Posts for Truncated


Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Tseralith wrote:
mostly due to the train in the fourth room which takes little more than one minute just to spawn the last part...
Oh, THOSE trains. I meant the ones that are already out when you enter the room, moving back and forth. I didn't think about how slow the other trains were, but you seem to be on top of things anyway. How is stage X coming?
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
In AGDQ 2014, when this was run, what seemed like a new trick was shown. On the weird water boss (at 29:50 in the Youtube encode), he took a hit just as the boss died, which led to his sprite becoming glitched and then falling off the bottom of the stage. This led to the boss explosion being cut short. Video link: http://www.twitch.tv/speeddemosarchivesda/b/493407760 At 04:18:30. I'm not sure if this video will stay up, watch it while you can.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Tseralith wrote:
At this stage I actually meet the most dangerous enemy so far into the run. That UFO is so nasty I have to use the Ice gun just to get rid of his projectiles, but at least I got rid of him quite quickly despite his 30 hp.
Rackarns alltså. That thing shoots way faster than the boss of this stage... I looked over the maps a little. The alternative to the UFO room has those brown bubbles which take forever to spawn. But the bottom right path looks like it has fewer enemy spawners overall. The miniboss and the trains can all be piledriven. But maybe you test all rooms before you pick your path? In any case, this is going to be a great TAS!
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Tseralith wrote:
The barrels would probably help me out a bit, though I'm not sure how much of a difference it would've made anyway considering the sheer amount of enemies I'll be facing. It could very well make no difference at all short of visual effects, but do you feel it's unprofessional of me not to make use of every little timesaver? ...or actually it could even be faster to remove them since all elves are now nicely gathering around me for me to piledrive them, clearing up bigger groups much faster. I haven't tested with keeping the barrels around, though do you think it's worth trying? I might just end up saving time from minor optimization alone anyway. Just so I'll know for sure, I've been intending to do the Secret Santa stage in order to get the best ending since I feel a difficulty like this deserves that kind of ending. So I wish to check if people would approve of this detour for the best ending category?
It's completely possible that not blowing up the barrels is slower for whatever reason. Enemies don't clump up as well, the barrels kill too few to make a difference since so many enemies die at once with the piledriver anyway, etc. I'm just throwing ideas out there. I won't think less of you if you decide they are not worth testing. :) ReallyJoel's Mom wouldn't be pro if she didn't blow up the galaxy at the end! I complete agree with your decision to play the extra stage to get the best ending.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
I couldn't find any maps online, so I went ahead and made some myself. As you can see they are extremely professional and well-made, and not at all thrown together in Paint with copy paste. Anyway, as you probably already know, all paths are equidistant (same number of rooms), so it's all about choosing the fastest rooms. Also the maps are pretty trivial, except for Stage 4 where the map wraps. Stage 1: http://postimg.org/image/48amtfmyr/ Stage 2: http://postimg.org/image/x8cwk4kkh/ Stage 3: http://postimg.org/image/yoof29nhd/ Stage X: http://postimg.org/image/xyt9ointh/ Stage 4: http://postimg.org/image/xno6d56ht/ EDIT: Updated Stage 1.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Tseralith wrote:
This should save time since it'll take about 64 frames from grabbing to throwing and landing again. That's a total of 20 damage (12 for getting rid of the turret and 8 on the boss) which would've taken ~80 frames from just shooting them normally with Brick... not too much time saved when you consider it, but it's still cool! Stage 2 is finished now as well actually, with loads of piledrives when the situation starts to look grim and some fancy shooting in between. With this stage I also discovered there's actually a cap on how many enemies may be on screen at once. When the cap is reached, more enemies just won't spawn until I kill off some of them. You won't be able to see it here yet as this was something I just discovered when I was messing around. Here's the link to the latest .WTF file for anyone who wish to watch it. I'll really try to see if I can get a decent enough encode for everyone else to watch sometime soon. http://tasvideos.org/userfiles/info/11761786865506540
Nice to hear that the idea on the second boss worked out! The movie looks really good. Questions: - It seems you cannot shoot the boss turrets constantly since their green shots block the bricks. Would it be worth it to use some Rainbow ammo to bounce off the walls and kill them quicker? - In the room directly before the boss, there are some explosive barrels. However, you shoot them down before there is any significant amount of enemies, so they don't do you much good. Would it save time to leave them standing, so they could take down some enemies save them for later to kill more enemies with their explosions? On another note, the game locks up in Hourglass if I try to use savestates, so I have to watch it from the beginning every time. :/
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Tseralith wrote:
It works! You just grab one of the upper turrets, fly straight down and throw it at the boss doing 8 damage to him.
