Tool-assisted game movies
When human skills are just not enough

Submission #2811: Kirkq and Nitrodon's SNES 7th Saga, The "any%" in 1:50:17.43

Console: Super NES
Game name: 7th Saga, The
Game version: USA
ROM filename: 7th Saga, The (U) [!]
Branch: any%
Emulator: (unknown)
Movie length: 1:50:17.43
FrameCount: 397046
Re-record count: 216794
Author's real name: Kirk Iler and Jason Schoenfelder
Author's nickname: Kirkq and Nitrodon
Submitter: kirkq
Submitted at: 2010-08-18 19:02:54
Text last edited at: 2010-09-15 04:29:26
Text last edited by: Mister Epic
Download: Download (14780 bytes)
Status: published
Click to view the actual publication
Submission instructions
Discuss this submission (also rating / voting)
List all submissions by this submitter
List pages on this site that refer to this submission
View submission text history
Back to the submission list
Author's comments and explanations:

7th Saga in 1:50:17 by Nitrodon and Kirkq

  • Aims for fastest time
  • Takes damage to save time (I suppose this is expected in an RPG.)
  • Contains a Death Warp
  • "Non-glitched"

This run foregoes an infinite stat-boost glitch discovered during the making of the run. The run was around 75% completed when we discovered this glitch. The glitch can be well defined if we classify this submission as "Does not walk through walls." A run executed with the glitch would approach 1 hour and 30 minutes because the point it occurs at is quite late in the game.

This movie was recorded in Snes9x v1.43+v11 (beta15 + lua 0.06) It is also known to play with Snes9x v1.43+v13 (lua 0.05) The movie will not play back correctly in certain emulator versions; it will probably desync around frame 240,000.

This run is around 9.5 hours faster than any single segment completion of this game. We have spoken with the author of that run, and consider his run to be very well executed given the difficulty of this game. This run may in fact be the fastest completion difference that exists between a TAS and a single-segment run. (Feel free to correct me if I am wrong. - Kirkq)


Part 1
Part 2
Part 3
Part 4
Part 5
Part 6
Part 7
Part 8
Part 9
Part 10

Nitrodon’s Comments

I started hacking this game sometime in 2005, before I even joined TASvideos. (In fact, it was the first game I ever hacked.) When the subject of TASing it came up, I quickly came to the conclusion that the final boss would probably require a level somewhere in the 30s. While this is considerably less than what is needed in normal play, it still meant that the grinding involved would be excessive, and thus any TAS would probably be too boring for the viewers. Still, I supported any TAS attempts in hopes of being proved wrong. Finally, in early 2009, Kirkq found that Gorsia could be defeated at a level in the early 20s, which meant that the required grinding would no longer be excessive. After numerous route discussions, we decided to collaborate on this TAS.

Thanks to Dr. Fail and Nati for the ROM data and Game Mechanics guides on GameFAQs.

Kirkq’s Comments

This game is perhaps one of the most unforgiving RPGs of all time. You have to grind for hours. When playing for speed, you will enter fights that are probabilistically impossible to win. You will die often. There are no shortcuts. I can elaborate on this for quite a while, but I’ve briefly addressed some of the grinding requirements in this post.

Here is a rough timeline of my experiences with the run. I first TASed this game in December 2007, eventually performing a test run up until R Pison. Around November 2008 some speculation began again on the thread and continued through February 2009. In February 2009 Nitrodon and I began formulating a route. This took a substantial amount of testing and planning to see what sort of levels and supplies we needed to beat bosses. Around March 2009 we had a complete route for a TAS. We reached Pandam in the run in June 2009, revised the route, and I killed the S Brains by August. Due to the RNG manipulation needed to beat the Dragon, progress was halted for much of the school semester and then resumed in February 2010. The last few bosses or the run were completed over the summer of 2010. I apologize that progress was strung along over such a large time span, but it became very difficult for me to work with such tedious manipulation in large doses during school.

