I redid Luigi Raceway and beat the test run by six seconds, or so. There is still a small room for improvement but I don't think there would be any drastic changes to what there is now. I didn't go for the best mylestyle at the end because I didn't have the speed and there was a trade off between that and jumping the wall closer to the finish line. Video.
Joined: 11/18/2006
Posts: 2426
Location: Back where I belong
Hahaha wow! That's very impressive xenos! And even though I do kinda miss the boost off of Wario, it's still really entertaining. Congrats on the discovery.
Behavior of Shells in Mario Kart 64:Green Shells
When you shoot a green shell, it continues in a straight line.
Collisions:
Wall: It bounces off of it and continues in a straight line. The first time it hits a wall, it loses some of its speed. I am not sure if it loses any more speed on subsequent bounces.
Items on Track: The shell and the other item are simultaneously destroyed.
Player: The shell is destroyed and the player tumbles but is not launched very high in the air. If the player is using a star, the shell is destroyed. If the player is using a boo, the shell passes through them and continues.
Red Shells
When you shoot a red shell, it continues in a straight line for a short while (maybe about 1 second) until "heat-seeking" turns on. Then, depending on what place you are in (when heat-seeking turns on, not when you originally shot the shell), the shell behaves in varying ways:
If you are in 2nd, 3rd, 4th, or 5th place, the shell begins to seek the player in the place ahead of you. The shell becomes bound to that player and then it doesn't matter what place they are in - the shell will always seek that player. The shell heads straight towards the player it is bound to. If the player is far ahead and there is a wall between you and the other player, the shell will crash into the wall.
If you are in 6th, 7th or 8th place, the shell also becomes bound to the player in the place ahead of you. However, the shell does not head in a straight line towards the player. Instead, it follows the racetrack until it comes within a certain distance of the player it is bound to. Only then does the shell seek that player in the same fashion as described in the above paragraph. This reduces the likelihood that the shell hits a wall.
If you are in 2nd through 8th place and the player in the place ahead of you is using a boo when you shoot a red shell, the red shell can't "find" that player and after its initial one-second movement in a straight line, it tries to find a player to seek, fails, and destroys itself. If the player ahead of you is not using a boo when the red shell becomes bound to them but then uses a boo afterward, the red shell destroys itself at the instant the boo is used.
Finally, if you are in 1st place when you shoot a shell there is obviously no player ahead of you. The shell just follows the track as described in the above paragraph but it is not bound to any player. It just circles the track until it collides with something.
CollisionsWall: The shell is destroyed.
Items on Track: The shell and the other item are simultaneously destroyed.
Player: The shell is destroyed and the player is launched much higher in the air than they would be if they were hit by a green shell. If the player is using a star, the shell is destroyed.
Blue Shells
Blue shells do the same thing no matter what place you are in. They behave exactly as red shells do when you shoot them from 1st place. The difference is that when a blue shell comes within a certain distance of the player in 1st place, the blue shell then becomes bound to and seeks that player. Unlike red shells, the blue shell seeks the player that is in 1st place at the time it comes in range of that player, not the player in 1st place at the time the shell was originally shot. If the player in 1st place is using a boo, the blue shell still follows the track until it comes within a certain distance of the player in 1st place. Only then does the blue shell destroy itself like a red shell does.
CollisionsWall: The shell is destroyed.
Items on Track: The other item is destroyed and the blue shell continues.
Player in 1st: The shell is destroyed and the player is launched in the same way they would be if they were hit by a red shell. If the player is using a star, the shell is destroyed. If you are in first when you shoot a blue shell, it will come back and hit you.
Any Other Player: The player is launched and the shell continues. If the player is using a star, the shell is destroyed. If the player is using a boo, the shell passes through them and continues.
Note: This is mostly from memory but I think it is all accurate. I would never have written this post if I had realized how long it would need to be to cover all the details of shell behavior. I don't think any other version of Mario Kart has such complicated shell behavior.
Silent_Slayers wrote:
That was amazing. I don't even understand how you get the red shell to hit yourself lol.
The red shell was shot backwards from first place. The shell continued in a straight line for a short while. Since it was shot from 1st place, the shell then began to follow the track and would have circled it forever but xenos jumped in front of it before it had a chance.
Joined: 6/13/2006
Posts: 3300
Location: Massachussetts, USA
The reason is because Fried's vid and Xenos's vid are both attempts at the non-shortcut flap of MR. In Mk64, most of the tracks are divided into shortcut and non-shortcut world records. MR's non-sc was in theory the least improvable, since it is the most heavily competed in and has the least room for saving time.
Xenos: Again, awesome vid, but again, I think it can be improved some. At this point, do you think you want to complete the entire run on your own using the level of optimization you've been doing, or would you be willing to work as a team that tries to optimize absolutely everything?
