Maybe so we can figure out how you did certain tricks without reverse-engineering it from a two and a half year old video? Seems pretty obvious to me why we would want it.
Everything is clearly noticeable in the videos provided, and the inputs are obvious. Besides, as mentioned before the RNG will be different due to another ROM, so you're on your own when it matters (which is during most enemy/boss encounters) anyway.
Regarding ROM choice, there is no official English NTSC version for this game, so one has to stick with PAL when that language has priority. Since a TAS is for entertainment purposes and not a speedrun competition for realtime, I didn't see a problem with it.
Everything is clearly noticeable in the videos provided, and the inputs are obvious. Besides, as mentioned before the RNG will be different due to another ROM, so you're on your own when it matters (which is during most enemy/boss encounters) anyway.
K is this input file some rare resource that you're afraid will end the world when you dole it out? Clearly the inputs aren't obvious because multiple people are asking for the input file. Can you please just post it so we can move on to more important matters?
Saturn wrote:
Regarding ROM choice, there is no official English NTSC version for this game, so one has to stick with PAL when that language has priority.
Language hasn't had priority since like 2008. The idea is you do it on the fastest version, and if you're obsoleting a movie that was on a different version there has to be SOME sort of new optimization to the game besides the new version choice.
Saturn wrote:
Since a TAS is for entertainment purposes and not a speedrun competition for realtime, I didn't see a problem with it.
A TAS is for entertainment purposes AND speedrun competition. Plus, slower text and gameplay isn't very entertaining, unless you live on a different planet.
effort on the first draft means less effort on any draft thereafter
- some loser
Everything is clearly noticeable in the videos provided, and the inputs are obvious.
This is quite simply not the case. It wasn't the case in Secret of Evermore either, though thankfully that game was rapidly improving at the time so any real issues were resolved in the process due to rerouting and better optimizations along the way by multiple different people.
The problem with Terranigma is that the game is kind of already at that heavily optimized point, and is likely not to see any major breaking occur anytime in the very near future even though it would be heavily welcomed. So with this information stated, I'll leave my own personal reasons I'd love to see an input file:
Reason 1:
Optimizations. These are the bread and butter of TAS'ing, they are what takes a mediocre TAS to an amazing one once you remove all the fancy tricks. These are my personal interest in TAS'ing, as evident in the Secret of Evermore Any% TAS I did. You should know very well just how much movement/pathing optimization goes into that kind of game, and Terranigma is within those same sort of game styles. That being said, I am not nearly as familiar with Terranigma as I am with a game that I rerouted, ran, and developed research/guides for. Having input files to look at significantly improves my overall performance when it comes to TAS'ing this game, and it shows in the progression we've had thus far with it. Without Samsara's movement optimizations and critique of my own pathing, I'd be behind my current frame count by about 80-90 frames. The reason yours would be nice is because it's completed, it's something tangible that I can say "On this map, I'm 5 frames behind, what does he does that I don't and how can I learn to fix this in the future maps." Once that's done, I can further compare and challenge myself based upon your pathing and other decisions.
Reason 2:
A video on youtube is not the same quality as the recording/input file/bk2/whatever that is considered a "source" file. Because of this, we're limited to youtube's controls for movie playback (which are shit, and everyone knows that) when trying to do research on various aspects of the game. Case in point, back in the day, I had to ask on here (in this very thread even) what the hell was happening with the Elder jump in the beginning because trying to figure it out from the video w/o inputs was next to impossible at the time. Had someone not come along and mentioned the specific inputs for it, I'd have continued to brute force them until something came of it. That's inefficient, silly, and unnecessary. It's also just one example out of probably hundreds, all of which could be very easily answered with some kind of source file. I upload mine all the time, happily, because if someone can do it better, I want to see how so I can replicate and hopefully improve the process even further.
Reason 3:
The inputs are not obvious. They may be obvious when running in straight lines, diagonally, or basic things like that, but that doesn't overly matter. The stuff that matters like Elder jump (to recycle the example earlier) is next to impossible to decipher with just some random video on youtube that doesn't explain anything and just says "Hey guys! This is a thing! Yay!" This doesn't do anything for anyone in the future who wants to improve the current times/TAS on any site, all it does is frustrate them because the person ahead of them didn't have the foresight to properly document anything and/or provide a file of some sort that could explain things for them. This is basic stuff in the real world, if something is important to the future, you document it and you keep that shit around for a good long while.
