This submission of Archon is a 548 frame improvement over the currently published TAS ( https://tasvideos.org/2211M ) mainly due to a new route that skips a fight.
Game objectives
Emulator used: Bizhawk 2.8
Aims for fastest time
Manipulates AI
Doesn't use the dragon
MESHUGGAH did a great job with the published TAS, but in my quest to optimize/TAS games few people care about, I decided to take a crack at this one. His route manips a fight in the center and uses the dragon for the final fight. I route out that fight and take a different unit to the wizard.
The premise is the same, take all 5 power squares to win. During the first 3 turns we both are capturing the 3 squares in the center column. Here is where the route really differs. On my 4th turn, I move the Shapeshifter into range to attack but not be attacked by the Wizard. The Wizard is manipped to cast a spell on turn 5. That spell is imprison, which will either target a unit within his attack range or the strongest threat if nothing is in his range. Our strongest threat is the dragon, who is still sitting at home watching reruns of The Price is Right making this a wasted (and fast) turn.
Turn 5, we attack the wizard with the Shapeshifter. True to its name it turns into whatever enemy it's fighting, so we become a wizard. The wizard can shoot the second shot faster than the dragon (80 frames vs 120) which mostly mitigates the time spent on the shifting animation. I shoot the first shot on frame 6 of the battle because I have to aim around the tombstones. This is unavoidable in this route, unless there's a way to manip the level layout to be more open without spending more than those 5 frames getting it to happen. Beyond that, I don't think there are any frames left to save on this route.
Thanks MESHUGGAH for doing these TASes and giving me a high bar to clear! It's been a fun challenge and I look forward to bringing down more times with you.
slamo: Replaced movie with a 16 frame improvement, and judging!
slamo: Looks like a valid improvement to me, and you even have the original author praising it. Accepting to obsolete the previous movie.
I played a lot of the c64 version back in the day. now I wonder why I always used to teleport the piece I'd use to attack the light wizard over one of the other light pieces, leading to an extra fight. silly me.
Another great job at improving / obsoleting TASes, keep up the good work ShesChardcore!
- Route: Nice, didn't even knew about Shapeshifting or other features of this game because I've thought it's unavoidable to fight less than 2 times. Until now :)
- Tombstones: The battlefield is a 14x11 grid and tombstones have a variable height, assuming 4 different values: max, smaller, smaller and zero. I think they are manipulatable by frame time (don't know exact place of RNG) before engaging a battle and by doing different actions, for example using the same route but moving with a different unit to the same place of this route
Easy yes vote!
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Another great job at improving / obsoleting TASes, keep up the good work ShesChardcore!
- Route: Nice, didn't even knew about Shapeshifting or other features of this game because I've thought it's unavoidable to fight less than 2 times. Until now :)
- Tombstones: The battlefield is a 14x11 grid and tombstones have a variable height, assuming 4 different values: max, smaller, smaller and zero. I think they are manipulatable by frame time (don't know exact place of RNG) before engaging a battle and by doing different actions, for example using the same route but moving with a different unit to the same place of this route
Easy yes vote!
Thanks! I think the fight screen layout is rolled earlier than that. Possibly before the match starts. When I changed up moves or induced a different fight it would still always be that layout for the first fight of that particular game. It is likely just rolled on one of the 3 initial starting menus along with the AI movement logic.
Requesting a judge to replace the submitted file with this one: https://tasvideos.org/UserFiles/Info/637850226122032054
This is a 16 frame improvement by ending the first Shapeshifter move one block sooner. I could have sworn I tested that. I will be more careful in the future.
Tested the route, it looks like very good.
- RNG is rolled between selecting which side starts first and placement of units
- If our side starts first, enemy will be way too aggressive to make a fight in turn 4 by casting a spell
- Light side submitted version route: (being light while dark starts first) 4 frames faster (3260 versus 3264) at the start of battle however in this case, Djinn gets imprisoned, so Phoenix is used which only has "aoe dmg", not sure if it's faster, currently testing it. If your new userfile has a new strategy, it will work for light too.
edit: 3725 end because phoenix needs to go to enemy face to be able to hurt, requires 5 hit / 1 attack * 2 to kill enemy, so its much slower.
Don't know if there is any possible improvement, will check out your new userfile later today.
It's perfectly fine to find new improvements after initially submitting it. Going to say both: don't be too hard on yourself, and I really like your attitude for keep improving TASes :)
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Tested the route, it looks like very good.
- RNG is rolled between selecting which side starts first and placement of units
- If our side starts first, enemy will be way too aggressive to make a fight in turn 4 by casting a spell
- Light side submitted version route: (being light while dark starts first) 4 frames faster (3260 versus 3264) at the start of battle however in this case, Djinn gets imprisoned, so Phoenix is used which only has "aoe dmg", not sure if it's faster, currently testing it. If your new userfile has a new strategy, it will work for light too.
It's perfectly fine to find new improvements after initially submitting it. Going to say both: don't be too hard on yourself, and I really like your attitude for keep improving TASes :)
I tested the Phoenix this morning as well, the sorceress runs away into the corner and it takes forever for the Phoenix to get there and do enough damage.