My future TAS projects at this point include first finishing up Bomberman Hero (Okay, Kyman?) and then working on Golden Sun and a separate personal TAS, which I may keep to myself until completion.

Special Thanks to:

Nitrodon: I had no clue what I was undertaking when I started this run and it quickly became clear that I alone would not have been able to achieve the level of manipulation we did in this run. Nitrodon provided all of the game mechanics information necessary to allow the run to be played with a high level of precision. While I may have done around 60-70% of the testing and execution, Nitrodon supplied 90% of the information. It was a pleasure working with you, and I learned a lot that will help me to tackle future runs. Hopefully we both learn to procrastinate less in the future.

Healblade: Your spreadsheet was extremely helpful in planning the run.
Nati: For making a great Game Mechanics guide for reference on GameFAQs.
Starion: For collecting the images of all the hidden items in the game.
Sp00ky: For providing the strategy of killing the Trick early on.

Thanks to my bros Erokky, BagOfMagicFood, Atma, and everyone else who chills in #rockman

Lastly, thanks to everyone who posted in the thread or on the videos. It was nice to know that people out there appreciate our work and enjoyed seeing this game slaughtered. It would’ve been hard to finish if we didn’t know that people were awaiting the final product.

THE RUN ITSELF: (Written by Kirkq)

General Discussion: Overworld RNG Manipulation

On the overworld (or to a lesser extent in a dungeon) the RNG will increment very radically. Differing step patterns create differing enemy patterns which cause differing incrementation patterns in the RNG. Changing the path from point A to point B, while still maintaining a shortest path, is the best way to prepare the RNG for the next use. Brief pauses can be a very useful way to only lose a few frames while preparing the RNG. A suitable RNG is selected for engaging boss battles and recruiting apprentices with high stats.

General Discussion: Boss Battle RNG

Once a boss battle is entered, the RNG is no longer random; it is entirely deterministic. It will ONLY change based on character and enemy actions. Enemy action types are on a purely deterministic number generator initialized at the beginning of the game and dependent only on previous enemy action types. Essentially the boss is set on using a certain attack type unless you adjust the previous battle. We can only adjust the increment by one action at a time. Typically, one extra action in the previous battle means that the next boss will start on what would’ve been his second attack.

Brief Narration of the Run:

Game Start

A couple frames are wasted before the file select screen to manipulate apprentice positions and join values off the start. Lux the Tetujin is chosen for various reasons. He is easily the strongest early game character. He and Wilme the Alien are the only characters that can beat Pison and Red Pison at low levels. Lux also gets a large stat boost at the end of the game to help with the final bosses. After some testing there was no doubt that Lux is the best main character. Selecting the name "E" wastes no frames since there is one frame for free directional input before the 'A' button can be pressed.

Lemele Castle

A mirror is grabbed in the castle. A woman in town gives us a Potion 1, and we buy 2 B Power (doubles power stat excluding gear) with our starting gold. These will all be of use to us later.


Every dot on the minimap (crystal ball) is a random encounter. Nitrodon wrote a very neat script to walk a shortest possible path from point A to point B, reloading when monsters were encountered until a successful path was found. This was very useful in the production of the run. It was later modified to help radically manipulate the RNG through brute forcing while walking to the next destination.

Castle Aran

We grab an Opal on the way to the boss, Romus. Romus can kill us in one hit, but we make it two to increment the enemy action value an extra time.

Lemele Castle/Rablesk

Well we died and are put back in Lemele, let’s try something else and head to Rablesk. There are hidden items scattered throughout the game. We acquire a potion 1 on the way into town. By dying to Romus, the game feels sorry for us and lets us acquire the Whistle in Rablesk. We grab a hidden Protection seed; this will raise our guard by 4 when manipulated. We sell the opal and purchase a B Protect (double guard stat excluding gear.)