Reason 4:
It would be nice to be able to credit the person who provided a significant amount of pathing/movement through many of the areas that was already optimized and well-tested. I cannot do this without some kind of tangible data to go along with it because I'm having to guess and speculate constantly at every turn, jump, RNG scenario, and multiple other situations in every area repeatedly. Then I upload my files, have Vaan, Samsara, and anyone else interested check them out for me. They provide feedback, I change things to address their concerns, and we leave with a pretty damn amazing map/area. Most of this could be cut out if we already had a file to compare against/adjust/mess around with/etc... because it removes the initial step and immediately goes into the review and optimize steps. This would drastically reduce the amount of time a GOOD TAS takes to create. It's why things like the SM64 TAS are so insane and amazing to watch, multiple people pour effort into it and the results show through very clearly.
I honestly could go on and on about it with more and more reasons, but there's not really a need for that. If you don't wish to upload an input file or whatever, that's fine and I understand your decision. But saying that things are clearly noticeable and obvious is just outright wrong. There are things which are noticeable/obvious, but they usually aren't anything of significance.
And I don't mean for this to be a degrading or asshole-ish post, these are simply the reasons and ways I would use an input file if I had one, nothing more. That's all I'm going to say on the matter as well because I'd rather put my focus towards something more productive.
And I don't mean for this to be a degrading or asshole-ish post, these are simply the reasons and ways I would use an input file if I had one, nothing more. That's all I'm going to say on the matter as well because I'd rather put my focus towards something more productive.
Let's be frank. Saturn has decided that he doesn't want to work with TASVideos because we didn't publish one of his Super Metroid runs once. So he's made it a goal to troll us. Please stop feeding him.
Build a man a fire, warm him for a day,
Set a man on fire, warm him for the rest of his life.
And I don't mean for this to be a degrading or asshole-ish post, these are simply the reasons and ways I would use an input file if I had one, nothing more. That's all I'm going to say on the matter as well because I'd rather put my focus towards something more productive.
Let's be frank. Saturn has decided that he doesn't want to work with TASVideos because we didn't publish one of his Super Metroid runs once. So he's made it a goal to troll us. Please stop feeding him.
Couldn't care less. My only interest here is figuring out ways to improve the TAS. If I happen to "feed a troll" or two along the way, then I genuinely don't give a flying fuck. At least I'm not calling them out on it and just fanning the fire, as that serves literally no purpose. I don't think we'll see an input file and I'm okay with that, but I still want my reasons for why I want one up there in case someone else later on comes around and needs some sort of "incentive/reason" to post their own input file.
On that note, I'm out of Tower 2 finally, hopefully I'll know how good I did after people review it some and suggest changes (though lately everyone has been super busy). Starting in on Tower 3, playing around with the bridge mechanics currently. Should make some solid progress on it either this morning or later tonight if I'm not dead tired from work today.
Updated nonsense:
.tasproj: https://goo.gl/eIgh7P
.bk2: https://goo.gl/p5UL5f
Discord group: https://discord.gg/KDhsRYj
If anyone wants to take a look at the pathing/movement/whatever, there you go. I will check on any suggestions that aren't "do better on the 4 red huball fight" because that's not possible on our RNG seed for that area. The 2 blue huball kill(s) in Tower 2 along the top are 2 frames slower than an optimal situation, yet getting an RNG seed which allows for a double kill is 10, 22, and 24 frames slower. Other than those 2 things, any suggestions on improving/fixing anything is more than welcome and greatly appreciated, even if it doesn't lead to anything.
As said before, the run is not suitable for this site in its current state, so there is no need for an input file when a video exists that shows everything in detail. Surely having it just to make use of the copy-paste function is alot easier and faster than optimizing something on your own, but at the same time it's also lame and kills the uniqueness of a TAS.
TheAngryPanda wrote:
The reason yours would be nice is because it's completed, it's something tangible that I can say "On this map, I'm 5 frames behind, what does he does that I don't and how can I learn to fix this in the future maps." Once that's done, I can further compare and challenge myself based upon your pathing and other decisions.
