Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Kirkq wrote:
About 6000 frames are spent killing Romus in Dromiceius's test run at level 3 as opposed to 9000 spent acquiring the whistle and starting grinding at level 3. It seems faster to me.
Unless you have a way to get to level 3 in 50 seconds, and account for the gold spent on B Protect and B Power, getting the whistle is fastest by far.
Please watch the WIP at the top of this page and read through our various route posts, I assure you this has all been calculated.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Any time we want an event to occur on a specific RNG, we have to polish the RNG up to that point. Varied walking patterns and pausing the game (costs minimum 4 frames) in an area with monsters will fairly drastically change the RNG to a more favorable state.
Things manipulated thus far:
-Wilme is manipulated to be in Bonro Weapon/Armor Shop waiting to join upon the second start press of the game (only costs 3 frames surprisingly).
-Romus is manipulated to kill us in two hits the first time to set the trick's first attack to petrify. (The next enemy action is directly based on the previous.)
-Romus is engaged the second time to allow for the most optimal set of two levels possible (there are only 256 possibilities when not on the overworld).
-The Trick is engaged such that he will stay petrified enough turns, and give us an optimal level and an emerald while dying in a timely manner.
-Pison's battle gives us another great level. We selected the RNG we wanted to end on out of 256 and found an attack combination that produced it.
-In Bonro the RNG is polished throughout town such that Wilme is level 6 and has good stats, and various people are manipulated out of the way.
-In Zellis people are manipulated out of the way, and Brantu is manipulated to walk towards us.
-The RNG is rigorously polished up until R Pison to get us the exact value we selected out of 256. We selected the value to end on and traced about 17 rounds backwards to produce the levelup. After determining the fight from the RNG, we then purchased the exact amount of potions and went to engage the fight we seletced. The Red Pison fight took me over 5 hours after Nitrodon had compiled all the information, and made his own attempts. It was traced backwards from the levelup. Nitrodon probably spent more than that on his own attempts to figure out all of the subtle behaviors of the RNG based on actions.
-In Eygus, Valsu is manipulated to be the apprentice, and he is manipulated to have low stats to die more easily.
-In Guntz we manipulate the P Seed to give +4 Power before the Elder moves, then the RNG is such that the elder moves towards us after the P Seed.
-Patrof Cavern floor 2 - the pause is necessary to get by the monsters
-Patrof Cavern end of floor 2 and all of floor 3, the RNG is polished to allow Valsu to die in two hits.
-In Bone Perole in manipulated to walk towards us.
Right now we are trying to produce an optimal two levelups off the first vacuumed S Brain.
(1/8 chance to engage 1 S Brain, 71% chance for a normal attack to connect, 30% chance of Vacuum success, and RNG increments based on waves hitting the shore.)
That is approximately the luck manipulation.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Pressing comma and period will display the input and the current frame, I probably should've mentioned that.
It seems autofire functions as so:
Frame Advance, Set autofire, Frame advance ....is the same as
Frame Advance, A, Frame advance, A, Frame advance.
Frame advance = "no input + Frame advance"
A = "A + Frame advance"
Autohold seems to function the same. Neither skips a frame, and begins input on the current frame.
There isn't really a "before and after frame X". It's more, the button is held during frame 200 or during frame 201
It isn't recommended to use autofire in a TAS, because you often lose one frame where you could have pressed the button but didn't happen to. It's great for testing though.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Zero100 wrote:
1. Most importantly, what game should I do? I'm thinking Golden Sun, but you guys say the RNG is hard to manipulate. If not, I'll most likely hope to do a gba game, and/or an RPG that has a easy to manipulate RNG that hasn't been done before.
The Golden Sun RNG is extremely rigid from what I've read. Every action has some set consequence. I personally would like to see a TAS, though much of it would be story. (It takes at least 45 minutes to get out of Vale.) Any contribution to such a run would be a good one. You may just want to start with a test run, get the hang of rerecording and perhaps find a couple tricks, route suggestions, and recommended leveling. The final product would need to be well routed and well optimized, but tests can contribute knowledge and motivate others to contribute as well.
