A new WIP which progress up to Cerulean. Squirtle turned out to be weaker than I had anticipated against the Lass in Mt. Moon. This means that I had to manipulate two 1/256 misses back to back against the Oddish. I think Tilus is one of the few who could truly appreciate this battle. I also rolled DVs of F7DF for Gyarados; a little higher speed would have been nice, but overall very good. I had also thought that one could release pokémon directly without storing them first, but apparently not. It looks like I will have to switch boxes now and then, but I'll be making enough trips that this shouldn't be a problem. Next stop, Nugget Bridge.
I think that many viewers, myself included, have never seen the certificate from the head programmer. I thought that for this reason, it would be interesting to end the run with this displayed indefinitely. Although there won't be a credit roll, I still think that this is enough closure to catching all 151 pokémon. Were someone to do a run which has the goal 'Fastest time to obtain all 151 pokéman and defeat the Elite Four', it would no doubt be at least an hour or so longer than this run will be. I think that such a run would begin to drag on towards the end.
That's what I thought originally, but I don't think it offers the same closure. It would be as if FODA had ended his SM64 input after speaking to Yoshi on the roof instead of beating the game.
I think it should be after beating te elite four because Oak's rating could be used as the closure on the run.
Unlike Mario 64, how many pokémon you have in no way corrilates to how close you are to beating the game. Basically, the run you're suggesting would be a completely different, much longer run. For example, if I were planning to go through victory road, there would be a lot of pokémon I would catch there, instead of evolving them from their earlier forms. As it is, I am aiming for the fastest time to catch the 151st pokémon, and to view the diplom afterwards.
You mention in your submission text that you would get 8 escape ropes in your run if you the money. Where would you use this 8th rope?
Saffron Gym. You can use and Escape Rope to exit instead of having to take the telporters. If I recall correctly, this is only a minimal gain though, something like ~400 frames.
Chamale wrote:
I had an idea for a new route. After pewter city, there is a Youngster facing downwards. He has 4 pokemon. There is also a Lass with 2 pokemon. However, if you try to sneak behind this youngster the lass sees you too soon and you either have to fight the youngster or walk back to a building and go in. However, this is my idea:
Don't go to the mart immediately, fight the Lass at that junction, go to the mart and get stuff.
This has 2 potential ramifications:
1. Having 8 ecape ropes
2. Charmander staying at level 15, and not trying to evolve.
I'll test these, and tell you the results.
I guess the question is if walking back to the Mart and to the Lass again is quicker than the time it takes to Cancel evolve, and defeat two pokémon. My instinct would say that it would be faster to defeat the Bug Catcher with four pokémon, but I couldn't be certain.
So, 22000 frames and 14400 re-records later...
and I've finally got progress through Mt. Moon. This should prove to be the most difficult segment of the run, and I'm glad to finally be through it. I got rather lucky manipulating the Spearow and Jigglypuff; I was able to get them both without dropping a frame. It's interesting to notice how much cash I have when I buy the Magikarp... also notice how the man moves afterwards (subtle glitch). From this point, I'll be catching a Clefairy in Mt. Moon, and glitching both a Gyarados and a Gengar while I'm there. Afterwards, there'll be a small glitch-fest in Cerulean (including a MissingNo. encounter).
Mukki wrote:
I agree. Also, Primo, can you post a link to the route?
Sure thing. This is not complete, and it also contains a few errors... namely I forgot to pick up a Drowzee while I was in Vermillion, and catching a Growlithe won't work exactly as written. I'll most likely end up having to let two pokémon be sent to a box and retrieve them later.
I mean, with the name ASH, you may as well name the player GARY, which also works. But I choose ambiguous names.
This would actually save time. Stylistic choices give the run character, however, so I don't think that your decision is a bad one.
Chamale wrote:
On another note, I've spent 1000 rerecords luck-manipulating for Charmander's stats. The best I've gotten is D4F9.
How are you attempting to do this? Are you systematically searching? If I recall, the last button press before the stats are rolled is when you enter a new name.
Chamale wrote:
Judging by the fact that Charmander has the second DV very low in your run, I figure that it's for defense or HP, but what do the other 3 mean, exactly? Will having a 4th DV of 9 result in any battles being non-optimized relative to FFFF DVs?