In case you missed it, there's a full 50 fps HQ video file available that can be played back on various media players on a frame by frame level, which allows an exact comparison for each room just as well when aiming for realtime.
OmnipotentEntity wrote:
Saturn has decided that he doesn't want to work with TASVideos because we didn't publish one of his Super Metroid runs once. So he's made it a goal to troll us.
That's not the case. Jumping to quick conclusions isn't always a good choice.
As said before, the run is not suitable for this site in its current state, so there is no need for an input file when a video exists that shows everything in detail.
Except it has been said multiple times that the video file is more a nuisance then helpful
We weren't asking for you to submit the run for publication. We were asking you to upload the movie file so we can see what input is done for each trick.
For example: http://tasvideos.org/userfiles/info/34192786141103242 this is Arc's user file for Zelda 2
I don't mean to sound like a jerk and walking you through such a basic process, but since we're like a page in and you still haven't posted this file I'm assuming you have no idea of basic site utilities that everyone knows about, or you somehow missed basic English and are reading a completely different topic from what actually exists.
Saturn wrote:
Surely having it just to make use of the copy-paste function is alot easier and faster than optimizing something on your own, but at the same time it's also lame and kills the uniqueness of a TAS.
It's a little condescending that you would assume Panda would just copy+paste your input. First off, it's on a different version so that wouldn't even work. Secondly, your input is probably not so godly that he'd want to copy+paste it anyway. Then again, I'm just assuming,
since you're being a moron and not revealing the thing already.
This'll be my last post on this issue. If you don't want to give the user file, fine. But at this point you're just being an annoying nuisance about it, and from here on I'll assume you're actively baiting for reactions over this userfile in order to try and pretend you're still important.
effort on the first draft means less effort on any draft thereafter
- some loser
Joined: 11/13/2006
Posts: 2822
Location: Northern California
Saturn wrote:
As said before, the run is not suitable for this site in its current state, so there is no need for an input file when a video exists that shows everything in detail. Surely having it just to make use of the copy-paste function is alot easier and faster than optimizing something on your own, but at the same time it's also lame and kills the uniqueness of a TAS.
You don't have to release the input file, but could you just stop lying and admit the truth? Your excuses are completely nonsensical, and we all know you're just too much of a stubborn egotist to release something that would actually be genuinely helpful to Panda and everyone helping out with the TAS.
Fair's fair, isn't it? You stole information from this thread and passed it off as your own work. It's only fair that you give us the input file that we made possible.
TASvideos Admin and acting Senior Judge 💙 Currently unable to dedicate a lot of time to the site, taking care of family.
Now infrequently posting on BlueskywarmCabin wrote:
You shouldn't need a degree in computer science to get into this hobby.
In case you missed it, there's a full 50 fps HQ video file available that can be played back on various media players on a frame by frame level, which allows an exact comparison for each room just as well when aiming for realtime.
I did miss it, thank you for pointing it out. DL'ing now.
And regarding the copy+paste comment, fuck you. You're pathing in Secret of Evermore was outright abysmal. There was no real testing involved, because character swapping is not faster than movement with the spear or diagonal inputs throughout multiple different areas. The same can be said of your Terranigma TAS video linked previously in this thread. Samsara alone has provided numerous places in which we beat your pathing by pure optimization alone, and this isn't even including the fact that you likely haven't reached out to anyone to look into properly tested routing for various parts of the game. It shows in your work, and it's sad because this game has a great community around it that is incredibly helpful to people learning things about it... And you just ignore this asset that could significantly improve your project. I get that working alone on something can test your "skill" and knowledge of something, or that it's nice to say "I did this, it is mine, all by myself!" I get it, it makes sense, but it's also very selfish, idiotic, and embarrassing at the same time because just having a few people look over a project segment can lead to multiple frame improvements, a better overall resulting movie, and honestly it just feels good to know very different people consider it to be better than anything else out there up to that point.
To reiterate, if you think that I want your input file just to copy and paste segments of it, you're a fucking idiot. Have a good one.
Assuming bad luck at tower one boss, it wouldn't pay off to kill extra enemies, as you lost at most two seconds. But who knows, getting hurt one or two times and hitting enemies without killing them might get you better luck.
What about the magic rocks, do you have a plan where you collect them? Three could be collected in tower one with only minor detours.
