Secret of Evermore (SNES)

Edit history:

Grumly: 2014-01-12 05:51:46 pm

2014-01-12 05:39:14 pm

64 is an example I took for a better understanding of what happened when I change the number of this variable, using Action Replay (64 in hexadecimal means 100). In the memory of the game, without intervention, the value change when you cast a spell but i don't know precisely yet how this number is set. If we can know how this number is set, maybe it can be possible to manipulate this number and do extra damage.

It could be quite useful for TAS but maybe for speedrun as well if it is possible to found some setups.

There is no doubt that Thraxx has got a special pattern for calculating damage since I tried to launch HB with a fixed power of 64 at Magmar : the result is ALWAYS 35 damage. No another RNG on this boss ! For Magmar, the value of the variable 7E358E fixes sharply the amount of HB damage.

Here is the link :

TheAngryPanda

Edit history:

TheAngryPanda: 2014-01-12 06:49:24 pm

TheAngryPanda: 2014-01-12 06:18:16 pm

TheAngryPanda: 2014-01-12 06:06:19 pm

TheAngryPanda: 2014-01-12 05:54:28 pm

TheAngryPanda: 2014-01-12 05:52:34 pm

2014-01-12 05:51:55 pm

Speed > Safety

Yes, that's what I'm trying to find is the relationship between damage and other factors. Currently, these are my two known influencing factors on the formula when the value is the same:

positional data on the non-player controlled character
whether or not the R/L buttons are pressed, which is a player-input RNG modifier for the game

These two are the only things changing when the damage changes on the same frame with differing data. I'm currently testing multiple 8x at once and getting some strange effects, though it's likely due to Thraxx damage being weird.

I have a value of 18 at positional data x = 132, y = 244, when I do one cast without doing anything else at all from this position, I get the value 18 and 27 damage. Doing a 2nd cast on the very first frame possible afterwards (which is possible in real-time) yields another 18 value because the positional data and other factors aren't changing. You can actually let go of R and the game won't recognize that you have because the ring menu is brought up so rapidly. This gives me 41 damage. The 3rd cast using this gives me 88.

so 27, 41, 88. let's say 27 was the 100% value for 18. 41-54 gives us -13, this is a loss of around 50% damage. 88-81 gives us 6, which is an increase in damage by a relatively low %.

If this trend was to continue, perhaps there is some logic to it. Or maybe the Thraxx damage is calculated upon being hit and re-rolling the damage very rapidly since it would be hit every other frame after an 8x cast. Finding the value/address of this factor is what will break the Thraxx fight.

I'm looking forward to playing with this on things other than Thraxx, by the way. It's just that Thraxx is the single most important part of the TAS currently, as it could mean gaining around 500 or so frames if I can find a way to make all of it work together properly. The numbers above are from very brief testing results, nobody should be viewing them as solid data... it's just something I've noticed. Smiley

edit: ~~video's being processed, I'll check it out as soon as I can.~~
edit2: Alright, that's actually pretty cool that Magmar's is static. Good to know, I can definitely use that information later.

edit3: Okay, so looking back on the formula(s) Nitrodon's provided us. It seems the minimum damage for Hard Ball at level 0 is 11 and the maximum is 20. So what I'm wondering is if this value we're messing with is actually THAT value instead. Which would indicate that Thraxx damage on the heart is based on a % of this value and that each hit is, in fact, determined on frame of contact as opposed to cast by the player.

Nitrodon/Assassin17, if either of you are around, any ideas/theories for this that may help figure out Thraxx heart damage formula?

edit4: So the story develops more it seems.

At value 18, I obtained 28 damage
At value 14, I obtained 34 damage

There were cast in succession, perhaps there is a cumulative damage factor going on with the heart formula? More testing....

edit5: Continuing with Thraxx progression testing:

-Utilizing the dog to inflict damage on the cage prior to damaging the heart via casts definitely shows an increased damage amount by some degree. The variance isn't leading me to any major discoveries though towards a distinctive formula. The possibility that Thraxx damage is a cumulative % based increase is starting to really stack up. This, however, does not explain why without the cutscene damage is reduced significantly to the point of me having almost no possibility of killing this boss. Perhaps number of objects present on screen leads to another factor within the fight itself? Currently there are both arms, boy, dog, cage, #HB, Heart on screen, as opposed to Heart, both arms, boy, dog, 4 maggots, #HB, and cage.