Castle Aran

We use the Protection Seed while manipulating the RNG for the boss just a bit. The RNG we enter/end the (one turn) battle on determines our stats at the end. The whistle kills Romus in one hit gaining us two levels. It is much faster to kill Romus with the Whistle than it is to level up enough to fight him. We look to acquire primarily power and speed in this run. When leveling, Lux can get anywhere between 3-5 power and 2-4 speed. We get max power and speed from these two levels. Romus drops the Key of Earth, which opens the door in the next cave. We expend about 30 seconds acquiring a Power Seed (raises power by 4 when manipulated) from a chest in the basement and a Potion 1 nearby. It is faster acquiring these potions than making the money to buy them at this point.

Cave of Earth

We grab an Agility Seed (raises speed by 4 when manipulated.) Speed is a very necessary stat. Crossing a threshold in speed can mean the difference between a 30%, 50%, or 61% hit rate against a much faster boss. Inside the next chest we open is an enemy Trick. A Trick is next to impossible to beat at level 3. It would kill Lux in one shot, and it has a 98% chance to connect. The only way to kill it is B Power, Defend, Attack, Defend, Attack. Defending increases the power of the next attack. (Someone in Lemele tells you this.) We would need it to miss us 4 times, which is impossible when the RNG is deterministic in battle. We use the mirror from Lemele Castle to reflect a petrify by the trick, and it is manipulated to last 5 turns. This was sp00ky’s idea originally. The trick gets us to level 4 and drops a random gem. We manipulate the 1/8 chance for an emerald (sells for 5000 Gold.) We heal before engaging the boss Pison, our arch-nemesis or something. The Pison fight is fairly straightforward and not too probabilistically unreasonable compared to the trick fight. It gets us to level 5.


We need to find some guy named Gain to get the map. He is busy running around town avoiding us. On this quest to find Gain, we recruit Wilme in the weapon and armor shop. His location and response was manipulated at the beginning of the game, as we haven’t fought enough battles to change it since then. Apprentices are recruited with level-representative semi-random stats at your level plus or minus one. Thus, we recruit Wilme at level 6 with the best stats possible. Wilme’s stats were based on what RNG we entered town on plus each increment by an NPC walking. Pauses in towns are used to manipulate the RNG either for apprentice stats or to get an NPC to move a different direction. Gain gives us the Map, this is required to advance the plot (or lack thereof.)


In the Inn we recruit Brantu. He is necessary to advance the plot.

Melenam Cave

We acquire the Wind Rune, 1 of the 7 runes we set out on a journey to collect, remember? The Wind Rune allows us to teleport to any town we have visited. Brantu opens the way to Melenam.


We wind rune to Rablesk to go on a shopping trip. Buy 2 B Protects, sell 1 emerald. Wind Rune Lemele. Buy 7 potion 1, 4 Potion 2, 5 B Power. Potion 1 heals 40, Potion 2 heals 90.


We wind rune back to Melenam. There is an item here called “Brwn” or Brown Coat hidden on the floor. It is a mild upgrade to Lux’s armor. We forego it because it provides very minimal gains. It really amounts to 2-3 less damage per hit after everything is accounted for. After the next fight, 2-3 less damage is nothing. We heal to prepare for the next fight.

Red Pison

Didn’t we kill this guy like 5 minutes ago? This battle was traced BACKWARDS from the 1/256 chance of getting 3 +5 Power levels and 2 +4 speed levels. It is extremely difficult to gain an edge in this fight. Lux can only take one hit, Wilme can take two if he is defending. Both characters have a 50% hit rate, and we must defend then attack to deliver a successful blow. We use all of our resources to barely beat Red Pison at level 5 Lux and level 6 Wilme. We kill off Wilme to funnel the experience to Lux. Lux is now level 8. There is a hidden M Water under Red Pison we grab to revive Wilme later.


