Submission Text Full Submission Page
This TAS of Super Mario 64 aims to grab all 87 trees in the game in record time. The run utilizes TAS-only tricks and ignores the RTA ruleset for entertainment reasons. The final time for this run (based off VI/s) is 8:28.483. The encode provided includes a 2.733 second timeloss in Wet-Dry World, hence the time discrepancy. You can find a video of the improved segment in the detailed breakdown.
Disclaimer: This was submitted by kierio04 with express permission from (and submission comments written by) the main author, frKieran

Game objectives

  • Aims to grab all trees in the fastest time
  • Takes damage to save time
  • Heavy glitch abuse

Comments

Work for this project was started on October 31st, 2024. It helped me improve my skillset as a TASer and I believe most movement in the run is completely optimized. There are segments in this run replicated, hexed in, or referenced from runs made by other TASers, who will be listed below:
  • Krithalith - Castle Grounds beginning BLJ, scattershot VCutM BLJ, (replicated) SSL beginning movement, THI SBLJ. (Death%, 1-Key, SSL IL, THI IL)
  • Crackhex - Movement to enter SSL. (16 star SBLJless TAS)
  • Manama - Triple jump glitchy wall kick in THI
  • FeijoadaMolhada - Beginning movement of WDW (Quick Race Through Downtown IL)
  • ChouxZi3 - Beginning movement of WDW, (replicated) SBLJ strat from HMC entrance area to the BitFS SBLJ spot
  • Jeaze - I used the routing from some of his TASes as reference, namely CCM.
  • 1-Key Team (Krithalith, MKDasher, SilentSlayers, Alexpalix1, Crackhex, Superdavo0001, tesserakt, Plush, snark, IsaacA) - VCutM BLJ, MDS, basement water enter, BitFS course & fight

Route

The TAS begins in the Castle Grounds, where a large number of trees are. I then go to the basement using the Moat Door Skip (MDS) from the 1 key TAS. In the basement, I visit SSL and BitFS (giving me the key to upstairs). I briefly visit DDD to pause exit and go to the lobby, before visiting BoB, CCM, and WF. I then go upstairs to visit SL, before pause exiting back to the lobby (skipping the necessity of collecting a star). I do this again for WDW and THI, before going to the Castle Courtyard to finish the TAS.
CourseTrees
Castle Grounds26
Shifting Sand Land1
Bob-omb Battlefield17
Cool, Cool Mountain13
Whomp's Fortress1
Snowman's Land9
Wet-Dry World2
Tiny-Huge Island2
Castle Courtyard16
TOTAL87

Detailed breakdown

Castle Grounds

We start off the run in this stage, where we immediately head over to a BLJ spot instead of going to a nearby tree. The reason is simple, we need a path we can follow in a line to grab every tree with no backtracking. The trees on this map can be thought of being in 3 different groups. Starting near the middle one, we can go to the left one, which requires us to back track all the way over to the right group, and vice versa. This is way we do a BLJ to the back of the right group. The BLJ uses pause-buffered inputs to build up as much speed as we can on the tiny slope. We end up being in the long jump state next to the first tree with hyperspeed, and something interesting happens, we instantly leave the tree and conserve our speed in a different direction. To understand why, we must look at how grabbing a tree works. If Mario and a tree (or any pole)'s hit box collide, Mario's X and Z position will be modified to be the same as the tree/pole object. However, his Y position can remain the same, which causes a bug when Mario grabs a tree below its base Y position. Instead of updating to the grab pole action, Mario instead either becomes idle or enters freefall, depending on how close the floor is below him. If he enters idle, he can instantly change his direction. But being set to idle sets your speed to 0, so how do we conserve speed? Simple, we press Z on the frame we go idle. See, sliding speed conserves through certain actions like grabbing a pole, and is updated whenever horizontal speed is. However, in this case, sliding speed isn't reset when we become idle, so we can start sliding with the speed we had earlier. Grabbing a tree in this way also has the benefit of converting your negative speed to positive depending on the angle you slide out. It may not look like we slid here because it only happens for 1 frame. We initiate a slide kick from this crouch slide which also instantly lands, so we then do a forward rollout, which again instantly lands. We now would enter walking here which caps to 48 speed, but instead we start turning around. Turning around usually depletes your speed quickly, but since we have such a large amount and there's no speed cap on this action, we can conserve our speed over to the 2nd tree at the back. Once we get there, we walk for 1 frame and dive below the tree, initiating another instant-release (the name of this exploit)
After we do that, we expertly conserve speed with instant releases on the remaining trees leading up the final one in the right group of trees. We then head over to the BLJ spot used in 1 key, except here we use it to butt slide off the bridge and instant release the tree at the back of the left group. We grab all the trees in that group, and then the middle one. Once we get to the final tree, we then do the BLJ used in 1-key to enter VCutM.