In order, the DV values represent Attack, Defense, Speed, and Special. HP is derived from the other four. It's not necessary to roll a perfect Charmander, as you won't be using him long enough for a small variance to make any difference. Minimum DVs I would recommend would be D0BD. Gyarados, on the otherhand, will need at least F0DF (although I had speed problems with Gyarados, it wasn't until the very end, so D should probably be ok. Ideally, you would roll F0FF).
The name "AL" is the fastest name to input (and a shortening of my name) and will save 150-ish frames, through the fact that it is 2 frames shorter. It takes 80 frames to input, but it is very worthwhile.
According to my calculations, this will only save ~65 frames, whereas a one character name would save ~142. Another choice would be to use Pokémon Blue, and name the rival Ash, which would save 77 frames (probably my choice of preference).
Chamale wrote:
How far along should I be before posting a WIP?
In general, there is no such thing as too frequently. In fact, it often happens that a runner waits quite a while before posting, and then someone points out a major mistake early on. Although in your case, I would probably wait until after you finish the first battle.
Uh, not to sound naive but...I don't suppose you could glitch a Mewtwo, could you?
You sure could. However, the only attack it would know would be Psychic, which only has 10pp. This would make it a bad option, because you'd need to teach it several attacks in order to be able to finish the game. Plus, because there is no Thrash TM, this would be inherently slower than using Marowak.
These are the only problems I anticipate:
1. Agatha. She's troublesome for any pokemon except Mewtwo or Alakazam, but this is the problem: How the hell do we kill Golbat? This really worries me. I suppose the best way to do it is to luck-manipulate for having Golbat and Arbok at the end of the fight, then killing them with Thrash.
2. Lance's Aerodactyl. Again, the only attack that works is Thrash, but this time it's not very effective. Still, these problems are not very large given the huge improvement (potentially) offered.
You need to learn Strength anyway... you could teach it to Marowak. That would solve the Golbat problem. Although this probably wouldn't be a great Idea because then you'd end up having to overwrite Rage, which is a great asset against weaker trainers, as it saves your Thrash PP. Plush Strength has a slow animation. Ice Beam seems like it might be good option. You wouldn't be able to teach it to Gyarados, but Bubblebeam should be enough for everything he needs it for. This would solve both the Golbat and Aerodactly problem, and the potential Rhydon problem in Veridian Gym (although, you could have used Bone Club here as well).
Chamale wrote:
But, wow, this really hangs on the speed at which Missingno. glitch can be done. Another method involves trading. This would mean catching a Spearow as a fly slave, eventually flying to Vermillion and trading it, then teaching Farfetch'd Fly and going to Cinnabar. It seems too long to me.
Just grab the Pidgey where I do in my run. This is how I would imagine this run is total:
Start - Cerulean: Similar to existing run.
Route to Bill's: Fight the Jr. Trainer instead of the avoidable Hiker (saving 2 Bubblebeams).
Bill's - SS Anne: Similar to existing run.
SS Anne: Use Bubblebeam on Rival's Pidgeotto (instead of Hydro Pump).
SS Anne - Celedon: Similar to existing run.
Celedon: Pick up Poké Doll and Ice Beam, but don't teach Ice Beam to Gyarados.
Celedon - Fuschia: Skip the Gym in Celedon, and come back later with Marowak.
Fuschia: Teach both Surf and Strength to Gyarados upon entering the Gym (but *only* if Strength results in OHKOs against the Drowzees. If not, keep Bite and teach the HMs after the Gym). Manipulate a Explosion miss from the Weezing (this was a mistake in my run).
Fushia - Cinnibar. Fly to Pallet, Surf to Cinnibar. Enter the Pokémon Mansion and get the Secret Key. Escape Rope out. Fly to Veridian and Talk with the Old Man. Fly to Cinnibar and Catch Marowak. Fight in the Cinnibar Gym while you're there.
Cinnibar - Saffron: Do Celedon Gym, Office Building and Saffron Gym. Probably the Gym before the Office Building, but the other way may prove faster.
Saffron - Indigo: Similar to existing run, but don't heal if you don't need to :)
You can't glitch a high-level Gyarados. Not even in the Japanese version.
Fact.