Cool. I know it's possible then, I just have to practice more. :) Looking forward to the next WTF file!
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Tseralith wrote:
Aw, that's a shame you can't remove those projectiles. Oh well, I'll just have to find another way. The Ice shoots 1*0.5 the first two frames, then 2*0.5 the next three frames, not two. That should be 1.5 damage every 5 frames, which is 0.3 damage/frame. I guess I didn't explain it correctly. If I throw on the first possible frame, it'll take just as long as the piledriver would. Otherwise it'll take longer depending on when I throw them, though throwing them lets me reach enemies I wouldn't reach otherwise. Of course its preferable to piledrive them normally, though sometimes it doesn't really matter when enemies are still spawning after I've cleaned up the room. Speaking of the piledrives, I have to use it a HELLA LOT if I wish to survive at this difficulty, which makes me wonder how entertaining it would be to watch me do hundreds of piledrives over and over. Of course there'll be some variations to the best of my ability, but I believe most rooms will be watching me piledrive enemies for half the game. I've also almost cleared Stage 2. Just the last room before the boss, where I'll have to fight elves which spawns from four different exits for 25 seconds. Fun times!
Thanks for clearing up how the ice works! About the stage 2 boss: after you go one round and new turrets drop down from the sky, it's possible to piledrive the turrets as they come down. I'm not sure if that saves time in itself. But I am thinking, maybe you can throw one of them at the boss? It looks like he just might be in reach. I can't get it to work when playing normally, but perhaps it's possible with TAS precision. I know for a fact that the boss is not invulnerable when he has his spikeball shield up, only just not hittable with normal weapons. (He takes damage if you turn on cheats and use gold bricks.)
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
>I believe I should be able to use my Ice weapon to remove the projectiles instead. The green spikey projectiles in the picture cannot be removed by the Ice weapon. >A little side discovery that every second time Ice fires, it shoots two projectiles for two frames instead of only one. That should mean it actually does 0.3 damage/frame. If I understand this correctly, Ice shoots 1*0.5 dmg the first 2 frames, and then 2*0.5 dmg the next 2 frames, and then repeats. So that's 1.5 dmg every 4 frames, which is 0,375 damage/frame. Is that correct? That would make it stronger than Rainbow. EDIT: I got it running on Windows 7 even though it was recorded on Windows XP. Very nice! One question: is the piledriver the same length of time as the throw, or shorter/longer? Or does it depend on when you throw the enemy?
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
L-Spiro wrote:
Unless you know Japanese stars you probably wouldn’t know anyone I am about to list but they are pretty famous in Japan at least.
I wouldn't say I am very knowledgeable, but I only recognized Patrick Harlan and Louis Garrel. And well, Garrel is a French actor, not Japanese, so no points to me for that. Anyway, this got me thinking, actually I know someone reasonably famous in Japan as well: Camilla. (She seems to go by only her first name in Japan.) She works as a model and actor and turns up in CMs every now and then. The most well-known show she has been in was probably Nihonjin no shiranai nihongo where she played Elene. We had common friends in Sweden and used to go to the same parties. I've met her once in Japan after that. I tried getting a google image search link in here, but no luck, so google yourself for "カミラ スウェーデン モデル" if you want to see what she looks like.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
I voted yes. I really enjoyed all the damage boosts. How do you get the enemies to cooperate so well?
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
In other words: - Bricks 0.25 damage/frame - Ice 0.30 damage/frame - Rainbow 0.33 damage/frame So Rainbow is the strongest. I thought enemies you piledrived/threw into were destroyed, but I guess I just haven't hit big enough enemies. Looking forward to the rest of this! EDIT: updated Ice damage according to Tseralith's info.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Tseralith wrote:
I'm actually trying to see if I can finish this game on "Reallyjoel's Mom difficulty," TASing it on my Windows XP laptop. And I've finished stage 1 already, with no "unlockable shutters" as yet. Although I have no idea if one might appear later on. I can also mention this difficulty is close to impossible to finish outside of TAS.