Vanish Cap Under the Moat

The movement here is the same as 1-key, we long jump over to the elevator and BLJ to millions of speed. The moving elevator displaces us to the death barrier. Once we exit VCutM, we arrive near the waterfall in the castle grounds, in a swimming state, one of the states that doesn't reset sliding speed. Once we get up to the waterfall, we re-activate our speed by pressing Z, and navigate parallel universes to start walking behind the moat door, allowing us to enter it without collecting the basement key from Bowser.

Shifting Sand Land

After navigating the basement, we enter Shifting Sand Land. We use the strat for Shining Atop the Pyramid to conserve our speed on to the ! box. Instead of jumping into the star, we strain a bit left in order to walk for 3 frames to jump dive on the side of the pyramid, making sure that when we rollout we have the perfect angle facing the tree. Once we rollout, we do a low-mag punch to conserve our speed after landing, and do a few frames of turning around to do the same thing. We initiate a sideflip, which immediately collides with the sole tree in this course. Since grabbing a tree with less than 10 speed imitates a faster pole grab animation, and the first frame of a sideflip gives you 9.15 speed, we grab the tree quickly. We jump dive off the tree and immediately die to exit SSL, since pause-exiting would take us to the main lobby (we don't have a basement key to get back)

Bob-omb Battlefield

The routing in this level starts out the same as Jeave's, but changes quickly since we don't have access to the wing cap. Despite that, we still get pretty close to the same time. We grab the trees where the wing cap would be, then BLJ to the tree next to the bridge. We can't instant release it since we always fall at the same Y level and same Y level rate, unlike when you have the wing cap. It's theoretically possible to grab the tree two ones after that since it's on a sloped surface, but it doesn't matter since we are just waiting on an elevator cycle. We start the BLJ on the elevator just barely within time. We then do a BLJ into a hyperspeed instant release, into a slide kick that we jump dive out of. Jumps get higher the more speed you have, so we are fast enough to land on the island. This landing is incredibly precise, requiring you to be neutral while landing, floor clip into the island without getting your speed negated by the ceiling, etc. After that we grab the tree on the island, and the remaining ones behind the mountain.

Cool Cool Mountain

The routing in this level is the same from Jeave's TAS, with a few exceptions. Besides small time save improvements, we don't dive early on one of the trees since the cannon lid is still present. We also land closer to the last tree and side flip into it to save some time.

Whomp's Fortress

We jump kick to the nearby tree and side flip into it. We can grab it earlier with a long jump, but as mentioned earlier, grabbing a tree with less than 10 speed allows us to pause exit earlier since the animation is shorter.

Snowman's Land

The route here is the same as Jeave's, but execution is pretty different and we save a great deal of time. We do two long jumps and a jump dive on the sloped surface to reach the tree quicker, we do a sideflip to get a fast pole grab animation on the windy tree, we do a jump kick and low-mag punch instead of sliding to get to that tree, and again do a fast pole grab animation at the highest tree.