Emulator Coder, Site Developer, Site Owner, Expert player
(3576)
Joined: 11/3/2004
Posts: 4754
Location: Tennessee
Comicalflop wrote:
Xenos: Again, awesome vid, but again, I think it can be improved some. At this point, do you think you want to complete the entire run on your own using the level of optimization you've been doing, or would you be willing to work as a team that tries to optimize absolutely everything?
Minus the pretentiousness + rudeness, I think comicalflop is asking if you would like to work together on a MK64 TAS.
Xenos: Again, awesome vid, but again, I think it can be improved some. At this point, do you think you want to complete the entire run on your own using the level of optimization you've been doing, or would you be willing to work as a team that tries to optimize absolutely everything?
Given that this game has not had a run published, I think it prudent that the first attempt of this game should not be one of optimisation but rather a demonstration covering all known techniques with the reasoning that new strategies shall inevitably be conceived in addition to the various smaller refinements that would make such an attempt at perfection futile. We can create a set of inputs, look at it and say "that's optimal", but it won't be and I think you are being naive in thinking that you can "absolutely optimise everything" especially without already having an archetype with which to compare it. Theoretically of course, one could redo the same track a multitude of times to circumvent the need to create such a predecessor but I find such repetitive approach tedious and ineffectual in light of my aforementioned premise. In short, I could work on a team that attempts to TAS this game but not one with the specific design goals you seemingly have in mind.
Good to see so much progress.
Great job with LR xenos :-)
Weatherton: I'd like to watch your efforts for all the courses you've worked with.
Maybe I'll test Grand Prix and VS. in the near future to see if I find anything.
Myles
Joined: 6/13/2006
Posts: 3300
Location: Massachussetts, USA
xenos wrote:
Given that this game has not had a run published, I think it prudent that the first attempt of this game should not be one of optimisation but rather a demonstration covering all known techniques with the reasoning that new strategies shall inevitably be conceived in addition to the various smaller refinements that would make such an attempt at perfection futile. We can create a set of inputs, look at it and say "that's optimal", but it won't be and I think you are being naive in thinking that you can "absolutely optimise everything" especially without already having an archetype with which to compare it. Theoretically of course, one could redo the same track a multitude of times to circumvent the need to create such a predecessor but I find such repetitive approach tedious and ineffectual in light of my aforementioned premise. In short, I could work on a team that attempts to TAS this game but not one with the specific design goals you seemingly have in mind.
That's fine. I'm ok with you making a version 1; it gives me time to finish my other projects and stuff so that I can work on a v2 (with hopefully nothing else on my plate) after you finish.
Joined: 6/13/2006
Posts: 3300
Location: Massachussetts, USA
It won't be SMK until we can find some way to extend those mushroom boosts. SMK and super circuit have it, but why not MK64? :-(
Nice strat. I think the 3rd lap can be improved if you curve out more and hit at an angle more in so that rather than land, spin, then turbo start to cross the finish line, you just land on the other side of the finish line.
How do you aim those red shells? Optimizing the shooting of those is going to be a major pain I can tell.
This type of strat would only work for LR and MR as you showed, and maybe only KTB as well. Perhaps the use of red shells can allow for routes like rainbow road's.
Wow, I just read through the whole thread (I wasn't active on tasvideos one year ago), and I must say that I liked all the WIPs posted. Comicalflop, you're awesome.
I'm looking forward to this.
Joined: 6/13/2006
Posts: 3300
Location: Massachussetts, USA
I'd like to point out that the awesomeness extends to 4 other people, not just me:
-Weatherton (creation/testing of many routes, CPU jumps, glitches)
-Fried (taught me to be extreme in optimization. He and Weatherton TASed some sections of the LR TAS.)
-Xenos (coming up with newer routes and tricks)
-Adelikat (came up with memory addresses, and also was able to beat my first LR attempt, if only a small portion of it.)
First off, I just watched the trial run of the flower cup, FREAKING AMAZING!! Comicalflop, I know that it takes a long time, and people complain (myself included a little), but the entire point of this site is tool-assisted speedruns, not tool-assisted prettyfastbutnotreallyruns, and I think you whole-heartedly embrace that concept. Keep up the amazing work!
I am curious as to what WIPs are up, I looked back about 8 pages and couldn't find any of completed cups, can someone post the link or quote the post with the link? Thanks.
I produced a test run (though not intrinsically improvable) of KD to test one of the two feasible routes on this course. While the difference between the two is slim, I personally would favour this one. Convenient clickable hyperlink.
I really thought i knew Mario Kart 64 pretty but i still have absolutely no idea what happend in that video :D Defintely one of the greatest WTF feelings ive ever gotten from a TAS =)
THe idea of a full game TAS just seems to get more awesome every day =D