I would love you if you completed a TAS of this game. :) It doesn't matter that much if it's completely optimal, given as you say the very large number of possibilities. It would still be incredibly impressive, since it's only possible to get through 1 or maybe 2 rooms when playing normally. Some things I wonder about: - Isn't it possible to skip the title screen with the escape key? I deleted my .SAV file and it seemed to be possible. - Is it faster to go into the options and turn off the intro before you start, rather than skipping through it? - I realize there are no powerups and ammo refills from the enemies. What about the presents? - How strong are the weapons compared to each other? Is there any point in using the 8 Rainbow and Ice ammo you get from the start? In the Wishlist for Windows topic, the user Billy Johnson claims to have TASed this game up to the final boss. I have tried sending a PM to get the WTF file, but no reply. Post: http://tasvideos.org/forum/viewtopic.php?p=321194#321194 He also said that it was not possible to avoid the final attacks of Mecha Santa and more extra lives are needed. See his picture. It sounds strange to me that it wouldn't be possible to block the shots with piledrivers, but maybe it's true. In a normal game I always use the Goddess Buster on Mecha Santa once he's into his red health desperation, which kills him instantly. However, this move is forbidden in ReallyJoel'sMom difficulty. Unsurprising, given that the idea seems to be to give the player the worst possible deal.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
marzojr>Yes, pressing up+down (or left+right, for that matter) changes timing slightly because the game has routines to "fix" the input that are ever so slightly slower when it has to fix it than otherwise.' So THAT is why! It was a bit of a mystery to us why up+down worked when no other button seemed to have any effect. Patashu>So this is the first time in TASing history that not having a bug in u+d / l+r input has lead to a faster TAS? ;) Heh, I guess you could say that! :) The previous inability to get the right character for the first level is really just an emulator artifact though, on a real console speedrunners just reset until they get Nightcrawler. With better emulation, say the ability to set the TV electron beam position and saving that in the movie file, this problem would be gone.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
From the very useful hitbox info in feos' script, this is my conclusion about how melee attacks work: - The game draws a headbutt hitbox in front of the character when an attack is initiated. If it overlaps an enemy, headbutt, kick and punch are possible outcomes. - The game then draws a kick hitbox in front of the character, and if it overlaps, kick and punch are possible outcomes. - The game then draws a punch hitbox, if it overlaps, punch. - If none of the hitboxes overlap an enemy, a regular projectile is used. In other words there is still a random element. It is possible to force a punch, but not a headbutt. http://tasvideos.org/userfiles/info/11501746348848149 More updates to feos' script. The biggest change is the presentation of subpixels - they are just one byte as far as I can tell, not two. Also I changed it to actual value rather than rounded, because I think information is obscured in rounded format. When seeing the actual value, it is obvious that air drag is 9 subpixels per frame, for example. I was going to change the speed to the same format, but ran into a familiar problem. Due to the way negative numbers are represented in binary, it is not trivial to output as pixels:subpixels. I have a hack/workaround, but it's not pretty, so I'm not posting it yet. I have tried to locate the charge counter for the weapons, but no success so far.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Big thanks, Marzo! So essentially, the TV is used as a source of randomness? Very interesting. I have a vague memory of this being used in some NES game, maybe Blades of Steel? In any case, I guess this means that whatever Genesis emulator core adelikat was using in his tests in Bizhawk emulates the TV electron beam to start up on a random value. The 6-button thing being slower to read, throwing off the startup somehow, is pretty much what I expected. >So until there is a setting to change the initial value of the H/V counter, the best that can be done is using Wolverine in that level. Actually, our TAS manages to manipulate the starting character to be Nightcrawler, so there is something which is not accounted for in your analysis. We do this by pressing up+down on frames 7-39. Pressing it for longer/shorter gives other characters. (Sonikkustar discovered this by accident.) I tested, and this does not affect ffc670 (FrameCount_mod4), so if your info is correct, it must affect the TV electron beam position somehow. Probably indirectly by taking up calculation cycles somehow, like having a 6 button controller? The startup does not take longer, and game action starts on the same frame (character appears at frame 174), at least.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
I though it was OK, it looks good for a SMS game. But it will probably be vault material. >I got the WR so i wondered what could be improved with it. How much faster is this compared to the WR?
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
I made some changes and additions to the LUA script, updated version here: http://tasvideos.org/userfiles/info/11429271374803498
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
10 years huh... it's crazy how long it's been. I usually get tired of a hobby well before this. Let's hope we can keep at it for another 10!
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
feos wrote:
So you still need hit type prediction, right?