Right now my current plan is to manipulate a high-damage factor from various values at various positions via manipulation on the dog and/or boy AI behavior. I also figured out I can open the menu to 8x cast and attack on the exact same frame, so I guess that's a 1 frame time-saver regardless of anything else for a few places.

So this brings our list of potential things that influence heart damage to:

-Dog action (sniffing, not sniffing, attacking, idling)
-AI behavior (and movement, which is complex but able to be manipulated pretty easily)
-This value Grumly's posted about
-The distance between Thraxx Heart and Boy (frames wasted during distance traveled = more RNG in between contact of cast, not easily manipulated due to various factors)
-(theory) # of maggots on screen seems to have an effect, unsure of exactly how though
-(theory) Damage is showing to be cumulative in % based on damage incurred by other means previous to itself. Perhaps this is key in designing a formula?
-(theory) The current alchemy formula can be used to figure out initial damage, but not Thraxx damage reliably. Maybe they're related as a tangible factor tacked on during the fight? Not sure how this would be implemented or anything like that, but there's clearly a relationship there between the two. I don't see the % modifier being static either on Thraxx, but I also have no way of proving otherwise currently since testing isn't completely thorough due to other factors not being confirmed for involvement or not. Sad

And yes, I do think about this crap too much for my own good.

Yes, that's what I'm trying to find is the relationship between damage and other factors. Currently, these are my two known influencing factors on the formula when the value is the same:

positional data on the non-player controlled character
whether or not the R/L buttons are pressed, which is a player-input RNG modifier for the game

These two are the only things changing when the damage changes on the same frame with differing data. I'm currently testing multiple 8x at once and getting some strange effects, though it's likely due to Thraxx damage being weird.

I have a value of 18 at positional data x = 132, y = 244, when I do one cast without doing anything else at all from this position, I get the value 18 and 27 damage. Doing a 2nd cast on the very first frame possible afterwards (which is possible in real-time) yields another 18 value because the positional data and other factors aren't changing. You can actually let go of R and the game won't recognize that you have because the ring menu is brought up so rapidly. This gives me 41 damage. The 3rd cast using this gives me 88.

so 27, 41, 88. let's say 27 was the 100% value for 18. 41-54 gives us -13, this is a loss of around 50% damage. 88-81 gives us 6, which is an increase in damage by a relatively low %.

If this trend was to continue, perhaps there is some logic to it. Or maybe the Thraxx damage is calculated upon being hit and re-rolling the damage very rapidly since it would be hit every other frame after an 8x cast. Finding the value/address of this factor is what will break the Thraxx fight.

I'm looking forward to playing with this on things other than Thraxx, by the way. It's just that Thraxx is the single most important part of the TAS currently, as it could mean gaining around 500 or so frames if I can find a way to make all of it work together properly. The numbers above are from very brief testing results, nobody should be viewing them as solid data... it's just something I've noticed. :)

edit: [s]video's being processed, I'll check it out as soon as I can.[/s]
edit2: Alright, that's actually pretty cool that Magmar's is static. Good to know, I can definitely use that information later.

edit3: Okay, so looking back on the formula(s) Nitrodon's provided us. It seems the minimum damage for Hard Ball at level 0 is 11 and the maximum is 20. So what I'm wondering is if this value we're messing with is actually THAT value instead. Which would indicate that Thraxx damage on the heart is based on a % of this value and that each hit is, in fact, determined on frame of contact as opposed to cast by the player.

Nitrodon/Assassin17, if either of you are around, any ideas/theories for this that may help figure out Thraxx heart damage formula?

edit4: So the story develops more it seems.

At value 18, I obtained 28 damage
At value 14, I obtained 34 damage

There were cast in succession, perhaps there is a cumulative damage factor going on with the heart formula? More testing....

edit5: Continuing with Thraxx progression testing:

-Utilizing the dog to inflict damage on the cage prior to damaging the heart via casts definitely shows an increased damage amount by some degree. The variance isn't leading me to any major discoveries though towards a distinctive formula. The possibility that Thraxx damage is a cumulative % based increase is starting to really stack up. This, however, does not explain why without the cutscene damage is reduced significantly to the point of me having almost no possibility of killing this boss. Perhaps number of objects present on screen leads to another factor within the fight itself? Currently there are both arms, boy, dog, cage, #HB, Heart on screen, as opposed to Heart, both arms, boy, dog, 4 maggots, #HB, and cage.