However, I think Kingler might not be such a bad option. Crab Hammer, Stomp, Guillotine... good speed, awesome attack; seems like a solid package. Especially since Stomp is also a flinch attack (although, with Guillotine, why would there ever be a two hit battle?). Only drawback is that his call is not as short as Raticate or Snorlax.
Raticate seems to me a much better option than Snorlax, if from nothing else, from speed alone. Even at level 100, Snorlax would still but slower than Gyarados, who himself had many second turn attacks, losing ~150 frames for each. Raticate on the other hand is decent, has similar speed to Kingler, but unfortunately attack which is 27% less*. Also only two real attacks, Hyper Fang and Quick Attack. Hyper Fang is good, but still weaker than Stomp. And Quick Attack is no stronger than Scratch (!), although with 30pp, I can see how it could prove useful. However, a very, very short battle cry.
Keep this in mind for the Ghosts though: Raticate really only has one attack to use against Agatha, Dig. Dig is a two turn attack, which means 300+ frames extra for each ghost = 900+ frames just for using this attack. Plus, you'll also probably need to use it else where to make it to Indigo Plateau without healing. This alone is 15 seconds, which is more than your estimated 10 seconds saved from battle cry. You could alternatively teach Raticate a special attack, but with his low special, it would probably result in a 2HKO as well.
Kingler, with STAB on Crab Hammer would have a much better chance of OHKOing the ghosts. Plus, you will most likely need to teach Raticate an additional attack, which I estimate to cost ~710 frames (assuming you were already in the item menu). Plus, as mentioned above, you have the OHKO wildcard, in the case of a particularly difficult opponent.
In conclusion, I advocate Kingler over Raticate. Although, there may be other, potentially better options.
*Comparing L100 without stat experience.
Edit:
You can ignore everything I just wrote about Kingler. I wanted to make a few suggestions of potentially better options, and I've convinced myself that I've found the best. Marowak. Marowak is appealing for only one reason, but for a very, very good reason, Thrash. Thrash works like this: you send an attack once, and your pokémon attacks 3-5 times in a row without having to issue a new attack. With 20pp, you could take out just about every opponent between Cinnibar and Indigo Plateau. This would save a lot of time during battles (~40 frames per enemy after the first) and you wouldn't need to teach him another attack. Possibly not even need to save at Indigo Plateau, but I'm not yet convinced of this. And then you still have Bone Club for the ghosts. Although It attacks twice in one turn, this is still a lot faster than Dig (first, because you only have to issue it once, and second, because you opponent doesn't get to attack in the middle), and if you opponent faints after the first hit, that's it. He also knows Rage, which works exactly like Thrash, but starts out much weaker. But for many battles, it should do the job, allowing you to push Thrash through to the end. His cry is not as short as Raticate, but still shorter than Gyarados.
I'm probably the only one that still cares about this movie...
Nope, you're not! I've just gotten through my exam period, so now I have a lot more time (at least for the next month or so). I intend to utilize this time to play pokémon (wow, do I sound like a loser).
Anyway, I'm past that battle, but I'm still on route 3. Once I'm through, I'll have another tough battle (well, tough to manipulate at least) against the Lass in Mt. Moon, but after that progress should be very quick.
I also discovered something that will save a little time while I was redoing battles. I was constantly comparing against my first movie to make sure I wasn't losing time. Suddenly, I lost 60 frames on a battle which was, in my opinion, nearly frame optimal. And this was the reason: when a critical hit is used to defeat an enemy, the message remains on the screen about 15 frames less than when it is used in the middle of a battle. This means that if you need to defeat an opponent with 2 attacks, one critical, one normal (which is often the case), it is ~15 frames faster to do the normal hit first, and the critical second. For this battle, the first time through I had done it correctly for each of the four enemies (which is odd, because I normally do it exactly the opposite for easier manipulation), and the second time through, I had done it incorrectly. Sadly enough, I also made this mistake on that blasted Caterpie. I've decided not to redo it.
Long story short, I'll be posting an update as soon as I enter Mt. Moon, which shouldn't be long.
I was thoroughly entertained, and the route seemed extremely well planned. I was also very impressed by your health management:
"Hmm, he's almost out of health. Oh look, energy tank."