You may want to try to frame optimize periodic segments for practice, but any complete optimization in a test run is just tedious. Here's a non-TAS run being working on.
http://www.youtube.com/watch?v=gHaFG32B3nwZero100 wrote:
2. Where's the best place to learn HOW to TAS? I think that there might be a good from-scratch FAQ somewhere around here, but I'm not sure where to find it.
Here's the emulator.
http://code.google.com/p/vba-rerecording/
VBA 21.zip if you're on Windows.
We can't tell you where to get Roms at TASVideos, don't ask.
I do a lot of real time testing and then frame optimize later.
Start by recording a movie.
F1-F10 are your savestates.
In Read Only mode, you can view the video you are making and make relevant savestates.
When not in Read Only mode, you can alter your movie. Loading any savestate in "Recording" mode whether the savestate was created in Read Only mode or not will continue recording your progress from that savestate.
When switching between modes "Stop Movie" as a precaution.
Use frame advance, real time, pausing, and reloading to do testing and execution. (A finished TAS should be almost entirely done in frame advance, though walking in Golden Sun can be done in Real Time.)
If you make some progress feel free to upload in one of the existing Golden Sun threads. (Always use existing game threads if they exist.)
Zero100 wrote:
3. What programming languages will I need/Where can I learn to work with Hex? I know Java, but I'm not sure that it can make programs that seek for hex patterns in games.
You don't really need programming to make a good TAS. Programming offers helpful tools to make your life easier, and sometimes test exhaustive possibilities.
Look into memory watch to find variables to determine what fight is upcoming, and perhaps read damage formulas. I know the Golden Sun RNG has a lot of documentation, go check out GameFAQs.
Check out my post in this thread for a little bit on memory watch: http://tasvideos.org/forum/viewtopic.php?t=8169
As you may or may not know, modifying addresses can be good to test possibilities, but is not allowed in a finished product. (The movie will not synchronize for others if addresses were directly modified in the recording.) It is always good to view relevant addresses. Examples may include finding a value to determine what the next battle will be, finding values to determine miss values and critical hit values, a value to determine when the next battle will occur. These values are somewhat difficult to find, because you need to accurately project their behavior.
If you want to get a little more in depth, check out Lua.
http://tasvideos.org/forum/viewtopic.php?t=6539
EDIT: Any game with a difficult RNG isn't exactly friendly for a new user to start out with. I recommend playing around with games you enjoy to contribute some knowledge and progress. You can probably contribute useful knowledge and tricks to Golden Sun, but I doubt you can produce a sufficiently optimized publishable TAS for Golden Sun at this point in your career. So mess around with it for a while and see what you can come up with, or feel free to pick a more friendly game to start out with.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
This route has been modified to allow buying of the Kryn Fist, and neglecting of any armor. The Kryn fist changes Lux's projected damage against Gorsia from about 74 to about 97.
We concluded armor wasn't necessary. It saves around 5 damage per hit when Gorsia deals around 50 when defending of our Lux 200/Lejes 150 health. Armor was originally thought necessary due to spell resistance, but we later realized that it wasn't necessary when abusing F Shield. We therefore only need to steal Olvan's gear, because money is not nearly as much of an issue any more. We knew these were possibilities, but we knew there would be no route changes necessary until Pandam.
Here is the revised route. It probably needs some editing for clarifications that are currently in the back of my head but not in this post.
Pandam-
Collect:Ring (Ring = end game insurance money.)