Right now my current plan is to manipulate a high-damage factor from various values at various positions via manipulation on the dog and/or boy AI behavior. I also figured out I can open the menu to 8x cast and attack on the exact same frame, so I guess that's a 1 frame time-saver regardless of anything else for a few places.

So this brings our list of potential things that influence heart damage to:

-Dog action (sniffing, not sniffing, attacking, idling)
-AI behavior (and movement, which is complex but able to be manipulated pretty easily)
-This value Grumly's posted about
-The distance between Thraxx Heart and Boy (frames wasted during distance traveled = more RNG in between contact of cast, not easily manipulated due to various factors)
-(theory) # of maggots on screen seems to have an effect, unsure of exactly how though
-(theory) Damage is showing to be cumulative in % based on damage incurred by other means previous to itself. Perhaps this is key in designing a formula?
-(theory) The current alchemy formula can be used to figure out initial damage, but not Thraxx damage reliably. Maybe they're related as a tangible factor tacked on during the fight? Not sure how this would be implemented or anything like that, but there's clearly a relationship there between the two. I don't see the % modifier being static either on Thraxx, but I also have no way of proving otherwise currently since testing isn't completely thorough due to other factors not being confirmed for involvement or not. :(

And yes, I do think about this crap too much for my own good.

LovableLambchop

2014-01-12 08:02:56 pm

My head hurts after reading all this.

Amazing work though Grumly!

TheAngryPanda

Edit history:

TheAngryPanda: 2014-01-12 08:53:45 pm

2014-01-12 08:47:30 pm

Speed > Safety

Okay, that's enough for tonight...

I've done about 20 or so different formula variations in a rather sad attempt to figure out the way everything is related so I can manipulate damage properly. Sadly, no major results. Will resume torturing myself with this tomorrow I guess. On the bright side, it seems like Thraxx (and possibly Thraxx 2.0?) are the only major issues regarding this sort of thing, every other enemy seems to have static damage flow instead of a variable thrown in.

I think I can make cutscene skip work now... maybe... really hope so.

nayt

2014-01-12 09:14:05 pm

This is all really exciting stuff to read! I didn't know Evermore had a speedrun group.

It's one of my favorite games so I may have to give it a shot after I work on LttP.

TheAngryPanda

2014-01-12 09:34:36 pm

Speed > Safety

Unfortunately most of the stuff in the past few pages more or less only applies to the TAS of the game, which I'm pretty heavily focused on. That being said, I do most of the routing for the game as well so if you have any questions or anything, feel free to PM me on here any time and I'll happily attempt to help out. There are also some guides floating around I made, slightly out of date, but should give you a general make-up of the routes and how they go through the game.

General FAQ (not a short read, unfortunately):
http://pastebin.com/i88Affqj

Excel doc with routes (Crush WIP = Any%):
https://www.dropbox.com/s/3tngl51cd0rolxf/Secret-of-Evermore.xlsx

PDF of Any%:
https://www.dropbox.com/s/7msd574z0si7cho/Secret-of-Evermore-expert%27s-route.pdf

PDF of 100%: (basically a new route)
https://www.dropbox.com/s/hkkzhp4xepcov4c/Secret-of-Evermore-100%25-route.pdf

For the PDF's, it's best to just download them to view them. The scaling on dropbox is total crap. Your absolute best option, however, would be the excel document, because you can adjust things and have access to the alchemy damage calculator, monster stats, and various other useful things that pdf simply cannot accomodate. I don't update them much anymore because runs have gotten stale and most changes don't merit any major updates to the files. I believe Metasigma is learning a 100% + All Bosses route, LovableLambchop is planning to run 100%, and I'm doing the TAS. Most other runners have gone to different games lately.

Unfortunately most of the stuff in the past few pages more or less only applies to the TAS of the game, which I'm pretty heavily focused on. That being said, I do most of the routing for the game as well so if you have any questions or anything, feel free to PM me on here any time and I'll happily attempt to help out. There are also some guides floating around I made, slightly out of date, but should give you a general make-up of the routes and how they go through the game.