Wet-Dry World

This is the most complicated strategy in the entire run. We start off with the same movement for the IL TAS of Quick Race Through Downtown, glitchy wall kicking our way up to the water crystal on the tall white platform. We initiate the BLJ in the same way, this one being brute-forced to be optimized for the water crystal touch later on. This BLJ depends a lot on the tilt of the platform to keep you grounded, as well as your facing angle, which makes it especially difficult. There is really only one PU we can access this way since we also can't walk into it, we'd fall off. We have to land in a PU at the same Y level or 78 units above it such that we land the long jump at that point in the PU. Once we do that, we do some parallel universe movement I came up with to first jump into the tunnel that loads the city area of the map, then walk into the water crystal on the main map. Our Y level here has to be the floor the water crystal is on. Since object interactions are processed before the water surface upward, we can trigger the water crystal while being upwarped by the cage. I put a lot of work into trying to find a solution where Mario is facing in such a way he can merely swim out of it, but I couldn't find anything. The next-best thing was to water jump out, and do two long jumps to instant release the trees that most people don't even know exist. I pause exit a few frames earlier by entering first person first frame we land, which is an allowed action to pause exit from.
As explained earlier, this section was improved between encoding the full TAS and submitting it here. You can watch the improved segment below:

Tiny-Huge Island

We start this course off with a triple-jump glitchy wall kick to reach the higher portion of the map, a strat found by Manama. I then quick turn and long jump twice to the remaining tree. This tree isn't on a sloped surface but I still manage to do an instant release, due to me grabbing it while I am falling. I jump kick out of the tree, quick turn, then triple jump dive, into another jump dive, to enter the warp pipe. The reason I can't just triple-jump initially from the tree is due to the fire spitter. Also, the warp pipe's hitbox is entered by only 0.2 units, a feat that took me a few hours to brute-force. Once we arrive at the huge island, we start an SBLJ to quickly get over to the next tree. We have to do a dive and instant dive recover, otherwise the landing frames don't allow us to turn in time to navigate the narrow bridge. We crouch to lessen our speed and jump backwards to grab that last tree.

Castle Courtyard

We start by going over to the right tree, which is faster from my testing. Some simple movement that's been explained earlier by me, nothing too special here. The last tree grab can be done quicker, but timing ends with last input, which is the frame I press B to start the jump dive. After that, Mario goes neutral, so he loses some speed.
NOTE: In the original encode, the TAS ended on last input at the cost of a later tree grab, and was timed as such. This was adjusted when preparing it for submission, so an additional 17 frames were added.

Other comments

There might be an improvement in Castle Grounds, but that would be incredibly hard and time consuming. I will probably do an All Signs run next.

CoolHandMike: Claiming for judging.