Wind Rune to Entrance
Exit
Tiffana-
Purchase 9 Vacuums
Wind Rune to Tiffana-
Engage and Vacuum 9 S Brains while walking around Tiffana
(Level = 15)
Wind Rune to Tiffana-
Purchase 9 Vacuums
Engage and Vacuum 9 S Brains while walking towards Brush
(Level = 18)
Enter Brush
Wind Rune to Tiffana-
Purchase 9 Vacuums
Wind Rune to Brush-
Engage and Vacuum 9 S Brains while walking towards Bilthem
Walk the other 1/2 of the path to Bilthem, Enter town
(Level = 20)
*Apprentice Responses Change after S Brain 24* (Kamil = Join)
*Make sure 5121 (Locations) is at a value beyond M Pison.*
M Water Wilme
Recruit Kamil (Lv. 21)
Wind Rune to Pandam-
Padal-
Purchase: 9 B Agility
Purchase: 7 Potion2
Purchase: 8 B Protect
Purchase: 3 M Water (Serpent, Dragon, Doros)
Collect: Recovery (Check which first)
Activate Prosa (Say no.) (Check which first)
Wind Rune to Tiffana-
Exit
Baran Castle-
Collect: Potion2
Collect: Recovery
Kill Serpent (Costs 2 B Power.)
(Level: 21)
(Let Kamil Die.)
M Water Kamil
Wind Rune to Padal-
Exit
Telaine-
Walk to Prosa (Around the left?)
Collect: Moon Rune
Wind Rune to entrance
Cave of Beore-
Exit to Luze
Luze-
Kill Dragon (No items)
(Let Kamil die.)
(Do not check pedestal)
Sky Rune must be in Olvan possession.
*Apprentice Responses Change* (Olvan/Lejes = Join)
-----Somewhere after Dragon but before M Pison-----
Wind Rune to Pell-
Kill 2x Brain (Manipulate 2x Recovery)
---------------------------------------------------------------
M Water Kamil
Recruit Olvan (Lv. 22)
Collect: Sky Rune
Wind Rune to Bilthem-
Collect: Exigate
Doros Castle-
Collect: Amulet
Collect: Star
Collect: Potn2
Collect: Harp?
Kill Doros
(Costs 4-5 Potn2, 2 B Agility)
(Let Olvan Die.)
M Water Olvan
Wind Rune Brush-
Fortune Teller Rarsa (100 G)
Brantu - Flying Machine
Valenca -
Exit
Cave of Kapel -
(Loss of Magic)
Exit
Guanta -
Enter/Exit
Cave of Silence -
Collect: M Water
Kill M Pison (costs nothing)
(Level: 22)
(Let Olvan Die.)
Collect: Moonlight
Exigate outside
M Water Olvan
Wind Rune Bonro
Sell Olvan's Gear directly.
Recruit Lejes
Wind Rune Guanta-
Exit
Gorfun-
Kill Monmo
Kill Gariso (Costs: 1 B Agility?, 4-5 Potn 2)
Pharano-
Collect: Recovery
Purchase: 8 B Power
Purchase: Max Potion 2
Purchase: Max Potion 3
Purchase: 2-3 M Water
Ligena:
(Manipulate Airship)
Melenam:
Get Lux a tuneup
Exit
Palsu:
Get letter, Exit.
Melenam:
Purchase: Doom Sword, Kryn Fist
Tour Lab
Talk to one person on the way out.
Reenter Lab
Kill Foma (costs 2-3 Potn3)
Airship:
(Hopefully not buying anything)
Exit.
Barrier Cave:
Kill Goron
Kill Griffan
Talk to Saro
Death Warp (lose half gold)
Airship:
Purchase Max Potion 3
Purchase 2 B Agility
Village: Collect Runes (not Sky)
Gorfun:
Kill Gorsia
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Open VBA/game
"Cheats" in the emulator context is setting memory addresses.
Cheats -> Search for Cheats
In Cheat Menu -> Start (Sets original memory address values to reference)
Check Box for Search Updates Values (Updates values every time you search.)
Walk one step right, search "Not equal to" "Old value", walk one step right, search "Not equal to" "Old value" advance a few frames holding still, search "Equal to" "Old value"
After numerous searches like this, you should eventually get narrowed down to 20 values or so, one or more of which reflects your x position. (You also need to find your y position by walking only up and down.)