Talking to the Sage blocking the door assigns which apprentice we have to fight later in Patrof. You can only advance the plot if you have the Wind Rune when talking to him. The apprentice’s level is assigned to be 5 above yours at the time you talk to the Sage. The Patrof apprentice is never the main character, the partner, Esuna, or Lux.


We enter Pell to put it on the Wind Rune list and head to Guntz. We grab a Power Seed in the Guntz Elder’s house. We use various items while waiting for the Elder to walk towards us. He has the Water Rune, but won’t give it to us right now.

Patrof/Patrof Cavern

We must go into Patrof Cavern to get into the Castle. In the cavern we grab an Agility Seed and a Life Recovery. A Recovery is a limited quantity item in the game only found in dungeons. It heals max HP and MP for one character and is quicker to use than a potion. We use this later in the game primarily for the MP recovery. Towards the end of the Cavern the RNG is tightly manipulated to make the apprentice walk towards us correctly and engage us on a specific RNG. We kill the enemy apprentice Valsu in 2 hits and acquire the Star Rune. (2 of 7) The Star Rune is free B Protects in battle.


We go to Pell to recruit the Digger Quose who digs for water in Guntz allowing us the get the Water Rune from the Elder. (3 of 7) The Water Rune heals a random amount of HP in battle less than a Potion 2. The only consideration necessary for which of the Water Rune and Star Rune to get first is which one we want higher in the inventory.


We recruit Perole in Bone. This only occurs once we have both the Water and Star runes. We grab an Agility seed and Wind Rune to Bonro. By having Perole, we are able to take a boat to Pandam on the northern continent. (Various things can occur here with various characters, but having Lux and taking the boat to Pandam is the fastest.) Since the party leader determines the scenario, one could kill off the main character temporarily to force a different scenario. Tricking the game logic can create a humorous scenario with Lux as hero and Olvan as partner. The player starts by recruiting Perole. One can then kill off Lux and take Olvan’s route to the north continent. This allows Perole to follow you for the rest of the game, and he can be present for the ending sequence.


We acquire the Ring in the inn to later sell for 6000 gold. We then exit Pandam and head to Brush. We leave Brush and head to Tiffana. In Tiffana we purchase M Water and 9 Vacuums. Vacuums have a small chance to OHKO random encounters. We can only hold 9 of any item.

Grinding session:

We kill 27 S Brains with 27 vacuums in 3 trips for about 13 minutes manipulating the power and speed to be high levels. We choose to grind near Brush because the enemy spawn rate is higher than around Luze. An S Brain encounter occurs 1/8 battles around Brush. A vacuum has a 30% chance to connect against S Brains multiplied by our attack success rate based on speed (71%-98%). Accounting for all of this gives a (1/8)*30%*71% = 2.6% chance of success. We must also take into account that the overworld RNG is on a continuum, so we must manipulate the RNG in between each battle to get to the next successful kill. Sometimes after entering a fight, we have to wait for the overworld RNG to increment on a ~1.5 second timer, so we mess around in the menu. In the middle of this we we were required to set up apprentice locations and apprentice reactions. We we able to manipulate the RNG to give us the perfect apprentice locations value out of 4096 possible combinations. On the first trip we make our way towards Bilthem. The encounter zone walking towards Bilthem ends roughly where we stop engaging S Brains, so we avoid enemies the rest of the way. Perhaps we could’ve walked to Brush from Tiffana while engaging S Brains, but due to the mountains and water enemies don’t spawn as much. Our encounter rate doing this would’ve been lower thus losing most (if any) time to be gained. We grind S Brains around Brush from level 8 to level 20 with some extra experience left over. After the grinding is complete, we teleport to Pandam and head towards Padal.


In Padal, we buy potion 2s, B Agility, and B Protect. Kamil is recruited at level 21 in town 0 steps off of our path. We speak with Commander Prosa in Padal to force one of three scenarios with the Moon Rune. We refuse to help him, because we can. We grab another recovery on our way out.