General FAQ (not a short read, unfortunately):
http://pastebin.com/i88Affqj

Excel doc with routes (Crush WIP = Any%):
https://www.dropbox.com/s/3tngl51cd0rolxf/Secret-of-Evermore.xlsx

PDF of Any%:
https://www.dropbox.com/s/7msd574z0si7cho/Secret-of-Evermore-expert%27s-route.pdf

PDF of 100%: (basically a new route)
https://www.dropbox.com/s/hkkzhp4xepcov4c/Secret-of-Evermore-100%25-route.pdf

For the PDF's, it's best to just download them to view them. The scaling on dropbox is total crap. Your absolute best option, however, would be the excel document, because you can adjust things and have access to the alchemy damage calculator, monster stats, and various other useful things that pdf simply cannot accomodate. I don't update them much anymore because runs have gotten stale and most changes don't merit any major updates to the files. I believe Metasigma is learning a 100% + All Bosses route, LovableLambchop is planning to run 100%, and I'm doing the TAS. Most other runners have gone to different games lately.

DLDarklink

2014-01-13 01:43:23 am

...

I'm still practicing SoE without streaming Tongue

I'm trying to get at least sub 1:20 ^^
Also, a player/viewer named Jelsa asked me some stuff about the PAL run. He want's to run it (in the future) but he hasn't been able to obtain a cheap cart Sad

Our conversation actually sparked my interest in the PAL version (dafuq? :P). After I get a sub 1:20 on the NTSC version I'll try to improve that from time to time and switch to PAL and some Terranigma runs.

This week will be Evermore week for me with only a little bit of Terra Smiley

Many thanks to GA for the buffered setup (getting the dog at blimps cave) and Crawek for pointing that out Cheesy

TheAngryPanda

2014-01-13 06:16:11 am

Speed > Safety

Nice to hear, this time can we get a recorded attempt so we can update SDA's listing? Also, if you can PM me or talk on skype sometime, maybe we can even get together and develop a guide like the above ones for PAL categories? That would probably help quite a bit in regards to optimizing the run and making it more presentable for submission I'd think. Although, I'm sure you've already got it pretty well-rounded out.

If Terranigma gets broken as bad as I've heard it might, I may have to go hop on that train for a little bit as well, too tempting.

That being said, back to working on this Thraxx fight! Cheesy

TheAngryPanda

Edit history:

TheAngryPanda: 2014-01-13 08:33:46 am

TheAngryPanda: 2014-01-13 08:27:47 am

2014-01-13 08:20:49 am

Speed > Safety

I did it! Thraxx is down, time to move on, unless something horrible happens because of cutscene skip! Cheesy

756 frames saved. Fight's awkward looking due to RNG manipulations and just being a shit boss though, not happy about that. On to Magmar. Smiley

edit: And I have 4,XXX,XXX,XXX exp needed to level. Face = floored.
edit2: It's just a glitch with the display because the maggots aren't present during the fight to be auto-killed or something. Killing anything (mosquito for testing) gives me an instant level up.

CrAwEk

2014-01-13 07:14:34 pm

Secret of Evermore PB & WR 1:18:20 HYPE Smiley

http://www.twitch.tv/crawek/c/3542853

Patashu

2014-01-13 07:36:43 pm

Congrats, that is damn fast!

TheAngryPanda

2014-01-13 08:13:39 pm

Speed > Safety

And that's without BBMSkip. Soon as he gets BBMSkip in there, it'll be sub 18 no problem. Smiley

assassin17

2014-01-13 09:57:31 pm

I love YaBB 1G - SP1!

Quote:

edit: And I have 4,XXX,XXX,XXX exp needed to level. Face = floored.
edit2: It's just a glitch with the display because the maggots aren't present during the fight to be auto-killed or something. Killing anything (mosquito for testing) gives me an instant level up.

what level were your characters going into the fight?

TheAngryPanda

2014-01-14 01:57:11 am

Speed > Safety

It a known bug, apparently.

level 1, the Thraxx kill itself puts them at 2, 1st mosquito kill puts them at 3, I'd presume the 2nd mosquito kill will leave them at lvl 4.

AsFarAsIGet

Edit history:

AsFarAsIGet: 2014-01-14 10:31:54 am

2014-01-14 10:30:32 am

www.youtube.com/user/HumanMYTH

Hey Guys heres an RTA thought:

So, I might have found a way to increase the act1 flowers hitrate by simply standing in front of one and repeatedly press L, R, L, R. (The Dog should be sniffin, dont sniffin, sniffin, dont sniffin - he shouldNOT sniff all the time) Would be nice if u can test it and let me know if it works or if its just my imagination. It seems that the sniffing triggers the attacks way faster! Cheesy

greetings
AFAIG

p.S.: You can also do it with just L or R, but u have to mash way faster Smiley

TheAngryPanda

Edit history:

TheAngryPanda: 2014-01-14 10:51:28 am

TheAngryPanda: 2014-01-14 10:48:11 am

2014-01-14 10:46:45 am

Speed > Safety

This doesn't increase the hit rate.