I also tried going first but like you said the AI is much more aggressive and conjures elementals instead of using imprison.
edit: movie was not finishing the game
3611 User movie #637850288296553933I wrote the description there:
"Moved forward Dragon instead of 2nd row units to have a faster imprison and finish the game as usual"
notes:
- the single frame delay for starting the game before unit placement animation is the best
- and starting 2nd with dark (to avoid (edit:) fighting as phoenix and aggressive spell casting or enemy attacks) is good
- unsure, if it's possible to be light side and imprison our phoenix instead of djinn. there was a 4 frame fast for light side phoenix but it's dmg makes it worthless / super slow
I can't think of any other improvement but will think about it
edit: its possible to do it from light side and imprisoning phoenix. battle starts 3449 frame, going to upload once optimized it
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
3611 User movie #637850288296553933
I wrote the description there:
"Moved forward Dragon instead of 2nd row units to have a faster imprison and finish the game as usual"
notes:
- the single frame delay for starting the game before unit placement animation is the best
- and starting 2nd with dark (to avoid (edit:) fighting as phoenix and aggressive spell casting or enemy attacks) is good
- unsure, if it's possible to be light side and imprison our phoenix instead of djinn. there was a 4 frame fast for light side phoenix but it's dmg makes it worthless / super slow
I can't think of any other improvement but will think about it
Just looked at it. It doesn't end up finishing the game since you're moving the Shapeshifter off of the center power tile.
Oh god I've forgot it... :D
Yeah need to rethink it then
edit: my dragon route won't work, tried a lot of different delays. Don't have any idea for potential savings currently. Will retry light side
edit2: light side phoenix too slow, tried other order of placing units but your userfile seems to be the fastest (and complete) movie. Good job! :)
edit3: How nasty is that new route. Not only you bypass the enemy's aggressive spell castings and attacking chances, but also defeat him right after a wrong decision made: imprisoning your dragon. Therefore you beat him with shapeshifting, the close danger you set up.
PhD in TASing 🎓 speedrun enthusiast ❤🚷🔥 white hat hacker ▓ black box tester ░ censorships and rules...
Oh god I've forgot it... :D
Yeah need to rethink it then
edit: my dragon route won't work, tried a lot of different delays. Don't have any idea for potential savings currently. Will retry light side
edit2: light side phoenix too slow, tried other order of placing units but your userfile seems to be the fastest (and complete) movie. Good job! :)
Thanks as always for looking into it! Always good to have multiple eyes on it.
GG, but I feel like the NES version is one of the lower-tier versions of the original A800 port. Here's what I know about the differences between various versions:
A800: original. Also the absolute fastest version of them all.
C64: Phoenix attack stretches further to the right, though I'm not sure if that's its true hitbox or just some graphics jank.
IBM: delta tiles are a little hard to read due to limited palette. Shapeshifter can take permanent damage (it's supposed to fully heal after every fight).
NES: adds crappy music during strategy and fights. Graphics are a little hard to read for the fights for the sake of hitboxes.
Amiga: mechanically almost identical to C64 version, but with prettier graphics, sampled sounds, and somewhat hard-to-read hitboxes.
Also, I'm utterly astonished the author managed to manipulate the AI to cast Imprison on their DRAGON; by all rights, the AI should have either cast Imprison on the Shapeshifter or else summoned an elemental on it.
I'm quite the fan of Archon, and it makes perfect sense for the enemy being aggressive if our side goes first, as the delta tiles start changing color in favor of the side that goes second, essentially forcing the side that goes first into a defensive strategy. The advantage of fighting on a tile of your own color is an energy barrier on top of your icon's health meter that must be destroyed before your own icon takes any damage. There are six colors that delta tiles may be, and depending on their color, your icon may have an energy barrier equal to 0, 1, 3, 4, 6, or 7. To put all of this in perspective, icons with the lowest health (Knights, Goblins, and Archers) have 5 health, while the icon with the highest health (the Dragon) has 17 health. So while that barrier can mean an icon can soak one more hit, that advantage doesn't necessarily mean victory.
Fun trivia: while unlikely, a game of Archon can result in a draw if both the dark's and light's final icon kill each other at the same time, or a stalemate if enough turns pass without any combat or any spells cast.
Also, I'm utterly astonished the author managed to manipulate the AI to cast Imprison on their DRAGON; by all rights, the AI should have either cast Imprison on the Shapeshifter or else summoned an elemental on it.
Thanks! I've only ever played the NES version but the AI movement is seeded based on certain frames hit when starting a new game. Summon elemental and teleport are also able to be cast entering from different frames but we don't want them. With Imprison, it appeared to always target a unit within the wizard's own attack range and if no such unit exists it goes for the dragon which I can only assume is because the shapeshifter itself has lower "stats" pre-combat or something.
Also, I'm utterly astonished the author managed to manipulate the AI to cast Imprison on their DRAGON; by all rights, the AI should have either cast Imprison on the Shapeshifter or else summoned an elemental on it.
Thanks! I've only ever played the NES version but the AI movement is seeded based on certain frames hit when starting a new game. Summon elemental and teleport are also able to be cast entering from different frames but we don't want them. With Imprison, it appeared to always target a unit within the wizard's own attack range and if no such unit exists it goes for the dragon which I can only assume is because the shapeshifter itself has lower "stats" pre-combat or something.
Interesting.
I don't know all the AI's reasoning, but from what I've seen, playing the game casually, the caster will imprison powerful units within striking range of them. I would've assumed that both the champion (Dragon and Djinni) and the hunter (Shapeshifter and Phoenix) would've been almost equally high on the list, with striking range taking the priority.
But yeah, gotta respect the frames. Imprison is very likely the SECOND fastest move the AI could've done (fastest would probably be moving a 3-tile icon, either a Knight near the Wizard, or the other Valkyrie). Time Shift would probably not be favored UNTIL an icon was imprisoned (to extend the imprisonment). Summon elemental would result in a fight. Teleport almost DEFINITELY would've resulted in a fight. Heal would not be used as there were no injured icons. Revive could not be used as there were no dead icons. And the AI NEVER uses Exchange.
Quite frankly, this shows just how dumb the AI is; a pathetic KNIGHT could've one-shotted a Shapeshifter on a peak light tile, while the Shapeshifter would've needed three hits to kill the Knight. To me, THAT was the play that floored me!