CoolHandMike
He/Him
Editor, Judge, Experienced player (990)
Joined: 3/9/2019
Posts: 815
Lol Mario is a tree hugger. This was very entertaining and well made! Good job!
discord: CoolHandMike#0352
Spikestuff
They/Them
Expert player (2778)
Joined: 10/12/2011
Posts: 6514
Location: The land down under.
Unique TAS. Not interesting for me personally. Voted Meh. Good for Playground. Would've been a good candidate for April Fools submissions.
WebNations/Sabih wrote:
+fsvgm777 never censoring anything.
Disables Comments and Ratings for the YouTube account. Something better for yourself and also others.
Joined: 1/14/2016
Posts: 108
It is hilarious to me to see such a long line up of authors, see it's an SM64 TAS, and then see the "all trees" category. I did have to look at the calender twice, as it seems a very good joke. The joke is made more perfect by extensive author's notes, treating the whole business most seriously; that they reference one "Jeave" and his TASes, who does not actually exist, hightens the tension (the actual youtube channel is "Jeaze", btw). That the actual TAS presented here seems of high quality is important to the bit as well. I cannot but vote yes in confusion, because the one question remains, and that is why "all trees"?
Player (81)
Joined: 7/7/2008
Posts: 875
Location: Utah
This is such a stupid idea for a TAS… When can we get it published? :D
Reviewer, Skilled player (1244)
Joined: 11/18/2011
Posts: 364
Location: Morocco
This would have been awesome for April Fools.
I still learn more about English. https://www.youtube.com/user/McBobX100
I wrote:
Working is the best way to achieve goals in speedruning. Hardworking is a pain.
kierio04
He/Him
Joined: 12/25/2018
Posts: 15
Location: New Zealand
Hey everyone, just here to clarify some things!
  • This was originally planned to be submitted in December, so its proximity to April Fools is merely coincidental
  • As such, this is a 100% serious TAS, not an April Fools' joke ("All Trees" is a category extension for many games, including SM64)
  • Only one person (Kieran) actively worked on this TAS; the large number of authors is a result of recycled inputs from pre-existing TASes (eight are from the 1 Key TAS alone)
Site Admin, Skilled player (1208)
Joined: 4/17/2010
Posts: 11630
Location: Lake Char­gogg­a­gogg­man­chaugg­a­gogg­chau­bun­a­gung­a­maugg
MovieRules wrote:
The movie must be complete Your submission must play the game from the beginning, and must finish the game, or reach the most suitable endpoint the game allows. Level selects, single-level movies, or otherwise incomplete movies are not allowed. Examples of suitable endpoints are:
  • A definitive ending, such as a credits sequence.
  • A kill screen, assuming it is impossible to complete even when TASed.
  • If there's no clear ending, [...]
  • For non-standard goals, in addition to any of the above, the movie can also be ended at some logical or artistic stopping point that makes sense.
If the movie doesn't complete the game, it may be ended according to the last point in that list, but it isn't automatically considered as valid. There needs to be some explanation why it's equally good as completing the game. Playarounds are the primary example of when it's okay. Their goal is maximal entertainment, so when the author has accomplished everything creative they could think of, going out of their way to show an actual ending may damage entertainment, because it may only be a part of a story mode, while the movie uses tournament that has no ending. Or if the movie is a showcase of arbitrary code execution, game completion becomes kinda meaningless because it can artificially be triggered at any point, but even then it's better to have some kind of closure than to just end input mid-game. At the first glance, I don't see why ending input mid-game is an essential (logical or artistic) part of the "all trees" goal and why game completion would make it worse in any way.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Bigbass
He/Him
Moderator
Joined: 2/2/2021
Posts: 215
Location: Midwest
feos wrote:
At the first glance, I don't see why ending input mid-game is an essential (logical or artistic) part of the "all trees" goal and why game completion would make it worse in any way.
Personally, I feel that given the goal of the TAS is to grab all trees, going out of their way to do something that doesn't further that goal in any way is entirely unnecessary. The stated goal is to "grab all 87 trees in the game in record time," not to "grab all 87 trees, and beat the final boss, in record time." Would game completion make it worse? Perhaps. It would certainly lengthen the TAS, which hurts the "record time." The extra time needed to complete the game isn't related to grabbing trees, yet it would still become part of the TAS. Game completion solely for the sake of game completion doesn't add any artistic value in my opinion. Especially for this game, simply completing the game isn't special or unique (and thus doesn't contribute toward entertainment or artistic values.) Perhaps if the completion was performed in an uncommon way, like how "game end glitch" movies complete a game through unconventional means, then it could further an artistic goal, but still not the artistic goal of this movie.
