trying to figure out how to beat Mad Clown. I guess I have a few days to do this.
Dangit, just realized I'm going to have to beat the Special Circuit to participate in this properly.
Well, I needed an intro to the basic game mechanics of this anyway. Wow, I already knew intellectually this game is a lot faster paced than MTPO, but wow.
Well I'm 4-0 on the minor circult, 4-1 on major (not consistent yet on Sandman), and am trying to figure out how to beat Mad Clown. I guess I have a few days to do this.
B+Left, Left, Up+B, ★
So, the Narcis Prince tutorial is going to happen in about 50 minutes. I'm streaming some ultra rusty SPO practice at the moment.
http://www.twitch.tv/zallard1
http://www.twitch.tv/zallard1
All the things
Below is the RNG sequence in SPO. It calls this most frames while the game is active, regardless of whether you're in a fight or not.
What this means for real-time: You probably can't hope to manipulate events, but if you do some fancy math you may be able to (slightly) increase your chances. Frame counters are always messy to work with in real time, and the Button Sum that I note is only marginally manipulable. The main thing though is that it goes to a look-up table rather than calculating the raw value.
Different buttons correspond to different bits in the Button/Direction Masks. Pressing multiple buttons/directions at the same time will have both bits be set for as many frames as they are held. It also means that not pressing any buttons will make the Button Sum not change.
I can post a dump of the look-up table if any of you want to mathineer your way to some potential improvements, but I think it would be a very uphill battle. I believe it contains all 256 possibilities, so it's just spread out quite a bit. If there's some trend though, you may be able to find something to push it in your favor.
Code:
00b15a jsr $ccb7 [00ccb7] A:0002 X:0000 //Start RNG sequence 00ccb7 lda #$00 A:0002 X:0000 //Clear the accumulator 00ccb9 xba A:0000 X:0000 //Clear X 00ccba lda $0090 [000090] A:0000 X:0000 //Load the Button Masks 00ccbd clc A:0000 X:0000 00ccbe adc $0091 [000091] A:0000 X:0000 //Add to the Direction Masks 00ccc1 adc $004e [00004e] A:0040 X:0000 //Add to current Button Sum 00ccc4 sta $004e [00004e] A:0089 X:0000 //Store back into Button Sum 00ccc7 lda $0043 [000043] A:0089 X:0000 //Load decrementing frame counter 00ccca sbc $004e [00004e] A:00c7 X:0000 //Subtract the Button Sum 00cccd tax A:003d X:0000 //Transfer Accumulator to X 00ccce lda $bc2f,x [00bc6c] A:003d X:003d //Load from RNG Table 00ccd1 sta $004d [00004d] A:0085 X:003d //Store into RNG address 00ccd4 rts A:0085 X:003d //End subroutine
What this means for real-time: You probably can't hope to manipulate events, but if you do some fancy math you may be able to (slightly) increase your chances. Frame counters are always messy to work with in real time, and the Button Sum that I note is only marginally manipulable. The main thing though is that it goes to a look-up table rather than calculating the raw value.
Different buttons correspond to different bits in the Button/Direction Masks. Pressing multiple buttons/directions at the same time will have both bits be set for as many frames as they are held. It also means that not pressing any buttons will make the Button Sum not change.
I can post a dump of the look-up table if any of you want to mathineer your way to some potential improvements, but I think it would be a very uphill battle. I believe it contains all 256 possibilities, so it's just spread out quite a bit. If there's some trend though, you may be able to find something to push it in your favor.
HELLO!
Great stream zallard. Meanwhile I'm still working on catching up to y'all to be able to start learning Narcis. Getting closer...
B+Left, Left, Up+B, ★
I'm really curious about the effects of the button sum, since it sounds like that's all I really have to work with in increasing my odds in certain events. Like for example, are there any combinations of buttons that tend to give the proper luck for the Bear Hugger uppercut? I know it won't be consistent or anything since the RNG sequence happening during most frames, but even slightly better odds for just 1 random event would be great if possible.