For the Magirocks there is no need to pick up more than 1. You need exactly 5 + the free BonePin you get from progressing in the game to warp out of Areas. One of the Magirocks will block your optimal path in Ra Tree so it basically only takes the amount of time to pick it up. The other 4 are provided in form of the GrassPin you get from the Ra Tree after defeating the Parasite boss. Equipping the Magic Chest when doing the menu in Grecliff to drink the Strength Potion and equipping the RocSpear will be the only time loss aside from -I would assume at most- another second casting it since there is exactly one frame at the Stormkeeper boss in Zue where the animation of the spell doesn't play but you get all 4 Magirocks back from it. Picking up 4 Magirocks manually would definitely be a bigger time waste as the only other somewhat fast to gather Magirocks are in Tower 2 and Zue. The other 2 are probably the ones in Grecliff you get somewhat close to. And since you want to have 5 at the end of Louran you are somewhat limited in where to get them.
Assuming bad luck at tower one boss, it wouldn't pay off to kill extra enemies, as you lost at most two seconds. But who knows, getting hurt one or two times and hitting enemies without killing them might get you better luck.
What about the magic rocks, do you have a plan where you collect them? Three could be collected in tower one with only minor detours.
Thanks for this, it actually highlights what I was concerned with in Tower 1 (the -15). I'm pretty sure I know where I can fix this at. Boss aside, I feel that I can still shave some time off with some knowledge I have now that I didn't before. It'll be some time before the next update though I think, just because the fastest place to change RNG pre-T1 is NPC movement in Crysta, so I'll have to work around that. Plan is to get a better manipulation on the huball movements in the beginning, which could save maybe 2-4 frames depending on things. This, in tandem with other manipulation improvements, could result in some changes to the boss AI, which might lead towards a faster time there.
Another part of the equation is that Terranigma has some pretty inconsistent load times, something I wasn't aware of initially. There's a console vs Snes9x 1.51 video by lemon (will provide the link below this) which raises some concerns, especially when the Crysta >> Tower 1 section (last comparison in the video) comes around. I don't think this should be viewed as an excuse though, or something to disregard the data you've provided, but it is something I'll have to keep in mind since the majority of people would agree that bizhawk is more accurate than snes9x 1.43, though by how much exactly is up for debate completely. I suppose I could test this myself, but I honestly would view this as a crutch and until I'm confident that my time in Tower 1 cannot be improved at all, will not be looking into it.
Lemon's video:
https://www.youtube.com/watch?v=QQ5tUSL7L0E
Like I've stated before, if anyone (and I do mean anyone) points out a possible improvement, I will look into it as much as I possibly can.
The timing in the table above for 1-boss is just before the fight starts. So the boss fight is maybe about 120 more frames in your video. The -15 is maybe just the imperfection of using a video instead of an input file. The difference from 2-cutscene to 3-start where everything could be done frame-perfect at full speed suggests that about 60 frames are added for each tower for language and emulator difference.
The following script (which I derived from another script you linked to, so it's mostly not my own work) allows you to test different RNG values in advance. I guess you can create a boss fight with a certain RNG value that way and test if the randomness is good, and if yes, go back and copy-paste the exact same fight after having done the necessary adjustments to the route.
--hit F11+frameadvance(!) to test next seed
local SeedsToTest = 50
local currentRNG = mainmemory.readbyterange(0x0408, 16)
savestate.save("damageteststartsave")
for i = 0,SeedsToTest do
for j=0,15 do mainmemory.writebyte(0x408+j,currentRNG[j]) end
while(true) do
local diff=mainmemory.readbyte(0x417)-currentRNG[15]
if diff<0 then diff=diff+256 end
gui.text(0,0,i.." additional advances of RNG. "..diff.." advances during play")
emu.frameadvance()
t=input.get()
if t["F11"] then break end
end
savestate.load("damageteststartsave")
local carry = 0
local temp
for j=15,1,-1 do
temp = currentRNG[j-1]+currentRNG[j]+carry
currentRNG[j-1] = bit.band(temp,0xff)
carry = (temp-bit.band(temp,0xff))/256
end
j=15
repeat
currentRNG[j]=bit.band(currentRNG[j]+1,0xff)
j=j-1
until j<0 or currentRNG[j+1]~=0
end
The difference from 2-cutscene to 3-start where everything could be done frame-perfect at full speed suggests that about 60 frames are added for each tower for language and emulator difference.