MovieRules wrote:
  • For non-standard goals, in addition to any of the above, the movie can also be ended at some logical or artistic stopping point that makes sense.
If the artistic intention of the non-standard goal is to grab all trees in the game, then it makes sense to me that the artistic stopping point is whenever that goal is complete.
TAS Verifications | Mastodon | Github | Discord: @bigbass
Site Admin, Skilled player (1208)
Joined: 4/17/2010
Posts: 11630
Location: Lake Char­gogg­a­gogg­man­chaugg­a­gogg­chau­bun­a­gung­a­maugg
That kinda implies that only a few other Alt goals should bother completing the game, because [2977] NES Super Mario Bros. "maximum coins" by TEHH_083, HappyLee & CuteQt in 26:10.25 could have just ended on collecting the last coin, and [4444] NES Super Mario Bros. "minimum A presses" by HappyLee, Kriller37, DaSmileKat, Kosmic & periwinkle in 10:24.39 could have ended right before the first required A press to truly minimize their count. Publication goals have always implied completion, even casual naming of some categories still has that implication in the form of the % sign slapped together with random words that actually define a given category (the most canonical example being "any%", meaning fastest completion). I can see how "fastest crash" contradicts completion. There's also #9270: InputEvelution's Flash Papa's Donuteria "Rank 5" in 10:56.27 that uses the dominant RTA goal but not something intuitively acceptable on tasvideos, and it was allowed for Alt because otherwise the run would have to be several hours long at minimum and there was a bunch of negative implications from that. The goal of only a few cycles makes sense as a workaround. But with this movie, I don't see what fundamental problem we're working around.
Bigbass wrote:
Game completion solely for the sake of game completion doesn't add any artistic value in my opinion. Especially for this game, simply completing the game isn't special or unique (and thus doesn't contribute toward entertainment or artistic values.) Perhaps if the completion was performed in an uncommon way, like how "game end glitch" movies complete a game through unconventional means, then it could further an artistic goal, but still not the artistic goal of this movie.
MovieRules wrote:
  • For non-standard goals, in addition to any of the above, the movie can also be ended at some logical or artistic stopping point that makes sense.
If the artistic intention of the non-standard goal is to grab all trees in the game, then it makes sense to me that the artistic stopping point is whenever that goal is complete.
IMO you're reading that rule backwards. Its point is not to allow game completion as an option if it makes sense as a part of the main goal. Its point is to require some closure, most traditionally game completion, but there are other options if their reasoning can be articulated and agreed upon. If there was a poll about whether we should consider game completion fully optional and up to the author, even if only for Alternative, I'd vote No. It's already optional in Playground, which is fine for all kinds of wild experiments. The only problem with it is that it never got as much promotion on the site as regular publications, but every time I try to dedicate time to that, other site duties take over (or burnout). Even then it doesn't mean that PG should fully merge with Alt, because there are still technical rules that make publication problematic.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Bigbass
He/Him
Moderator
Joined: 2/2/2021
Posts: 215
Location: Midwest
feos wrote:
That kinda implies that only a few other Alt goals should bother completing the game, because [2977] NES Super Mario Bros. "maximum coins" by TEHH_083, HappyLee & CuteQt in 26:10.25 could have just ended on collecting the last coin, and [4444] NES Super Mario Bros. "minimum A presses" by HappyLee, Kriller37, DaSmileKat, Kosmic & periwinkle in 10:24.39 could have ended right before the first required A press to truly minimize their count.
You're not wrong. Although, those submissions also included completing the game as a goal. The first chose to complete the game in addition to the primary artistic goal. While the second was to minimize the number of A presses needed to complete the game.
feos wrote:
Bigbass wrote:
Game completion solely for the sake of game completion doesn't add any artistic value in my opinion. Especially for this game, simply completing the game isn't special or unique (and thus doesn't contribute toward entertainment or artistic values.) Perhaps if the completion was performed in an uncommon way, like how "game end glitch" movies complete a game through unconventional means, then it could further an artistic goal, but still not the artistic goal of this movie.
MovieRules wrote:
  • For non-standard goals, in addition to any of the above, the movie can also be ended at some logical or artistic stopping point that makes sense.
If the artistic intention of the non-standard goal is to grab all trees in the game, then it makes sense to me that the artistic stopping point is whenever that goal is complete.