What this does is roll the RNG again to override what you already have.

Here's what happens (roughly):

-No input scenario, let's say this is RNG case 0:

Flower damage = 6-10
Flower chance to hit = 40% (this is just a number I'm picking, don't read into it too much, it's just an example)
Boy chance to evade = 40% (again, just an example)
Dog Sniffing = off
Dog Attacking = off
Boy Attacking = off
Positional Data goes here (This has an effect, but we can't control this either since it rolls RNG every frame, Thraxx's room is an exception)

So that case will result in 6-10 damage.

Here's RNG Case 1, with dog sniffing:

Flower damage = 6-10
Flower chance to hit = 40%
Boy chance to evade = 40%
Dog Sniffing = on
Dog Attacking = off
Boy Attacking = off
Positional Data goes here.

So this results in 6-10 damage. The only difference is that the dog is sniffing.

RNG case 2, dog sniffing + boy attacking:

Flower damage = 6-10
Flower chance to hit = 40%
Boy chance to evade = 40%
Dog Sniffing = on
Dog Attacking = off
Boy Attacking = on
Positional Data goes here.

Again, we still achieve only 6-10 damage, with no change in hit or evade %.

It's unfortunately nothing but placebo. It's nothing but a chance, and all you do is roll the chance again. It's kind of like the poison thing with spiders, it's a 25% chance, but you're not going to always see 1 out of 4 hits.

This really sucks for RTA because while I do know how to manipulate damage, hit rate, etc... even spell damage now thanks to Grumly. There's no way I can provide a 100% consistent method of achieving X goal on an RTA attempt, because this game's RNG actually does what it's supposed to. Sad

edit: The only thing I've noticed out of any and all testing I've done regarding flowers is that when the boy attacks the flower as the flower hits the boy, I see higher damage than lower damage numbers 8-10, opposed to 6-8, 8 obviously is the middle ground. Personally, I think this is just a coincidence and nothing more than that, but I suppose it's worth mentioning.

This doesn't increase the hit rate.

What this does is roll the RNG again to override what you already have.

Here's what happens (roughly):

-No input scenario, let's say this is RNG case 0:

Flower damage = 6-10
Flower chance to hit = 40% (this is just a number I'm picking, don't read into it too much, it's just an example)
Boy chance to evade = 40% (again, just an example)
Dog Sniffing = off
Dog Attacking = off
Boy Attacking = off
Positional Data goes here (This has an effect, but we can't control this either since it rolls RNG every frame, Thraxx's room is an exception)

So that case will result in 6-10 damage.

Here's RNG Case 1, with dog sniffing:

Flower damage = 6-10
Flower chance to hit = 40% 
Boy chance to evade = 40% 
Dog Sniffing = on
Dog Attacking = off
Boy Attacking = off
Positional Data goes here.

So this results in 6-10 damage. The only difference is that the dog is sniffing.

RNG case 2, dog sniffing + boy attacking:

Flower damage = 6-10
Flower chance to hit = 40%
Boy chance to evade = 40%
Dog Sniffing = on
Dog Attacking = off
Boy Attacking = on
Positional Data goes here.

Again, we still achieve only 6-10 damage, with no change in hit or evade %.

It's unfortunately nothing but placebo. It's nothing but a chance, and all you do is roll the chance again. It's kind of like the poison thing with spiders, it's a 25% chance, but you're not going to always see 1 out of 4 hits.

This really sucks for RTA because while I do know how to manipulate damage, hit rate, etc... even spell damage now thanks to Grumly. There's no way I can provide a 100% consistent method of achieving X goal on an RTA attempt, because this game's RNG actually does what it's supposed to. :(

edit: The only thing I've noticed out of any and all testing I've done regarding flowers is that when the boy attacks the flower as the flower hits the boy, I see higher damage than lower damage numbers 8-10, opposed to 6-8, 8 obviously is the middle ground. Personally, I think this is just a coincidence and nothing more than that, but I suppose it's worth mentioning.

