You might think that adding infinite jumps wouldn’t make much of a difference compared to the regular speedrun, and that most levels could just be flown through, maybe being a minute or two faster than the base game. But with The Lion King (SNES) it’s completely different! The TAS for the original version barely gets a sub 11, and here we get a 6:26! (TAS timing).
(Every time a God Mode–only trick is mentioned, I’ll say (Found by xxxx))
Category Explained
In God Mode, we call the Bug Toss / Bug Hunt games the “Bonus”.
The No Bonus category is just beating the game without collecting the bugs that lead you to those games.
Kid Simba Segment
The Pridelands
The ceiling hitboxes there are messed up allowing me to pass through and for some unknown reason this makes the Hyena jump instantly. (Found by TASGrau) Right after the fight I do a glitch to skip the cutscene.
This exists in the base game, but is harder here because the higher jumps make performing a low jump trickier — it has a 1-frame window.
A lot of RAM addresses in this game are linked, and Simba’s vertical velocity happens to be connected to the flag that triggers the post-boss cutscene. By grabbing the ledge on the right from below, after beating the hyena, Simba’s vertical velocity “tricks” the game into skipping the cutscene as soon as it starts.
Can’t Wait to Be King
One of the only two levels where you can just fly through.
By doing a precise jump at the start you can save a few frames (While is not God Mode-Only was suggested by Tompa)
When the first monkey tosses Simba to the second one, there’s a tiny moment where no monkey is holding Simba, but he still has the speed from the toss. Since nothing is holding him, he can jump and fly through the level way faster. (Found by TasGrau)
The Elephant Graveyard
If you ignore the starting hyenas and go left, the end of the level is unloaded — so skipping seems impossible.
But going right and falling down to the area you normally climb activates the trigger that loads the upper part of the level. From there you can spam jump to avoid dying to the water and finish the level normally, saving time compared to the normal path on botton left. (Found by TasGrau)
The Stampede
It’s an autoscroller.
Simba’s Exile
Last one you can fly through. Normally you have to jump four times to beat the level, but with specific inputs on certain positions you can do it on three, saving time on the fall. (Found by me)
Hakuna Matata
Interesting stage. At the start, I do two jumps to skip one waterfall and partially the next to save a few frames, since waterfall speed is slower than Simba’s.
By precisely one-frame jumping with Simba you can grab a ledge down the stage and get below the water trigger. With that you can use the ground as ceiling and avoid dying to the void. To escape you need to take damage, luckily, you can use the frogs there to take damage as when the sprite gets bigger the hitbox gets too, barely getting below the water trigger and allowing you to proceed and being a few frames faster than the regular RTA route. (Found by Samsara)
I clip in the wall right before the monkey fight by getting simba in a very precise position with a well-timed left tap to save a few seconds.
I roll to the right at the end of the monkey fight to avoid lag frames.
Adult Simba Segment
Simba’s Destiny
When Simba takes damage, he can clip through ceilings for a brief moment. In the base game this isn’t very useful, but in God Mode you can jump during that window to not only jump higher but also clip through ceilings. (Found by Moxiemay99)
When leopards/hyenas get tired, Simba can throw them with X, and this movement actually makes you clip through walls. That’s used to skip the first leopard spawner.
Going to the vine with the "Bonus" bug and getting hit by the monkey projectile is faster with a bit of luck (Route found by Samsara)
Here is isn't done as fast as possible but it saves time later on as it sets a faster RNG on Be Prepred that compensates this and it ends up saving time.
Be Prepared
I start by clipping through a wall using the throw mechanic and go straight to the end of the level, RNG was manipulated to clip faster. (While not God Mode-Only, SBDWolf helped with ideas on how to manipulate the RNG as i can't do it the same way as done in the base game's TAS)
The end of the level works in cycles. Normally, you wait for all geysers to erupt five times before a platform falls, and the left one sends you to Simba’s Return. But it was discovered that after each geyser erupts at least once, if you go to the right geyser and then right after it erupts you go all the way to the right, their values start flickering between 1 and 0 as they are put in a waiting state. This allows you to manipulate the geysers in a way that the leftmost geyser erupts earlier by reaching 0 first, saving ten eruptions if everything lines up. (Original explanation was wrong, corrected with Akiteru's help)
Simba’s Return
You can combo hyenas/leopards by jumping on them, but if you don’t press anything, don’t mash fast enough or finish the combo Simba enters a damage state (even without losing HP). That state can also clip through ceilings.
By going into the right cave and clipping through the ceiling, Simba skips most of the caves and goes directly to the last one. You need to do it this way, otherwise you won’t kill the hyena and the game softlocks, since all hyenas must die before the changing caves. (Found by TenMicu)
In the last cave, I use specific movement to bait all four hyenas into the same spot for a quadruple kill instead of killing them one by one.