It's also a huge time improvement over the last any% run. Voting yes :)
Heh, I bet if you were allowed to specify the start time of the game, people would try to find whatever time gave the best randomness, no matter how far away from the present it was.
This could have some pretty interesting effects. I can imagine a programmer thinking to himself "no one will be playing this game in 150 years... I don't need to handle that special case in the PRNG."
I was wondering, is there any reason why the vote for movie are private ? I mean, it would be nice to see who vote what, and see if they explained why.
I can only see this as counter-productive. Some people might go through all of their submissions, and come to the conclusion that there is a team of people who are against them, because they always vote no, or whatever.
Others might be discouraged from posting at all. Quick rant. Online forums seem to support mob behavior. I've seen this cycle a few times: A new tas is submitted by a respectable runner. The run isn't terrible, but it isn't anything spectacular either. The first 5-10 comments are positive, 'definite yes' and so on. Then finally someone with enough guff states what they think, that the run is boring, that certain parts are repetitive, that certain things could be better optimized, etc.. Suddenly, everyone is of the same opinon, as if alleviated from the responsibility of standing behind their own opinion, which was, until now, the minority.*
Or vise versa: A new tas is submitted by someone no one's ever seen before. The run is good and even beats an existing tas by a few seconds. The first few comments are negative, stating that this run, while faster, is not as entertaining as the previously published tas, and other subjective qualities (although entertainment is always subjective, but that's a different topic). Then someone with enough sense points out that the previous run was praised for being technically perfect, and that this run is even faster. Suddenly everyone sees the movie for what it is: an improvement to an already excellent run.* End Rant.
ZeXr0 wrote:
I think that voting without giving a reason should just be ignored, because it does not gives any feedback to the player what they did wrong.
This would be particularly annoying for anyone who just wants to vote 'Yes' without restating the same thing the last ten viewers wrote.
Deviance wrote:
If the lack of anonymity leads to conformity, it may do more harm than good.
Agreed.
*These scenerios are completely fictional and may or may not contain some or any semblance to reality.
True, although German is not my mother language, nor do I consider myself to be a good representative of the TAS (and even less real-time speedrun) community.
Perhaps a good place to start would be to make a list of all the qualities that would be desirable to make a TAS, and then eliminate them.
An easilly manipulatable PRNG. For example, one in which a single frame difference of input/delay causes completely different results. One would need to carefully design it so that small variances in time or input have absolutely no affect on the outcome. Be sure to give many occurrances a true 0% possibility if certain criteria are not met.
Alternate, obviously shorter paths which require near-super human ability to perform. Do away with all of them. Don't make 'impossible' jumps that are one or two pixels away from obtainable just to frustrate players, because someone will figure out a way over it that you didn't expect.
Game physics which break at 60Hz input, causing glitches, such as being able to walk through walls. If you know that the possibility of a certain glitch might exist (such as a player getting stuck between to bricks in Mario Bros.), make sure it does something which will impede the players progress. To use Mario Bros. as an example again, the game should have shoved the character _backwards_ instead of forwards. Imagine the Rockman run if zipping had forced the character the other way. The glitch would have been pretty worthless.
Assumptions about controller input, such as left+right/up+down. Treat every combination of key inputs as valid, and make sure your code handles each reasonably.
Knock-back. Always make knock-back reverse the direction of motion (and not the direction the character is facing). In addition, ensure that total velocity is never increased, and never produce a knock-back which forces the character in a vertical direction.
Limiting number of 'bullets' to a certain number on the screen. This is always a bad idea, because it allows players to stand right next to an enemy and turbo fire at 30Hz. Instead, limit shots to a certain number per second, like 3 or 5. Additionally, never have a stun affect for any difficult enemy, such as a boss.
Invulnerability of any sort. If you do want to have invulnerability, perhaps only make it against the enemy which actually damaged the character, and not all enemies/obsticles. Make sure you collision detection is decent too. If you need to estimate for performance reasons, make sure the hit box is larger than the character, rather than smaller.
I think that'd pretty much cover it. Although I'm very much the opinion that every game should be made with speedrunning/TASing in mind. For one, it increases the replayability enormously.
Then again we can use Gyrados' bite for the same purpose.