Also I uploaded the Narcis Prince tutorial on youtube for everyone's viewing convenience. I think it went pretty well.
Also I uploaded the Narcis Prince tutorial on youtube for everyone's viewing convenience. I think it went pretty well.
All the things
To figure that out would require looking into exactly which pieces of the RNG are used to determine a dodge. There's a pretty decent chance that it will be different for each fighter and mechanic.
Just from what I briefly saw, you will be limited by a couple things. I will try to set this up as simply as I can, but it is still a bit complex.
-The look-up table is the biggest limiting factor in this. You're not shooting for a particular value to fall out of the Button Sum + Frame Counter, but rather a certain frequency of results from the look-up table.
-The easiest thing to do would be to focus on single bits that you know give a better chance at producing an RNG value that you want. For the sake of this example, I will just start with odd or even. Through some fictional extensive analysis, you realize that any odd index will give you a favorable event, and any even indexes will give you unfavorable results. You can force the Button Sum to have the bit set for odd by pressing a given button/direction for exactly one frame, or any odd number of frames.
-The frame counter is a big limit to consistency. Since it changes between odd and even every frame, you're already stuck to no better than 50% consistency.
-You can force the least significant bit to be a consistent odd or even by holding the given direction/button for every frame from a certain point in time; this works because the frame counter will decrement a frame, but the Button Sum will also be toggling the the same bit so the actual odd/evenness will not change. Figuring out which one is set is an entirely different matter; it's 50/50 if you're doing it blindly.
The prime assumption of all of the above is that you've done some variety of frequency analysis to determine that some trends in the RNG values given by patterns in the index will be beneficial if you can hit them. This is not an easy process, and you will absolutely have to know how the RNG values are used to induce the outcomes you want. I can help out on that, but I need to know which specific events you want to dig into.
EDIT: Somewhere in there I forgot to mention that Bear Hugger seemed more likely to be hit by the uppercut when the 5th bit of the Button Sum is set (press right for one frame). I don't know if this actually has any added consistency to it, but that was the main difference between the "good luck" movie and the three "bad luck" movies that Zallard1 provided.
Just from what I briefly saw, you will be limited by a couple things. I will try to set this up as simply as I can, but it is still a bit complex.
-The look-up table is the biggest limiting factor in this. You're not shooting for a particular value to fall out of the Button Sum + Frame Counter, but rather a certain frequency of results from the look-up table.
-The easiest thing to do would be to focus on single bits that you know give a better chance at producing an RNG value that you want. For the sake of this example, I will just start with odd or even. Through some fictional extensive analysis, you realize that any odd index will give you a favorable event, and any even indexes will give you unfavorable results. You can force the Button Sum to have the bit set for odd by pressing a given button/direction for exactly one frame, or any odd number of frames.
-The frame counter is a big limit to consistency. Since it changes between odd and even every frame, you're already stuck to no better than 50% consistency.
-You can force the least significant bit to be a consistent odd or even by holding the given direction/button for every frame from a certain point in time; this works because the frame counter will decrement a frame, but the Button Sum will also be toggling the the same bit so the actual odd/evenness will not change. Figuring out which one is set is an entirely different matter; it's 50/50 if you're doing it blindly.
The prime assumption of all of the above is that you've done some variety of frequency analysis to determine that some trends in the RNG values given by patterns in the index will be beneficial if you can hit them. This is not an easy process, and you will absolutely have to know how the RNG values are used to induce the outcomes you want. I can help out on that, but I need to know which specific events you want to dig into.
EDIT: Somewhere in there I forgot to mention that Bear Hugger seemed more likely to be hit by the uppercut when the 5th bit of the Button Sum is set (press right for one frame). I don't know if this actually has any added consistency to it, but that was the main difference between the "good luck" movie and the three "bad luck" movies that Zallard1 provided.