It probably would not be that helpful. It's easy enough to see what attack comes out by just testing. What would be good though, is to figure out what causes different attacks to be used, so we know how to get the one we want. When making the movie I posted, for the most part I could get either punch or headbutt depending on my position. Once though, I always got kick, regardless of position, and waiting for a frame seemed to be the only way to change it. Are some frames always one attack type and other frames different attack types depending on the distance to the enemy? That seems like a very strange way to program something, and there is probably a better explanation.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
feos wrote:
Once enemy is more than 114 pixels back from the screen, he gets scrolled out (>209 for helicopters from level 2). Here's our messy script:
Big thanks for this feos! :) Looks like we are set to go then... I tested some more on how to cancel melee attacks into a good pattern, and those strange extra hits I noticed before. Here's what I found: - The 2 extra hits of 32 damage are from the jumpkick, which continues after you have landed due to a bug. - You do not need a 2 frame jump to get them, a 1 frame jump (lowest possible) is enough. - The the attack after you land from the jumpkick must be a headbutt. The other attacks do not trigger the two extra hits. - You can press towards the target some time during the jumpkick to adjust your position to get a headbutt. This usually works, but not always. The best pattern I found goes like this: - charged shot - canceled into jumpkick - headbutt - canceled into standing - punch/kick - canceled into jumpkick - headbutt - canceled into standing - punch/kick (repeat with charged shot) This cycle lasts exactly 86 frames, which is the time it takes for the red and green weapons to reload. The damage is 448+whatever the charge shot is. DPS is between 357 and 480 depending on weapon and powerup level. You can also skip the charged shot, and just go all out melee. The DPS is then 434. This means that it's better to skip the charged shot for some weapons. Here's a movie which shows how fast you can attack with this pattern: http://tasvideos.org/userfiles/info/11346644702737782
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Ok, so I found a pattern of punching and jumpkicking which increases the damage 50% compared to just crouching punches. Like the throwing animation, the punching animation can be cancelled by jumping. - Press A for 8 frames (or 11, if a headbutt comes out) - 64 dmg - Press B for 1 frame - Press nothing for 1 frame - Press vB for 9 frames - 32 dmg This does 64+32=96 damage per 19 frames (or 22 frames). The DPS is 303 (or 262). I also noticed something very strange, where hits seem to come out of nowhere... - Press A for 8 frames - 64 dgm - Press B for 2 frames (slightly higher jump) - Press nothing for 1 frame - Press vB for 1 frame - 32 dmg - Press nothing for 16 frames - Press A for 8 frames, this is actually the start of the next cycle - the jumpkick will hit a second time for another 32 dmg, this doesn't happen if you don't press A exactly here for some reason. While you are charging the punch in the next cycle, another hit of 32 dmg is added (?!) - the jumpkick seems to still be active somehow, even though you are on the ground in a different attack animation. If this is consistent, it would mean 64+32+32+32 = 160 dmg per 28 frames, or 363 DPS. Also, if the attack in next cycle is a headbutt, the next jump does not get Batman off the ground, but still cancels the attack animation (??!) - this only marginally affect the DPS. It makes it trickier to find a punch/charge shot pattern, though. Requires further investigation... It would be helpful to understand when different melee attacks are used, and if it can be affected somehow. It seems to be partly dependent on how far from the target you are, but waiting also affects it. Maybe a LUA melee attack predictor? :)
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
RT-55J wrote:
I might take another stab at this, but I would need to have some things in Real Life settle down before I could convince myself to dedicate so much time/effort to such a time-consuming and mentally-wearying hobby. Also, honestly, with how atrociously long the autoscrollers are here, I would be tempted just to use the level-skip code to go past the worst of them. One thing that would really help with the project would be a LUA hitbox display that would allow you to see offscreen enemies/projectiles. I remember that aspect was one of the harder things to optimize when I made that WIP up there (for reference, it has over twice as many rerecords as my Bio Force Ape TAS, if that means anything).
Whoa! Well, clearly I was wrong about you not coming back. Nice to have you here again. :) I had the exact same thoughts about LUA scripting as you. We need a script which tells us mainly three things about enemies: - Their position, to know when and where they spawn outside the screen. - Their HP, to know if our shots are hitting them outside the screen. - The flag (probably there is one) which decides if an enemy needs to be killed to unlock scrolling. I don't think all enemies lock the screen. Feos, I haven't figured out the object structure of this game. I can give you some boss HP adresses, they might be usable in figuring it out: http://tasvideos.org/userfiles/info/11123334719078501 RT-55J, I will PM you and see if we can set up something, when you feel you have the time later.
Editor, Experienced Forum User, Published Author, Reviewer, Experienced player (969)
Joined: 4/17/2004
Posts: 3107
Location: Sweden
"Mike Tyson: 1:58.xx" Surely the value of xx is only a RAM address away, even though it's not shown to the player?