I think you have a good point here. In retrospect, switching Bite for Strength was a bad decision. Although it would have caused more 2HKOs (most notably in Fushia and Saffron gyms), it would have saved half of the battle dialog (about 150 frames or so) for all of the 2HKOs afterwards, of which there were twelve iirc. Add in the fact that Strength has an animation which is 55 frames slower than Bite (nearly a second!), and I think you've got a pretty solid case. Gains would probably be in the 40-60 second range.
Mukki wrote:
This would cut out the need for glitching high-level pokemon, of which I am still in doubt over how much time it would save.
I don't think this will save as much time as Chamale suggested, but I do think it would save at least a little. The level up music is 161 frames, and Gyarados levels up 14 times Cinnibar on. Add in the fact that we can skip the Office building beforehand, and you've got another 3. Then there's 12 2HKOs times 300 frames each, and 150 frames or so for every battle in which Gyarados had second attack (I have no idea how many, but it was quite a few). After subtracting the minute or so to perform the glitch, this once again puts you in the 40-60 second range.
I have no clue which would be faster. I might end up doing two runs, one each way, and submitting whichever wins.
Another quick note about selecting attacks with the shortest name... this is usually not the best policy. Take our Caterpie friend for example. He only knows two moves: Tackle and String Shot. Choosing a Tackle miss is the wrong choice, even though it's five characters shorter. Here's why:
Caterpie used Tackle!
Enemy Caterpie's attack missed!
Caterpie used String Shot!
But it failed!
Tackle ends up being 12 frames longer. This is not always the case though. Disable also has the 'but it failed' failure message, but for some reason, there is a very long delay before it is displayed (something around 30 frames). I guess it has to go through a different logic check or something. This is why I opted for a Pound miss from Grimer in previous runs.
Anyway, I'm just rambling now. I had really hoped to be able to make a large improvement before submitting a new run, but I don't think there's too much room left.
It's time for another Good News, Bad News.
Good News:
Ultra-Blaine can be defeated. I actually decided to go with Kingler instead of Seaking (with an attack called Guillotine, can you blame me?). Blaine's lineup looks something like this:
L111 Drowzee, L111 Rhydon, L111 Grimer, L111 Pidgey, L111 MisingNo., L111 (No Name)
Rhydon went down with a single Crab Hammer, and the rest fell victim to the guillotine block. His sixth pokémon is a strange glitch pokémon of sorts, but fortunately not one that crashes the game. It has a silent cry that lasts for about 5 seconds, and the graphic looks kinda like spaghetti sauce flung at a white wall. Pretty cool, actually. Changing the character before the male symbol didn't seem to have any effect on the lineup.
Bad News:
Ultra-Blaine is just a normal trainer. He doesn't say anything after the battle, nor does he award the Volcano Badge. I'm not sure if there's a way to initiate the actual gym battle, but it doesn't seem likely. Quite a bummer, but it was worth a shot.
I've got good news, and I've got bad news. Bad news first. In Pokémon Red, Blaine's first pokémon is a glitch pokémon that completely crashes the game. Changing the letter directly before the male symbol has a mild effect on how the game crashes, but it crashes universally (admittedly, I only went through half the alphabet, so I'm just extrapolating).
There is good news, however. In Pokémon Blue, Blaine's first pokémon is a normal Drowzee... L111 that is. I was unable to defeat it, so I don't know if his subsequent pokémon (he has six in all) might also cause a game crash.
So, now we need a way to defeat 'Ultra-Blaine'. One thing that comes to mind is OHKO moves. Of the L100+ pokémon we can glitch, 2 of them would start with a OHKO move, namely Kingler (Guillotine) and Seaking (Horn Drill) (unfortunately, both of these attacks are normal type, and can't be used against ghosts). Kingler has an outrageous attack, and very good speed. He also starts with a good arsenal of attacks (Guillotine, Stomp, and Crabhammer). His only weekness is that he has a low special, which means he wouldn't be very effective against the ghosts. Seaking is more of an all-round pokémon (decent attack, decent speed, decent special), but only starts with two useful attacks, Waterfall and Horn Drill. Even so, I still lean towards Seaking, because it would have a much better chance against Agatha.