Baran Castle/Serpent

We grab a potion 2 and a recovery. This is our first real fight since we’ve leveled 12 times, so we are actually dealing some real damage now. Lux gets to level 21 after we kill off Kamil. There is a sapphire on the ground under Serpent, but we don’t want it.


This is where the RNG began to get quite painful. We need to force the incrementation to adjust by around 800 after beating Serpent in order to set up for the next boss. After the next boss we need to set two apprentice join values and set who gets the Sky Rune. Commander Prosa is now in Telaine. He gives us the Moon Rune (4 of 7) for defeating Serpent. The Moon Rune acts as an infinite supply of B Power.

Cave to Beore/Luze

Still fighting the RNG every step of the way, we kill the Dragon in Luze. If we were to search the pedestal, we would learn that the Sky rune has been stolen, but we already know this. The Sky Rune always randomly gets assigned to an apprentice who is not the main character, the partner, or the Patrof apprentice.


We make a side trip to kill 2 Brains, acquire 2 recoveries, and just enough experience to level later.


We Wind Rune to Bilthem Castle, the place we walked while grinding S Brains. Olvan conveniently joins us with the Sky Rune. (5 of 7) There is a high chance the Sky Rune apprentice will fight us, but we used this as an opportunity to get a new apprentice. The Sky Rune recovers MP when used in battle. We grab an Exigate in town for use later. Now we look to overthrow the evil emperor Doros. In Bilthem Castle we grab a Harp (blocks an enemy vacuum), a Potion 2, the Jail Key, the Star, and the amulet. The Star is the only way to kill Doros. The amulet is an accessory that blocks all spells except vacuum, which we make good use of later. After a fight with Doros (killing off Olvan of course), we acquire the Light Rune. (6 of 7) As a side note, when acquiring runes at this point in the game, there is a very small chance that our apprentice, if alive, will engage us after the battle and try to kill us. The Light Rune increases a character’s magic power when used in battle. We Wind Rune to Brush.


We must talk to the Fortune Teller to advance the plot, Brantu (remember the guy we took to Melenam?) has built a flying machine to take us to the next continent. He couldn’t finish it until we were done grinding and collecting 6 runes though.

Valenca/Cave of Kapel/Guanta

Once we crash-land in Valenca we wind rune to the entrance and head towards the Cave of Kapel. At the end of the Cave, both characters are cursed and lose the ability to use magic. We enter Guanta to teleport back to later.

Cave of Silence

We collect an M Water and find that Pison guy again. This time he is Metal Pison. He spends most of the battle trying to lower our defense and eventually remembers to raise his own. When playing this game normally, one can stall M Pison pretty heavily just by countering his debuffs with the Star Rune every turn. Killing M Pison takes us to level 22, which bumps our next apprentice up a level. This is the main level the run planned for. We acquire the Moonlight in the cave. The kind folks at Guanta would now remove the curse they placed on us, but that won’t be necessary. Lux’s magic is pretty terrible at these low levels anyway. We Wind Rune to Patrof.


We extort Olvan for every penny he is worth by stealing his gear, and then we recruit level 23 Lejes who is conveniently standing right where we need him. Lejes wasn’t cursed so he still has magic.


Monmo the dragon is the next boss. He is practically immune to everything but fire. We use the Light Rune to increase Lejes’s magic power and the Sky Rune to constantly recharge him. This is quite a long fight. Because this fight had such limited strategy, Nitrodon was able to solve for a flawless fight with a script. Later in the dungeon we finally fight Gariso. Gariso possesses the final rune, the Wizard Rune. We must use the Moonlight to make Gariso visible and vulnerable. Gariso foregoes an action every time he turns invisible, so it is to our advantage to use the Moonlight as often as possible. Lejes’s F Shield allows one character to completely block a magic attack (Not Vacuum or Effect spells.), making Lejes invaluable for the rest of the game. We allow Gariso to heal at the end of the battle so that he doesn’t use more dangerous actions. Healing is a bit separate from the traditional action increment. We brought all the runes together and of course something terrible has to happen. After the fight Gariso/Gorsia/Lemele knocks us out causing a fate we will learn later.