IMO you're reading that rule backwards. Its point is not to allow game completion as an option if it makes sense as a part of the main goal. Its point is to require some closure, most traditionally game completion, but there are other options if their reasoning can be articulated and agreed upon.
Nothing in the way the rule is written indicates to me that its purpose is to require closure (game completion) beyond the non-standard goal(s) of the movie. Maybe the rule's wording needs some improvement to clarify its intent. Right now, it reads to me like as long as the goal is non-standard (grab all trees), then it's allowed to end based on something logical or artistic (grab final tree). I'm curious how others interpret this rule, especially non-staff or people reading our rules for the first time.
feos wrote:
It's already optional in Playground, which is fine for all kinds of wild experiments. The only problem with it is that it never got as much promotion on the site as regular publications, but every time I try to dedicate time to that, other site duties take over (or burnout). Even then it doesn't mean that PG should fully merge with Alt, because there are still technical rules that make publication problematic.
Apologies if there was some confusion on my part. Originally it sounded to me as though you were saying the movie was invalid because it chose to end the TAS early; sorry if I misunderstood. I'm not arguing towards any particular publication class, I leave that up to the judges. Rather I'm voicing my opinion that it appears to be a sensible artistic choice to end the TAS when the goal is completed.
TAS Verifications | Mastodon | Github | Discord: @bigbass
Site Admin, Skilled player (1208)
Joined: 4/17/2010
Posts: 11630
Location: Lake Char­gogg­a­gogg­man­chaugg­a­gogg­chau­bun­a­gung­a­maugg
Bigbass wrote:
feos wrote:
That kinda implies that only a few other Alt goals should bother completing the game, because [2977] NES Super Mario Bros. "maximum coins" by TEHH_083, HappyLee & CuteQt in 26:10.25 could have just ended on collecting the last coin, and [4444] NES Super Mario Bros. "minimum A presses" by HappyLee, Kriller37, DaSmileKat, Kosmic & periwinkle in 10:24.39 could have ended right before the first required A press to truly minimize their count.
You're not wrong. Although, those submissions also included completing the game as a goal. The first chose to complete the game in addition to the primary artistic goal. While the second was to minimize the number of A presses needed to complete the game.
Does it make more sense for them to include it, unlike for this movie? And more importantly, why? EDIT: A good example is pacifist movies. If one absolutizes the goal alone at the cost of game completion, a lot of pacifist movies would have do end by dying on the first boss, simply because they'd have to break its own goal of pacifism to complete the game. Instead the consensus seems to be allowing kills that make the game beatable. With "all trees" tho, I can't even say that completing the game would defeat the purpose of this goal, it'd just expand it to include an essential part of a publishable TAS.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Joined: 1/14/2016
Posts: 108
I think that this movie is essentially different than the two referred to SMB movies. During normal gameplay in SMB, one collects coins, and jumps mostly all the time. Getting coins is rewarding by giving more lives to reach the end goal, and (less importantly) by giving more points if one is interested in a high score. Jumping helps with getting to the end of the levels and to safe the princess. So far, so obvious. The coin maxing goal is interesting in that it maximises this side-goal of normal gameplay, while the ABC, in having an extra challenge, goes for a minimalistic approach. (edit: I would understand pacifist challenges to be interesting for reasons similar to this.) The current run, SM64 all trees%, has none of this. Climbing trees in sm64 does not at all inherently bring you closer to the end goal of beating Bowser. It is not something you usually do during gameplay. There is nothing interesting in beating the game while maximising or minimising anything to do with trees. All trees% is completely uninterested in anything to do with beating the game, and that's why it would be a disservice to require a normal ending. All trees% is interested in trees only, and the movement between them. If there is any merit in this goal, it is (besides the humourous absurdity) in the joy of the movement of Mario. The one thing missing in this submission is therefore not an ending to the game, but an ending appropriate to what the category is all about. As a quick idea, something like a flourishing jump of the last tree, showcasing some tree-movement there wasn't time for earlier, but that is now earned. Maybe think of it this way. Most TASes play a game, like sm64, that runs on a specific console. It makes no sense to require the movie to beat the console. This TAS plays a game, hugging trees, that runs on Super Mario 64.

1743415013