Pride Rock
If you swipe precisely above an enemy’s head, you get a glitch called “glitch kill”, where Simba is put in a free state while the game thinks the enemy is being comboed. For some reason, some actions such as walking deals damage during this state. This state can also happen by bonking the enemy's head precisely.
Each Scar phase normally plays a cutscene before progressing, but if the Scar phase ends off-screen, the cutscene doesn’t play. Glitch-killing Scar lets you end the phase off-screen and save time.
I did all phases because you can’t skip Scar Phase 1, and skipping Phase 2 makes Phase 3 not work.
Samsara: Initially, I was on the fence about the optimization level of the run. Most of it seemed fine apart from a few noticeably suboptimal sections which were pointed out in the thread and acknowledged, with both
Tompa and
myself providing input files meant to be studied and learned from. I pointed out that while each new version of the run was faster than the previous, there were still regressions from section to section, similar to the way an RTA record improves. Not everything is always going to be consistently faster in an RTA run, but this is a TAS. There's no excuse for regressions like this in a TAS environment. You have infinite time to correct mistakes and infinite retry points to craft a single perfect run. You have the works of previous authors to look at and the exact inputs they put in to learn from. This is version 4 of the run and I'm still seeing those regressions. The hyena fight at the end of Elephant Graveyard was pointed out multiple times, and there are multiple input files containing more optimized versions, and yet it's still noticeably slower than it should be in this latest version.
I point out the RTA thing because in my nearly two decades of being around this community, the most common mistake I've seen from newer members and newer TASers is the mentality of treating TASing like RTA. RTA is forced to deal with pesky things such as "human ability" and "limitations" that make it ridiculously unlikely for any record improvement to be consistently faster at every possible moment. Every player is going to have their own approach to a run, they're going to have their own strengths and weaknesses, their own levels of practice and consistency, their own history of speedrunning informing their abilities. Runs have to be improved in one real time attempt, and only the final time matters when it comes to determining new records. That's what makes RTA exciting for some people. It's a grind, it's a challenge, it's active work to be the best at something.
TASing, on the other hand, isn't like that. There's no skill floor, there's no need to be good at anything, there's hardly anything that can limit you. I said it ten years ago on the former Bird Hell social media site and a hundred times since then, the only two skills you need to be able to TAS are patience and Google, or your web search of choice since Google's been sloppifying itself at a staggering rate for a very long and ongoing period of time. In other words, you just need to be able to look for information and set aside some time to sit down and implement what you find into a run. TASes are improved at the frame level, at any user-defined speed, with as many tries and retries as necessary. Because of this, a TAS can be seen as being made up of a nebulous number of miniature RTA runs, where any tiny section of a game is its own record which can and should be broken. That's what makes TASing exciting for some people. It's a puzzle, it's passive work that can be done at the author's own pace, and yet it can still be challenging depending on your choices.
The reason we put so much emphasis on optimization isn't because we're harsh and demand impossible levels of perfection, it's because we believe there's no such thing as an impossible level of perfection because anyone can attain what we're looking for regardless of their skill level. You CAN get better at TASing, but as I said before, there's no skill floor. Anyone can learn what TASing is on one day, and come back a week later with something incredible and well-optimized, just as long as they vibe with it in the first place. The process of optimizing a TAS is the exact same as making one, it's just ensuring that every single part of it is better than anything that came before it, and with the ability to fully control where and when you place each frame of input, anyone and everyone with the patience to sit down and experiment for a while can create better TASes than whatever is currently out there. This is made even more possible by the existence of every other TAS out there, coming in the form of input files that can be studied under the exact same conditions that TASes are made in. If you don't know how to do something done in another TAS, you can just open that TAS up and see the exact set of inputs needed to do it, and you can recreate them in your own TAS at your leisure.
I will say, though, that I don't fully agree with the other comments in the thread saying it needs a LOT more work. For a first publication of this hack, especially by a newer TASer, I'm perfectly willing to be lenient on minor optimization errors as long as I don't notice them while watching. A more casual viewer isn't going to notice the smaller things that the practiced Lion King TASers are seeing. The problem just remains in the bigger things, moments like that hyena fight that should have been improved but weren't, the parts that were noticeably slower than previous versions where they weren't particularly optimal to begin with. I can still see several seconds coming off of this run with a good optimization pass, ensuring that as much of it as possible either matches or surpasses everything else out there. You have access to that information, you have the input files we provided in the thread, you have the published runs of the non-hacked version of the game that you can compare to for anything that isn't being skipped by the God Mode stuff. Use it all!
If you can, I'd highly recommend having two BizHawk windows open and directly comparing two runs. If not, take note of frame counts for when certain things happen: Level starts, level ends, enemies taking damage, anything and everything that seems important. Write them down somewhere, a text file or a spreadsheet or something, and use them to divide the whole run into smaller sections. Then, just try and beat every one of those sections, noting down your new times in the same way. Then do it again if you want.
It's close, and I do think the hack is worth publishing so it's worth doing the work, it just needs that extra little bit of polish before it's acceptable.