Pharano/ Ligena

Well I guess the game isn’t over yet. Now our runes have been taken away, making the game even harder. We purchase Potion 2, Potion 3, and B Power. Potion 3 heals max health. We grab the hidden recovery near the shop and exit town. A long walk to Ligena later we talk to the village elder and learn that Gorsia knocked us 5000 years into the past. We manipulate the hidden cave in Ligena to be in the closest room, and sneak aboard the rich guy’s airship. A couple slightly different scenarios can occur here with different characters. We are on our way to the wealthy city of Melenam 5000 years ago. Wait, wasn’t this city in a cave underwater?


Lux gets a power boost in Melenam. We forego the "Infinite Stat Boost" glitch we discovered when we reached this part of the run. We head to Palsu to get a letter for Tafuro in Melenam. We buy the best weapons in the game in Melenam for Lux and Lejes. Tafuro takes us to talk to Dr. Fail to discuss his project to fight Gorsia. With a name like Dr. Fail, nothing could go wrong, right? All we have to do is leave the lab and talk to one person in town and things have already gone bad. Foma will wreak havoc on the world unless we destroy him now. This battle involves a 50% hit rate with Lux and Lejes and a lot of F Shield. The battles in this game now all require max buffs and plenty of F Shield. Any magic attack at this point in the game will most likely kill us. Foma was set up with a pretty terrible string of magic attacks. This is offset later by some actual good luck. The city sinks underwater like we found it. We escape on the airship.

Barrier Cave

Two bosses are in here. First we fight Goron with a lot of F Shields. Lux gains level 23 off of this battle. We then fight Griffan. Griffan had an extremely favorable attack pattern. She does nothing but attack and petrify for the battle. She has a fair chance to use deadly magic attacks, but the sun hath shown towards the end of this run. Saro is the legendary guy of the past who has been fighting off Gorsia. Now he leaves his problems to us. We death warp back to the airship.


In the Airship we purchase some final supplies and head to Guanas to get the Runes back. The Runes do nothing now except weaken Gorsia. We forego the Sky Rune, which would decrease Gorsia’s magic power.


We grab the Power Seed for the final fight. Our damage is low enough in this fight that 4 power really does help. In fact it saves about 7 turns and a couple potions. We kill Gorsia in an extremely long fight with 50% hit rates again. We have to use 6 of the runes on him to even have a chance.


We kill Gorsia. Gorsia kills us. "E" is reborn again as Lemele, who originally turned out to be Gorsia. Apparently a robot ("tetujin") can be reborn into a human. Nothing has changed except Lejes is stuck standing 5000 years in the past. Don’t succumb to the dark power again. If you want to try to process what exactly this means, maybe you should try reading this highly theoretical thread on GameFAQs


We planned extensively for this run, and it was very difficult to precisely predict our stats and how well we would be able to manipulate the final battles. At the end of the run, we outdid our expectations to the extent that one of our earlier decisions proved flawed. It ended up such that it was not necessary to return to the airship for a shopping trip at the end of the game. Since this was finalized so late, and since we did not have an exigate to leave the penultimate dungeon, we were already committed to returning to the airship. This lost us around 40 seconds. If we had needed even one more potion for any of the final three bosses, the shopping trip would have been absolutely necessary and we would've made the correct choice. With even more insane manipulations and boss executions, a non-glitched run could feasibly be achievable in under 1:45, but not much less.

Suggested Screenshots:

71284: Lux attacking Red Pison with 1 HP and a dead partner.

FractalFusion: I'll file this one under "must have played it to enjoy it". The "encounter tracker" made it a bit more interesting, I admit. Accepting for publication.

Similar submissions (by title and categories where applicable):