AsFarAsIGet

Edit history:

AsFarAsIGet: 2014-01-14 10:57:43 am

2014-01-14 10:56:33 am

www.youtube.com/user/HumanMYTH

Oh, I thouhgt that the flowers attack more often/frequently. But nice that u checked those things aswell.
Might try to test this with WSplit, but im not motivated right now.

TheAngryPanda

2014-01-14 11:18:13 am

Speed > Safety

Hmm... Here's my notes on it from my notebook for TAS'ing and testing routes:

Flower hits out of 10 attacks:
0 = No Hit
1 = Hit

Test 1 (Control Run) - no movement, no sniff, use code to place dog out of bounds to prevent positional data changing

0110000111

Total hits = 5

Test 2 (Sniffing)

1011001110

Total hits = 6

-did not affect attack rate
-did not alternate from sniffing > not sniffing
-positional data did not change either
-same savestate as well as starting frame

Test 3 (Boy Attacking as Flower Attacks)

1111001101

Total hits = 7

-possibly attacking reduces evade% on boy, test later (I never did...)
-facing different directions gave different damage, perhaps due to changing positional data during the initial frame of the attack OR it rolls flower hp again which gives a new roll on dodge/hit?
-most promising test so far

Test 4 (Alternate Sniffing)

0000001111

Total hits = 4

-lower chances? Seems unlikely, probably due to rolling RNG over and returning bad results
-dog position never changes, maybe if this were the case we'd have better luck?

Test 5 (Alternate Sniffing - starting 1 frame later than previously)

1110101000

Total hits = 5

-based on the tests, seems to just roll rng and return new result

So that's what's actually in my notes for this. I also did testing without locking positional data down and got similar results. Nothing shows consistent increases in damage range, dodge rate, evade rate, or AI's chance to attack. The way the TAS manipulates these is by careful movement and immediately going to the last flower I need to get hit by and knowing all the frames it will hit me for 10 on then managing the frames leading up to that very carefully. That's why I "lose" 3 frames on the 2nd flower in the TAS, so I can save 22 later, which is never noticed by anyone unless they knew they were looking for it.

It would be nice at some point to be able to lock down an AI patterns at least for flowers, but I don't see it happening anytime in the very near future. Not without a way to KNOW what RNG we have initially. IF we can figure out how to have that knowledge, I could probably figure out a way to get the first flower to hit for 10 every single time, player input varies too much after that to provide more setups. Usually when the player turns the corner leading to 2nd flower.

Hmm... Here's my notes on it from my notebook for TAS'ing and testing routes:

Flower hits out of 10 attacks:
0 = No Hit
1 = Hit

Test 1 (Control Run) - no movement, no sniff, use code to place dog out of bounds to prevent positional data changing

0110000111

Total hits = 5

Test 2 (Sniffing)

1011001110

Total hits = 6

-did not affect attack rate
-did not alternate from sniffing > not sniffing
-positional data did not change either
-same savestate as well as starting frame

Test 3 (Boy Attacking as Flower Attacks)

1111001101

Total hits = 7

-possibly attacking reduces evade% on boy, test later (I never did...)
-facing different directions gave different damage, perhaps due to changing positional data during the initial frame of the attack OR it rolls flower hp again which gives a new roll on dodge/hit?
-most promising test so far

Test 4 (Alternate Sniffing)

0000001111

Total hits = 4

-lower chances? Seems unlikely, probably due to rolling RNG over and returning bad results
-dog position never changes, maybe if this were the case we'd have better luck?

Test 5 (Alternate Sniffing - starting 1 frame later than previously)

1110101000

Total hits = 5

-based on the tests, seems to just roll rng and return new result

So that's what's actually in my notes for this. I also did testing without locking positional data down and got similar results. Nothing shows consistent increases in damage range, dodge rate, evade rate, or AI's chance to attack. The way the TAS manipulates these is by careful movement and immediately going to the last flower I need to get hit by and knowing all the frames it will hit me for 10 on then managing the frames leading up to that very carefully. That's why I "lose" 3 frames on the 2nd flower in the TAS, so I can save 22 later, which is never noticed by anyone unless they knew they were looking for it.

It would be nice at some point to be able to lock down an AI patterns at least for flowers, but I don't see it happening anytime in the very near future. Not without a way to KNOW what RNG we have initially. IF we can figure out how to have that knowledge, I could probably figure out a way to get the first flower to hit for 10 every single time, player input varies too much after that to provide more setups. Usually when the player turns the corner leading to 2nd flower.