An alternative might be Aerodactyl, which knows Hyper Beam from the start (as well as Bite, which isn't bad). Excellent speed and attack, but once again a low special. The low special doesn't even really matter, becuase he can't learn any special attacks any way, other than Fire Blast (which we won't be getting!). His only chance against the ghosts would be Fly, but it's a two turn move, which isn't so appealing.
I'll be testing the Seaking route and posting results later.
Well, I did have progress up to the entrance of Mt. Moon... and then I realized I had made a pretty big mistake. For the first segment, I only need 14 Poké Balls. I decided to buy 15 because I had enough money, and I figured that it would save menu time if the Pokè Balls remained near the top of the item menu. However, this turned out to be a bad decision, because I was about 50 monies short of being able to buy the Magikarp in the Poké Center. To continue from this point would mean having to buy it at a later point, losing about 90 frames for entering the building again, 90 for leaving, and around 144 wasted from extra walking distance, all in all about 5.4 seconds.
So I had to go back and fix it. This was rather painful, because one battle in particular (the very first Caterpie on Rt. 3) had taken over 2000 re-records to manipulate the optimal battle situation. While I was at it, I went back to the first gym and switched Bubble up to the top of the move list the first time it's used, which will save about 30 or so frames from switching between attacks (at the time, I had miscalculated, because I had planned on using tackle through Rt. 3, as I had originally done, but I didn't need to because I won't be using Squitle as long as I did before).
I'm currently stuck on the Caterpie again. He has 30 hit points, and maximum non-crit is 9 damage, maximum crit is 13. Seems quite doable; one near max crit of 12 and two max non-crits should suffice (which is actually what I had the first time). However, manipulating the first 9 point attack is like hell on Kanto. Critical hits have the advantage that you have an extra button press after the critical hit message box to manipulate an opponent miss. On a non-crit attack, both need to happen at once (basically). That, and there's the fact that Bubble has a 10% chance of slowing the enemy. This seems small, but of every one of the 9 point attacks I've manipulated (well over 100, I'd guess), all but two have slowed the enemy. Sometimes it is possible to manipulate an opponent miss after a non-crit attack by inserting a random button press during the 'Enemy X used so-and-so' dialog, but for those two cases I couldn't get it to work.
Long story short, I've used over 6000 rerecords on this single battle and have yet to make any progress. I think I might alter my strategy to getting a 13 point crit, and then an 8 point normal. However, max crits are somewhere in the neighborhood of a one in 200-300 chance. Add in the fact that I need to enemy to use String Shot, and the chances have just been halved. I'm very tempted just to lose the 20 frames and go with the trivial 11-11-8 victory, but somehow, I can't bring myself to accept that. I plan to get over this hurdle within the next few days. When I do, an update will be posted.
I think I've discovered something which should save a couple of minutes. By having the male symbol (♂) as the third or fifth character of the hero's name, one should be able to glitch an Encounter with Blaine using the Cinnibar Surf glitch. Because the Volcano Badge is rewarded during the fight sequence (and not in the dialog with Blaine afterwards), this should allow being able to skip both the Pokémon Mansion as well as the Cinnibar Gym*. This is a nice result, because the Cinnibar Gym is annoyingly long (all those trivia questions), and the Pokémon Mansion requires using an Escape Rope which could be put to better use (namely after Sabrina instead of taking the teleporters back out).
How much time this saves exactly is hard to calculate, but I estimate somewhere in the 3:20 neighborhood (I measured from entering Cinnibar to encountering Blaine). Considering the glitch takes about one minute or so to do, this should result in a net gain of around 2:20.
Since we're going to be performing the glitch anyway, there's no reason not to grab a L100+ Pokémon while we're there, such as Snorlax as Chamale suggested. The improvements from catching Snorlax alone are minimal, but in conjunction with the above, it should save another 1:20-1:30 (I'm not convinced about time 'saved' from critical hits and super effective messages... I think most will still be necessary, but level up music and 2HKOs will definitely save time). Overall time saved should be in the order of about four minutes; an improvement large enough to warrant another run imo :)
*NOTE: I am not certain that fighting Blaine in this manner will reward the Volcano Badge. But, for reasons stated (it is awarded during the battle sequence, not afterwards), I have good reason to believe it will. One could test this by GameSharking the hero's name and then performing the glitch. I'll be doing this later and posting results, if nobody beats me to it.