Perhaps it's tied to the screen transitions, as Lemon's video would suggest? Or am I misinterpreting what you're saying?
Also, thank you for the tip about testing RNG that way instead of the method I was using, that's quite a bit quicker than the way I was going about it. Will start up V2 of my TAS soon as I wake up a bit more with the info from this thread. Goal will be to beat or match V1 everywhere, but mostly to attempt to get a more tolerable boss fight since that's been eating away at me ever since I moved on from it the first time.
EDIT 1:
So I've been toying with just general manipulation of various things.
1) My point of manipulation prior to Tower 1 is actually the room before the pot throwing nonsense. Beyond this point, there is no other manipulation point until the Tower 1 blue huballs I'd be trying to manipulate (meaning they don't count). Jumping into the menu box or whatever it's called doesn't alter the value(s) either for our RNG.
2) I don't think manipulating the blue huball to move left is going to make much of a difference in the long term outside of just RNG rerolling/alteration. This is because I will be at a lower vertical position than it currently is, which means i'd end up wasting a frame to re-acquire that distance, even if I killed both blue huballs on the same positional value. So I think this point of manipulation is busted and I will be moving on to the next possible location to adjust at.
3) Not really a note of significant worth/value, but putting this here mostly so i'll see it again in case I forget later on. I probably should just do some straight-line testing to see just how much each action actually costs frame-wise sometime and post that data on here. This could possibly lead to explaining some frame loss in certain areas, though I'm pretty sure I already did this and just didn't write it down because... well i don't really know why i wouldn't write it down. :/
Anyways, back to working on it.
Yes. And even more so, as Snes9x 1.43 emlulates even less lag than Snes9x 1.51.
This is exactly what I assumed as well in the Discord-Channel for this TAS. The best thing to do would probably be to compare room by room and assume because of only having a video for Saturn's TAS to go and take the result and add/subtract lets say roughly 5 frames for human error to it. You can't really grant that the numbers you get are correct unfortunately. Also we don't really know if there is no difference in the PAL and NTSC-J version so you can just compare them like that.
arandomgameTASer:
You made your points, I made mine. Time to move on.
Panda:
It was a general statement, no need to take offense.
Samsara wrote:
You stole information from this thread and passed it off as your own work. It's only fair that you give us the input file that we made possible.
An outdated WIP till Tower 3 and a handful of single trick-demos (that are basic stuff to find for a TASer with some general knowledge) doesn't make you an entire, almost 4 hour long run that consists of 99% never before TASed situations.
Regarding timing, there should be no difference between NTSC and PAL when comparing frames (other than at text scenes) if done within rooms only, as long as they don't lag. Screen changes should be avoided for previously mentioned reasons.
arandomgameTASer:
You made your points, I made mine. Time to move on.
You haven't actually made a point lol. 10/10 trolling.
Saturn wrote:
An outdated WIP till Tower 3 and a handful of single trick-demos (that are basic stuff to find for a TASer with some general knowledge) doesn't make you an entire, almost 4 hour long run that consists of 99% never before TASed situations.
Except there was far more in the thread besides that, including lua scripts and an entire theory TAS along with those 'basic' trick demos, all done over a year before your run. But yeah don't worry that's not insulting at all to the people who did those.
Saturn wrote:
Regarding timing, there should be no difference between NTSC and PAL when comparing frames (other than at text scenes) if done within rooms only, as long as they don't lag. Screen changes should be avoided for previously mentioned reasons.
Do you know what a PAL is?
Anyway, after addressing the troll far too much for my own liking, let's get back on topic.
TheAngryPanda wrote:
1) My point of manipulation prior to Tower 1 is actually the room before the pot throwing nonsense. Beyond this point, there is no other manipulation point until the Tower 1 blue huballs I'd be trying to manipulate (meaning they don't count). Jumping into the menu box or whatever it's called doesn't alter the value(s) either for our RNG.