AsFarAsIGet

2014-01-14 11:24:27 am

www.youtube.com/user/HumanMYTH

Wow gj panda and thanks for the fast results! Wink

DLDarklink

2014-01-14 03:02:51 pm

...

I did a guard skip with the dog today, but it was kinda different:

http://www.twitch.tv/dldarklink/c/3549265

AsFarAsIGet

2014-01-14 03:17:46 pm

www.youtube.com/user/HumanMYTH

Thats interesting darklink, moving during textboxes could bring some nice things with it! We should try to reproduce it and mb gain some advantages out of it! Smiley

TheAngryPanda

Edit history:

TheAngryPanda: 2014-01-14 03:52:41 pm

2014-01-14 03:49:58 pm

Speed > Safety

The game will crash if you do it with the boy or dog during text boxes IF they're present while the screen fades out. I was tossing this into the TAS a while back on stream and talking about it quite a bit but I think the VOD has been long since deleted.

My observation was that it's a frame perfect switch as you hit the text box event trigger, which allows you to move as the other character. In RTA, you can simply wait out the text box and enjoy your silly skip, the TAS lost tons of frames due to having to wait. The game "finishes" the delay on the text box and auto-closes it, then allows progression to continue. However, in the TAS when I thought it'd be awesome to do this with the boy (or dog it seems...) the text never actually advanced, thus it was softlocked because I couldn't hit a button to progress the text.

There's a few other places you can do this with some weird results, as well as the frame perfect character switch at events that cause movement, as GA found a while back that lets you walk out of bounds.

Lots of fun stuff in this game with frame perfect nonsense. Cheesy

edit: re: moving during textboxes. You will still hit "triggers" at places, to date there's been no major discovery revolving around this but I'm certainly not going to discount viability of this trick. It's also how cutscene skips work, which are frame perfect as well. Act 2 and 3 should have something in there that might be exploitable, but anywhere I can think of yields nothing but it being an interesting trick as opposed to useful. Sad

The game will crash if you do it with the boy or dog during text boxes IF they're present while the screen fades out. I was tossing this into the TAS a while back on stream and talking about it quite a bit but I think the VOD has been long since deleted.

My observation was that it's a frame perfect switch as you hit the text box event trigger, which allows you to move as the other character. In RTA, you can simply wait out the text box and enjoy your silly skip, the TAS lost tons of frames due to having to wait. The game "finishes" the delay on the text box and auto-closes it, then allows progression to continue. However, in the TAS when I thought it'd be awesome to do this with the boy (or dog it seems...) the text never actually advanced, thus it was softlocked because I couldn't hit a button to progress the text.

There's a few other places you can do this with some weird results, as well as the frame perfect character switch at events that cause movement, as GA found a while back that lets you walk out of bounds.

Lots of fun stuff in this game with frame perfect nonsense. :D

edit: re: moving during textboxes. You will still hit "triggers" at places, to date there's been no major discovery revolving around this but I'm certainly not going to discount viability of this trick. It's also how cutscene skips work, which are frame perfect as well. Act 2 and 3 should have something in there that might be exploitable, but anywhere I can think of yields nothing but it being an interesting trick as opposed to useful. :(

TheAngryPanda

Edit history:

TheAngryPanda: 2014-01-15 10:51:51 am

2014-01-15 10:51:16 am

Speed > Safety

So then this happened... http://www.twitch.tv/theangrypanda1/c/3553566

I figured out how to make it work, pretty much saves around 93 frames.

DLDarklink

2014-01-15 12:22:34 pm

...

I've found a way to bypass that guard with the menu buffer Smiley

from a certain position you do the menu buffer while aproaching that guard, on the third frame (? i can't count so it might be 2 or 4) you switch to dog and you'll trigger the 'walking while talking' stuff. I think it's the same position where you'll 'park' the boy during a run so the dog can walk past the guard Tongue

I know it's slow as f*ck but just wanted to point it out Tongue

TheAngryPanda

2014-01-15 12:31:34 pm

Speed > Safety

it's a frame perfect character switch, i'm pretty sure of it. It's definitely slower in the TAS, might be quicker in real-time IF you could do it w/o buffering the inputs via menu nonsense and had a good idea of where to stop so the text doesn't become an issue like it does in the TAS.

That being said, you're looking at an extremely minor time save, if it even saves time. I guess Horace Skip has a new friend? Smiley

lol.