When you find the address you want to set, go to add cheat, set the value you have inferred for the position you want, and enable cheats. (The option is "disable" so uncheck it.)
I didn't revise all of this, but it should give you the correct idea of what you need to do.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Perhaps a valid option would be renting the $30-40/month server for a few months to make sure that it is what we want, and then perhaps attempting to raise the two year fee from donations a few months in.
I'd gladly throw $20 in every once in a while to fund my hobby, and I'm sure many others would do the same. Paypal is probably convenient for most people.
The user database will probably only expand with the development of new emulators, so I don't think lack of motivation by players to keep the site funded will be an an issue in the foreseeable future.
I'd like to thank those taking the initiative to make this transition for TASVideos, there aren't exactly a lot of people capable of doing so.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
I'm 95% confident it isn't faster in the any%. I'm not sure the time you chose to bomb the switch in your platform was optimal, I didn't look that close. Perhaps setting a bomb and waiting for it to blow up will make the platform better. It's 3 extra warps, an out of the way jump, a climb, and another jump compared to riding the platform up and jumping in the green warp.
May be useful as an option for the 100% route, I'm not sure that it will be applied in either route, but still another impressive find.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
That is in fact the Crysta Hole exit, so it's not exactly a route change in that regard.
The damage can be taken off the lava or a self bomb, I will check which is faster when I make it that far. It is also possible to fall down without landing on the top platform.
It seems I will be (probably?, I'll know after frame perfecting it.) be down to two health at the end of the level. I will either have to make up the health in Crysta Hole, or in Death Road (I may be able to grab the extra health in Crysta Hole depending on the platform cooperation, or I can just not take damage in Death Road.)
Previously I had this:
Kirkq wrote:
Health Rationing:
Groog Hills - Take one-two damage (2-3)
Erars Lake - Two Hearts (4)
Water Pool - Take one damage. (3)
Dark Prison - Take one damage (2)
Hades Crater - One Heart (3)
Magma Lake/Magma Dam - End these with 3 health having visited only one restoration area.
Crysta Hole - One Heart (4)
Death Road - Bomb the first bomb over the portal without taking damage, for the second portal jump in and take damage. Take one damage (3) Grab the heart at the end of the level. One Heart (4)
Float Zone - Take three damage (1), One Heart (2), Take one damage (1)
I was previously bombing the dam off of the relatively nearby yellow block, this will most definitely be around 5-8 seconds faster. Definitely another great find ZAR, I was barely able to make that jump in real time with multiple reloads.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
I'm pretty sure you completely overwrote the movie. If the movie file size is something like 10 kB or much less than it was before, you probably deleted it. When you clicked record movie, it basically started that filename over from scratch, I think. If the filesize is comparable to what it was, there may be some way to recover it.
Alternatively, if you had made a savestate error, I'm quite sure you would've been able to recover most of your movie.
I think all you can do is make more backups in the future as you progress, it's a good habit to get into.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
I don't think this game provides anything great through tool assistance. I would guess that someone could easily get within 5 or 10 seconds or your completion time with enough practice.
I agree with these statements:
Xkeeper wrote:
The music isn't bad and it isn't that long. Too bad the AI isn't seen at all after the first few seconds, though. :(
Ferret Warlord wrote:
I appreciate your enthusiasm, but this game is dull to watch. Basic steering, no crazy physics to abuse, very monotonous.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Slowking wrote:
Deku Stick: You could overwrite them and buy them back later. But that would require money and buying 4 sticks probably takes more time than jsut getting the Bremer Mask.
Could you just keep one bomb or bombchu and overwrite that before buying more?
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
I'm going to vote meh.
mmbossman wrote:
3) an abundance of Super Metroid material already published
mmbossman wrote:
4) an inability for this run to provide reasonably different material from those movie already published.
These two reasons seem to be the most relevant, no one really seems to question the entertainment and technical precision.
I still see this run comparable to the 16 star run. Stipulation based on previous knowledge, or "somewhat human feasible route".