Is manipulation in Terranigma extremely rigid? I'm so used to games where manipulation is entirely input-based or button-based, so I can't imagine how hard that must be to figure out.
effort on the first draft means less effort on any draft thereafter
- some loser
Joined: 11/13/2006
Posts: 2822
Location: Northern California
Saturn wrote:
and a handful of single trick-demos (that are basic stuff to find for a TASer with some general knowledge)
Atma wrote:
saturn did a test run of this a while back
Good job finding it all back then! You did a great job populating this thread with "basic stuff" that you definitely found during your first, complete test run of this game!
TASvideos Admin and acting Senior Judge 💙 Currently unable to dedicate a lot of time to the site, taking care of family.
Now infrequently posting on BlueskywarmCabin wrote:
You shouldn't need a degree in computer science to get into this hobby.
Is manipulation in Terranigma extremely rigid? I'm so used to games where manipulation is entirely input-based or button-based, so I can't imagine how hard that must be to figure out.
As far as I understand it, this is what affects RNG:
1) Idling X frames (i don't know the exact amount) so Ark does his little idle animation, which triggers the RNG because it's a sprite being altered in some manner.
2) Enemies performing any action, be it attacking, moving, etc...
3) NPC movement, pretty self-explanatory here.
The things I know do NOT influence the RNG, at least not that I've seen using the scripts thus far:
1) Changing your direction*
2) Jumping/Attacking**
3) Waiting w/o idle animation being triggered (IE; delaying frames w/o any other method of manipulating RNG present)
*The only way this "works" is if it somehow changes the method in which an enemy approaches you, so it kind of affects RNG, but it's second-hand rather than a direct influence. I've used this once so far.
**If the attacking animation triggers a different reaction than non-attacking input while around/approaching/bypassing an enemy, the enemy may do something different according to whatever it's AI tells it to. I haven't seen this first-hand yet, but I imagine it could lead to some kind of influence over the RNG.
May have left out some things, sorry if I did. I'm kind of running on ~4 hours of sleep at the moment.
Regarding timing, there should be no difference between NTSC and PAL when comparing frames (other than at text scenes) if done within rooms only, as long as they don't lag. Screen changes should be avoided for previously mentioned reasons.
Bullshit. All PAL releases of Terranigma run 16.7% slower than the NTSC-J release (known as Tenchi Souzou). It's already noticeable if you just let the intro sequence play.
To top it all off, the PAL releases have an unskippable "Nintendo Presents" screen that's not present in the NTSC-J release.
Bottom line: Stop with your uninformed assumption there's no difference between NTSC and PAL, when there's a noticeable one.
Bullshit. All PAL releases of Terranigma run 16.7% slower than the NTSC-J release (known as Tenchi Souzou). It's already noticeable if you just let the intro sequence play.
To top it all off, the PAL releases have an unskippable "Nintendo Presents" screen that's not present in the NTSC-J release.
Bottom line: Stop with your uninformed assumption there's no difference between NTSC and PAL, when there's a noticeable one.
You misunderstood. It's obvious that PAL runs slower with 50fps instead of NTSC's 60fps, but the actions that happen in this game between each frame are identical regardless of the version, since there is no speed adjustment in the PAL version. For example if you need 100 frames to run across a certain room on NTSC, you will also require the exact same amount on PAL. The seconds will differ with NTSC being faster, the frames not. That's why a pure frame comparison is possible between NTSC and PAL as long as text differences and lag are not involved.
Saturn, all of this fighting back and forth is getting tiring. Just release your gods damned tax returns input file so that comparative work can continue with the game.
Comparing against a video is a serious pain in the ass, especially when there are timing issues involved, and if the inputs line up, as you say, then working from an input file is going to be much more reliable than working from a video.
I am still the wizard that did it.
"On my business card, I am a corporate president. In my mind, I am a game developer. But in my heart, I am a gamer." -- Satoru Iwata
<scrimpy> at least I now know where every map, energy and save room in this game is
The seconds will differ with NTSC being faster, the frames not. That's why a pure frame comparison is possible between NTSC and PAL as long as text differences and lag are not involved.
You realize that PAL LOSES frames compared to NTSC? They're not directly comparable at all really, so saying that if you ignore the PAL differences and just compare the frames is kind of stupid and dumb.
But assuming this completely wrong claim is right, this would be far easier to compare if we had your input file and not a dumb video. Not to harp the same point even though I said I wouldn't, but uh........yeah.
effort on the first draft means less effort on any draft thereafter
- some loser