B+Left, Left, Up+B, ★
Hm, ok, so there's an entire set of values that we want for a single desired event, and we want to figure out what gives us the most likely chances to get any of the values that we want? One thing to note is, the buttons I press are pretty much set and don't vary until I get to the Bear Hugger uppercut (unless I press random junk like in some of the bizhawk videos, or type in a different name).
I could try to make a list of every event and try to provide bizhawk videos for them, but that would take quite a while, since I would be interested in getting events to line up in SS runs (or possibly Narcis Prince IL attempts). Would I be able to make videos of just going into "Time Attack" mode and getting the random event to happen (or not happen), or would it have to be done on a new file for every video?
I could try to make a list of every event and try to provide bizhawk videos for them, but that would take quite a while, since I would be interested in getting events to line up in SS runs (or possibly Narcis Prince IL attempts). Would I be able to make videos of just going into "Time Attack" mode and getting the random event to happen (or not happen), or would it have to be done on a new file for every video?
B+Left, Left, Up+B, ★
I've been slacking on the movies because I'm having second thoughts on this just because it seems like it will end up being more work than it's worth sadly. However, I'm planning on getting Narcis Prince Day going somewhat soon depending on when everyone is free. If you happen to be in the PM thread, it'd be great if you could update me on what time you're free. If you want to get added to the thread, let me know soon!
Lately I've been putting a lot of time into Narcis Prince using the old strat, and made some progress. I've completed it twice so far. The first time was 8"98 because he danced at the beginning of the 2nd phase. I actually beat the current time with 8"52 on the second completion. It had perfect RNG but the execution was not up to par. They're both on YouTube:
.
I may try out the current TAS strat depending on my luck using this one. Hopefully we can decide a good time to have Narcis Prince Day sometime soon!
.
I may try out the current TAS strat depending on my luck using this one. Hopefully we can decide a good time to have Narcis Prince Day sometime soon!
B+Left, Left, Up+B, ★
Sick time on Narcis! The IL table is now only 0.09 away from the current TAS (5 frames)!
So I might as well mention here that the current TAS strat where there's a frame perfect punch might actually be feasible in real time after all, thanks to some research by Omnigamer. I'll put together a video on how to time the frame perfect hit, as well as an updated buffer strat (hopefully) for when Narcis enters the ring from the back (timing is different than before due to the faster recoil of the FP hit).
So I might as well mention here that the current TAS strat where there's a frame perfect punch might actually be feasible in real time after all, thanks to some research by Omnigamer. I'll put together a video on how to time the frame perfect hit, as well as an updated buffer strat (hopefully) for when Narcis enters the ring from the back (timing is different than before due to the faster recoil of the FP hit).
B+Left, Left, Up+B, ★
Today, Akiteru and I will go after the Narcis Prince IL. I'm unsure if anyone else is gonna join, so if you are, just post your twitch in this thread when you start.
http://www.twitch.tv/zallard1
http://www.twitch.tv/zallard1
B+Left, Left, Up+B, ★
Dang I just saw these posts... thanks akiteru and doicm!!! I'm so happy that run went as well as it did. I did a breakdown of how the AGDQ run went on my twitch page a few days ago. Also McHazard will be submitting his new TAS of Super Punch Out sometime this week (I hope). He says there are 4 improvements that save in total a little less than a second. We'll have to revamp the IL table once it's released!
Oh btw, new Super Punch Out WR of 2:53.66. Garbage Major Circuit got bailed out by the god World Circuit and Narcis/Hoy cooperating somehow. Lost most of the lead on Nick because I got a bad pattern. I absolutely have to improve the run further now...
Oh btw, new Super Punch Out WR of 2:53.66. Garbage Major Circuit got bailed out by the god World Circuit and Narcis/Hoy cooperating somehow. Lost most of the lead on Nick because I got a bad pattern. I absolutely have to improve the run further now...