All I have to say is that if this run does get accepted, we are opening the door for more variations in categories for various games, which could be good or bad.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Where exactly would you draw the line for Super Metroid runs?
(Glitched = Out of Bounds used.)
We have:
100%, Unglitched, Real Time
Any%, Glitched, Real Time (also low%)
Any%, Unglitched, In-Game
Any%, Unglitched, Real Time,
You want:
Low%, Unglitched, "Real Time while In-Game time"
What is stopping:
100%, Unglitched, In-Game
100%, Glitched, In-Game
100%, Glitched, Real Time
What if there becomes a split creating these:
Any%, Glitched, In-Game (also low%)
Low%, Unglitched, In-Game (Hypothetically, humor me.)
Why not publish while we're at it:
Low%, Speed Booster route, In-Game
Low%, Speed Booster route, Real Time
100%, Unglitched, In-Game, Full map completion
100%, Unglitched, Real Time, Full map completion
100%, Glitched, In-Game, Full map completion
100%, Glitched, Real Time, Full map completion
I've listed 8 arguable categories the site could/does have. If the runs are well played why not have 16 categories? The rules to these categories are all clear and easily defined. Should we publish every hypothetical run for every game?
I neither support nor condemn this run, I'm just curious how many "well-played" categories you expect to be hosted.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Two pages ago has the link to ThMrksMan's Progress. I'm too lazy to encode my run, so you're quite out of luck until my run hits the workbench for publication.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
I've never played this game. From viewing, I think both categories for this run would have pros and cons. This run takes some interesting shortcuts to save real time, and doesn't spend up to 3 seconds waiting for a better rotation. I'm fairly confident there would be a lot of 2 second wait times in an in-game time run, which would detract from its entertainment. At the same time the in-game run would be comparable to world records.
We're missing bonus levels? Projected completion time of those?
I'm sure this game should have a publication, but I'm not so sure about the category. If created, the in-game run should take damage where it makes the ending clock time faster. I don't really see any reason to have a completely no damage run.
I think this run should have two categories, and if that is the case this should be published. If I had to choose only one category to be published, I'm not sure if this is it or not.
Yes vote, keeping that in mind. I'm also wondering if bonus levels should be included or not.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
This run is only 9 minutes slower without using a debug cheat which would generally be frowned upon these days. It could be argued that less of the game is better for someone who hasn't played the game before, but the majority of the viewers for the run will be people who have played it. As someone who played this game through a few times in the past, the sequence of this game gets broken by usage of the debug cheat.
The site currently shows the wild warps run as a concept demo, so we technically don't have a normal run of this game. I think it would be alright to keep the current concept demo and publish this as a normal movie. Unpublishing the concept demo is alright with me, but I don't think it's really necessary.
The quality of the run isn't perfect, but for the most part it is done quite well. The routing and glitching seem optimal, but the execution suffers mildly. At the very least, I would hope that this would get rejected on the condition that a more polished run of this category in the future would be acceptable.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Yes vote, I think this run would make a good addition to the site.
I've never played the game before and the run is interesting enough to keep me entertained.
I think the argument of the popularity of the game should play somewhat of a factor. I myself had a friend ask to see a run of this game a while back, and I did not find it published. (I didn't have the sense to check gruefood back then.) FODA's youtube channel has over 22,000 views on this video, I think that speaks quite strongly of the viewers that would enjoy this run.
( http://www.youtube.com/watch?v=CuT9gQnv1Ho )
Mild movement glitches, pizza battle manipulation, and an entertaining final sidescroller make this run pretty good for a game that "mainly walks to the right."
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Voting No, I really don't think this game is worth watching to anyone who has never played it before. (I've never played it, or even heard of it.)
The technical precision of the run seems alright, there's really just nothing technically impressive going on but one jump glitch. The run is short, it does have that going for it.
A few people may appreciate this run, but it really just doesn't seem to be an entertaining game.
I would suggest better game choice in the future.
