Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Thanks Steelix100. I fixed the game name and also put your encode at the top of the submission text. By the way, it is possible as the submitter to edit your submission as well.
By the way, I was interested in a full TAS of this game and game mode (T.A. Death) ever since DeHackEd made a WIP 11 years ago (never completed). I didn't have the means to do this TAS (or to trust MAME for that matter), but I'm glad you did. Also, I didn't even know of the Secret GM ending.
I noticed in real-time plays of this game mode that once you reach Lv999, the blocks become invisible or something. Is it something that is supposed to happen here but doesn't because of emulation error?
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
I just finished watching this (though I used the skip-forward button a bit near the end). While I don't find this game particularly exciting (having never seen it until now), I found that the strategies used were pretty interesting, like how Cait Sith completely breaks the game when you can manipulate luck. Though I really wish you didn't name three of the characters "C".
I got fooled by the Hojo skip at the end; not knowing that the sequence wasn't supposed to go like that, I wondered why the music cut out for the rest of the run and and thought at first it was an encoding problem.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
In case anyone is wondering, I used [url=
https://pastebin.com/bb9a3eEB]this C++ code[/url] to generate [url=
https://pastebin.com/J8CqxaDy]these results[/url] telling me the RNG values along with their index/cycle number (where 0 is the first RNG value 0x3FF) and delay to get the next blackjack. Then I analyzed the values in this table (by hand) to determine 14 consecutive blackjacks with not too big a total delay.
Other things:
- The RNG is a 16-bit linear feedback shift register.
- The title screen runs through RNG values much faster (~47.5 values/frame) than everything after that (1 value every 3 frames), so any starting value within 15000 cycles or so has potential (that's why my code searches up to 20000).
- The shuffle algorithm works as follows: the game puts all the cards in order and then performs 52 swaps. This causes the RNG to be cycled many times, and not consistently either (it will cycle at least 6*52 times for the swaps but will cycle extra times if RNG value mod 64 is greater than 51 (which is a 12/64 chance).
I guess this stuff isn't actually all that interesting to most people, but at least it's fun to see how old video games handle RNG.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Riddler Classic this week:
----
You and I are playing a game. It’s a simple one: Spread out on a table in front of us, face up, are nine index cards with the numbers 1 through 9 on them. We take turns picking up cards and putting them in our hands. There is no discarding.
The game ends in one of two ways. If we run out of cards to pick up, the game is a draw. But if one player has a set of three cards in his or her hand that add up to exactly 15 before we run out of cards, that player wins. (For example, if you had 2, 4, 6 and 7, you would win with the 2, 6 and 7. However, if you had 1, 2, 3, 7 and 8, you haven’t won because no set of three cards adds up to 15.)
Let’s say you go first. With perfect play, who wins and why?
----
This is a well-known problem with a trick. See if you can find the trick. (The answer to the question isn't as important even though I know what it is.)
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Here's a TAS that completes it in 1:04.953 by getting 14 blackjacks in a row.
Link to video
Movie file: http://tasvideos.org/userfiles/info/52849058282505417
I believe it can still be improved by reducing the demands on RNG manipulation; namely by allowing for the first game to be a "dealer stands" (you don't gain any more money by blackjack on the first one) and by switching another couple blackjacks with a "dealer stands" and a "double down, dealer stands".
In case anyone is wondering: I disassembled the RNG and used a C++ program to figure out which RNG values give blackjacks.
Patashu wrote:
All the games share the same money count, right? So surely it doesn't matter which game you play, the goal is just to hit the maximum amount of money, which is the end of content/best possible state to reach?
Even better: You can literally set your starting money to 9999 and get maximum money immediately. Nothing stops you from doing so.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Nice work, Flameberger! I wasn't sure at first how far you were into your project, but it looks like you got it completed pretty quickly. You did manage to kill the gnome in two hits; I wonder how hard it was to get the RNG required for that to happen. Also I noticed you manipulated a lot of triple rapier misses; doesn't seem easy to do, but you killed the gnome in two hits so anything is possible, really.
Some notes:
- (50:06) Against Accursed (invincible Isher), I noticed you have Lina use Hourglass four times. I think it should be possible to have Serene use Hourglass three times instead, if you have everyone use Hell Scythe on the first round (that should make Lina's WT high enough), and you manipulate the WT at the beginning.
- (1:29:38) In Chapter 4 where you have to dodge the arrows, I think it is faster to fail this one.
- (2:15:57) I forgot to mention before, but it seems the VBA-Next core has a problem with rendering LV3 Disaresta, such that it lags heavily. I don't think that is even supposed to be the proper emulation. The old VBA emulator as well as the mGBA core don't seem to have this problem. The previous TAS also had that happen as well whenever LV3 Disaresta was used.
- (2:25:41) I think it is slightly faster to let Ein take the trap hit instead of Lina (unless you weren't able to manipulate it in time).
- (2:39:19) Did you consider the Slime Ball strat to poison Malice? Or did it not work out?
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Jan. 2019 rankings
By the way, there was a TAS on Nicovideo for the game "Getting Over It with Bennett Foddy" (I heard it's quite the troll game, from the same person who made QWOP) which has over 130K views. It was uploaded last November some time, but did not show up in any rankings. I have no clue how it was missed by whoever is organizing the rankings.