EDIT: Use the Minor Edit checkbox if you don't want your posts spamming the IRC channel.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
RNG = 006C, 006E
Pausing messes with the RNG if you pause while people are walking or while enemies are moving.
I'm wondering if it is actually faster to save in Rablesk or not. It may be faster to only visit Rablesk after dying to Romus. It is very close, I will test it tomorrow. I don't know why I didn't consider this before.
EDIT:
Latest WIP: Buying Vacuums in Tiffana
http://dehacked.2y.net/microstorage.php/info/519452733/Tiffana%20Test.smv
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
The file name may speak for itself. =)
http://dehacked.2y.net/microstorage.php/info/259136904/Gorsia%20Kill.smv
*May not sync on some SNES9x versions*
Nitrodon's pathing lua script takes me to Guanas and Gorfun. It should take care of all the pathing throughout the game once modified to work in the necessary caves/dungeons. The only thing left really is botting the fights to get a more accurate idea of what kind of potions we will actually need.
Remember I'm one level too high for Goron and Griffan. This was all performed in real time with numerous reloads. I believe this is an accurate depiction of how the final fights will go.
There seems to be exactly 3 types of menu options that change the RNG. Attack, Defend, and Item. The others overlap these. Sometimes i purposely miss attacks to force a hit to connect for the other character, etc. Attacks miss Gorsia an obscene amount at this level.
I hope this makes the route a little more convincing. I believe recruiting apprentices at level 21 is a good threshold for this run. This is what is required to get the apprentice gear boost, and thus afford the expensive gear at the end of the run. Having Lux/Lejes at 23/24 seems to be a near minimal case for a timely completion.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
Lux EXP - 62399 Currently 22
Lejes EXP - 71276 Currently 23
EXP left to distribute after recruiting Lejes: 28418
Lux to 23 - 9052
Either to 24 - 10077
Either to 25 - 10984
We are looking at Lux at level 23, and Lejes at level 24.
2 Pr Seed
2 P Seed
4 A Seed
1 V Seed
(all to Lux)
Level 23 Lux Max Stats:
HP - 194
MP - 69
Power - 115
Guard - 138
Magic - 91
Speed - 91
Add seeds and Melenam boost.
Level 23 Lux Max Boosted Stats:
HP - 228
MP - 99
Power - 143
Guard - 166
Magic - 111
Speed - 127
Average stats of Elnard Boosted Level 24 Lejes
HP - 151
MP - 122
Power - 93
Guard - 93
Magic - 92
Speed - 116
(The run only has him recruited/boosted to 23, but recruited apprentices have values slightly plus or minus average, so we can round up.)
I'm going to make a convincing .smv attempting the final three fights with these stats, and I will post my best attempt later.
EDIT:
I should've used potion 3's instead of 2's, but here is my attempt thus far. It should kill Griffan, if it dies at Griffan you aren't using the same emulator as me, we think. http://dehacked.2y.net/microstorage.php/info/1940234083/7th%20Saga%20GoronGriffan%20Test4.smv
I used turbo religiously which doesn't make the viewing all that pleasurable.
I am one level too high, because I forgot I level up after Griffan, but the difference is almost negligible. I used 3 potn2, 6 potn3, and 3 recovery. This is by far my best iteration, I am quite confident I can kill Gorsia on this attempt, but I will try later.
EDIT2: http://dehacked.2y.net/microstorage.php/info/321785173/Gorsia%20Kill%20Weapons.smv
A mediocre attempt without the armors and with the Kryn.
Experienced Forum User, Published Author, Player
(208)
Joined: 7/7/2006
Posts: 798
Location: US
I've done numerous estimating tests in real time with approximate stats we should have at that point. That's where most of the potion amounts come from. The problem is that the stat gain pool has something around 256 possibilities, only a couple of which are optimal with respect to this run. Some levelups occur in fights with numerous other scenarios needing produced as an outcome. I'm not positive what kind of stats we will have at the end of the run, or how optimal brute forcing can get a fight to be with respect to potion count.