Link to video
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
p4wn3r wrote:
Suppose you have an amount of capital C you want to reach. Your plan is to do so by investing an amount A periodically. Suppose that, by the time you have to deposit this amount A, the capital you already had increased by x. That means, if you had M in your account, after a time unit passes, you'll have M*(1+x). Find the amount of deposits n of value A in order to reach the goal C. Prove that the formula tends to C/A when x->0
If you make n deposits of value A with interest rate x, then the value in your account will be:
which simplifies to
(*)
Setting equal to C (the goal) and solving for n, we get:
(**)
For the example that you gave, plugging in C=1000000, A=2000, x=0.01 gives approximately 180.07 months, so it takes 181 months (~15.08 years) to reach the goal.
But that's not your actual question. The actual question is to prove that the formula tends to C/A when x->0. We could use L'Hopital's rule on (**).
Or instead we could just see that (*) approaches An as x->0. This is because ((1+x)^n - 1) / (1+x - 1) is precisely the value of the derivative of z^n at z=1, by definition. This gives An=C, so it follows that n=C/A in the case x->0.
OmnipotentEntity wrote:
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Watched this recently. A very fast-paced TAS and the game is pretty good (though Chapter 7 is long).
Looking forward to 100%, but feel free to exclude golden strawberries (all that does is force you to run chapters multiple times).
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
2018 rankings (list and translations of titles)
Some notes:
- I know nothing about Romancing Saga so unfortunately some translations may be really wrong. It is not something I can figure out on my own.
- The Pickup Special has three weird entries, one which is the July 2018 rankings video itself, one which is a comparison of over a dozen Dr. Mario videos, and one which is literally titled "草" (which I have translated as "lol").
- Some videos are repeated between both Pickup Special and the main list. I guess anything goes when it comes to Pickup Special.
- There are only 6 Dr. Mario videos, so at least Dr. Mario isn't hogging the whole thing again. You can hope that there won't be any in 2019.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Warp wrote:
In one of his latest videos, blackpenredpen states that you can't use L'Hopital's rule in order to solve lim(x->0) sin(x)/x because "that would be circular reasoning". I didn't really understand why.
If all you want is the answer to "What is lim(x->0) sin(x)/x?", then sure, there is nothing whatsoever that stops you from using L'Hopital's rule to evaluate it.
The reason he talks about "circular reasoning" is because when someone brings up lim(x->0) sin(x)/x, the question isn't so much "What is lim(x->0) sin(x)/x?" so much as "Why exactly is lim(x->0) sin(x)/x=1?" If you bring up L'Hopital's rule, the implied question then becomes "Why exactly is the derivative of sin(x) equal to cos(x)?". And to prove that the derivative of sin(x) is cos(x) and the derivative of cos(x) is -sin(x), you first need to show that lim(x->0) sin(x)/x=1 and lim(x->0) (cos(x)-1)/x=0, without using L'Hopital's rule.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
I find that the normal encode is too fast and the slowed encode too slow. Is there one at a speed that is intermediate to the two?
Anyway, nice find on skipping Voltman's stage.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
I didn't expect you to use a water hazard the first time! (Before then I thought you were going for lowest score.) I like the creative strategies you used to save time, like hitting the flagstick.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Oh, I didn't even know there were this many improvements, and that's even without considering the Hard Knuckle glitch.
By the way, I really recommend encoding at 720p (minimum height 720); it is easily doable (especially since this is GB) even with my 8-year-old computer. This will allow Youtube to display it at 60fps, which is very important for a number of games, especially Mega Man games.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Thanks for the TASScript; it really helped to see the exact inputs. That being said, the "slowed encode" seems pointless to me since it is not a slow video of the normal encode (which I thought it was at first), it doesn't even use the same commands, and I still can't read it. Personally, I'd just dump the contents of the entire console input/output as text.
I can't really say that this was entertaining in any way (similarly for a number of hyperspeed DOS TASes). However, it is complex enough that I can admire the amount of work that went into optimizing this. It also helps that I like adventure games, which are known for this type of complexity.
On the other hand, if we look at, say, CD-Man, I don't see this kind of complexity; the only thing I am left admiring is the author's apparent masochism.
(Submission descriptions are very important in influencing opinion. This is true especially for hyperspeed DOS TASes.)
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Just a few seconds in and it already looks like a Metroid reskin.
The TAS looks pretty good to me. Near the end is where things get very interesting, with all the Drone Teleport going on.
Editor, Experienced Forum User, Published Author, Skilled player
(1941)
Joined: 6/15/2005
Posts: 3247
Hello Alyosha. The first half of the run looks nice. Though I'm not sure if this is faster (touching the secret square removes the walls):
In the second half, it looks like there are some places to improve (especially using deathwarps). Can input easily be hex-edited into the run?
I also appreciate that you ended the run on the good (2P) ending, which I had never seen before until now. There is a 2P RTA at speedrun.com (one of the runners is the_kov) but that run kills off one of the players on the last level in an attempt to save time.
which simplifies to
(*)
Setting equal to C (the goal) and solving for n, we get:
(**)
For the example that you gave, plugging in C=1000000, A=2000, x=0.01 gives approximately 180.07 months, so it takes 181 months (~15.08 years) to reach the goal.
But that's not your actual question. The actual question is to prove that the formula tends to C/A when x->0. We could use L'Hopital's rule on (**).
Or instead we could just see that (*) approaches An as x->0. This is because ((1+x)^n - 1) / (1+x - 1) is precisely the value of the derivative of z^n at z=1, by definition. This gives An=C, so it follows that n=C/A in the case x->0.
In the second half, it looks like there are some places to improve (especially using deathwarps). Can input easily be hex-edited into the run?
I also appreciate that you ended the run on the good (2P) ending, which I had never seen before until now. There is a 2P RTA at speedrun.com (one of the runners is the_kov) but that run kills off one of the players on the last level in an attempt to save time.