1 2
6 7 8
20 21
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
I decided to revert back to the previous number of souls to bypass the fire. I decided to take advantage of cheat roads to get to the end of the Fillmore sim much sooner, because Fillmore will still grow slowly while I'm away. WIP
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
The choice of skipping all magic related items in Filmoa is an interesting one. So how many scrolls are you gonna end up with in the end? I'm also still curious as to why you build your road at the beginning up, down, then back up again.
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Sir VG wrote:
The choice of skipping all magic related items in Filmoa is an interesting one. So how many scrolls are you gonna end up with in the end?
Look towards the end of page 7 of this thread. I have a big post there explaining my tentative plan for which scrolls to get and why.
Sir VG wrote:
I'm also still curious as to why you build your road at the beginning up, down, then back up again.
From earlier in this thread:
Sir VG wrote:
One thing about the up/down construction at the beginning is seems you have to go up first, then down, then back up. Going down then up seems to cause me to get that stupid house in the fire zone.
I tested this myself when I was making the 1.43 version, and got the same results as you did, so I stuck with going up then down then up this time.
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
Haha...my brain is fizzling lately. Gomen. ^^;; Oh a side note, in comparing my latest run attempt on a console to the TAS, I've now officially beaten it by 9 seconds. If there wasn't any motivation to improve it before...now there is.
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Still TASing at a snails pace. Fillmore Act 2 is now finished. I ended up having to avoid one of the spikes in the last part of Fillmore Act 2 in order to have enough life for the boss. See any possible improvements? WIP
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
Overall, looks good. I wonder if there's a way that the apple at the end there could be incorporated into things that would allow you to burn through all the spikes, plus if you have more HP at the end it'll take less time to fill. If there's a way that wouldn't cost you much time, that would be my only suggestion.
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
I experimented a bit with getting the apple. I had to slow down my movement a bit, but that was balanced out by being able to get hit on the last spike. And since I ended the boss battle with more life, I ended up saving about 10 frames. Thanks for the suggestion; this kind of strategy may be usable in some other levels too. WIP
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
The strategy on the apple is good. One new thing is about the stairs just before the boss. Is it faster to jump up them or to walk (like you do in the WIP)?
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
It's faster to walk. Your horizontal walking speed does not change when walking up stairs. The horizontal speed of a jump is the same as the horizontal speed of walking, with the exception of the "running start" you get depending on when you jump. So walking will be faster, because when you land from a jump, there is a delay before you can start moving again, and there's no increase in horizontal speed over walking to make up for that.
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
Well, if (and a big if at that) I ever go back and do another attempt at ActRaiser on the console, I'll remember that. Jumping seemed faster to me on stairs (I also do it on Casandra going up sandy slopes).
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
I haven't done any tests on sloped areas other than the stairs in Fillmore, so it's still possible that jumping would be faster in Kasandora. EDIT: I see that your latest ActRaiser attempt is about 3 minutes faster than the one you have posted on SDA! Have you been making recordings of any of these newer attempts? If you have, I'd be really interested in taking a look at your latest recorded attempt.
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
I have my latest attempt ripped from DVD, but haven't sent it to SDA at this point since I haven't gotten my notes typed up yet. And yes, it's about 3 minutes faster, and about 39 seconds faster than OgreSlayeR's TAS. A lot of improvement came about because of your work, so thanks for that. :) I'll try to encode something later, if I have time.
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
OK, I've completed Bloodpool Act 1. Most of this attempt is exactly the same as what I did in the 1.43 v9+ WIP. The boss strategy changed, since I don't have a fireball to use this time, but I managed to only lose 22 frames during the boss fight compared to the previous TAS. WIP
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
No complaints here, your boss fight was excellent. I especially liked that jump early on from the ledge just above the first green lizardman unto the floating board. That must have taken some work.
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Skilled player (1640)
Joined: 11/15/2004
Posts: 2200
Location: Killjoy
Arcade TASer of 2010PCE TASer of 2010Arcade TAS of 2010Lucky TAS of 2009Lucky TAS of 2007
What happened with this run? I was really amazed at how optimized the first act was.
Sage advice from a friend of Jim: So put your tinfoil hat back in the closet, open your eyes to the truth, and realize that the government is in fact causing austismal cancer with it's 9/11 fluoride vaccinations of your water supply.
Experienced Forum User
Joined: 7/12/2004
Posts: 524
Location: USA
2 Month bump. I guess this is abandoned?
Working on: Command and Conquer PSX Nod Campaign
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Hi everyone, I'm back from the dead! :) Sorry, but I haven't posted to announce that I've completed the game, sims included, in 45 minutes. Rather, I've finally started to TAS ActRaiser again. So... Is anybody else secretly TASing this game? I'd be willing to work with you. In any case, I'm back up to the beginning of Bloodpool again. I managed to shave frames off of Fillmore Act 1 and Act 2 again. In Act 1, I now take a route along the tree branches, which saves about 15 frames, probably because there's a part where I speed up while falling from the trees. In the Sim itself, I now build another road while heading down to the third lair. I built a table with population growth for Ogreslayer's TAS, Sir VG's SDA run, and my previous WIPs. Looking at it, it looks like Fillmore will probably grow enough after I've left that that square will be used (remember that I'm using "cheat roads" to leave Fillmore sooner, so I'll be waiting more in later towns). I lose about 300 frames building that extra road, but there about 720 frames between construction phases, so it should be worth it. In Act 2, I let myself get hit by 2 of the enemies since it saves time due to going through the water faster. Overall, I saved 20 frames in Act 2. I'm now making it my official policy to go through every level at least twice, because I always manage to find another way to save time. Competing against myself seems to work well. I also did some investigation of the game mechanics. It turns out that there are two memory locations (7E08C4 and 7E08BC) that when used together take the guesswork out of finding out how much of a forward boost you will get when jumping, and whether or not you'll get a height boost by using the sword at a certain point in the air. I can't exaggerate how much easier that's made things for me. I also found miscellaneous other memory locations, like the X and Y positions of the angel's arrow, but I'm not sure how useful they'll be yet. It's interesting that I regained interest right now, right at the time gocha is starting to add/enhance lua support in Snes9x 1.51 (I only noticed after restarting my TAS). I'm not sure whether I'll have any use for lua, but if you guys can think of anything, I'm open to suggestions. Anything to automate stuff would be awesome :). Here's the wip: http://dehacked.2y.net/microstorage.php/info/247010340/ActRaiser%20%28U%29%20%28Zidanax%20WIP%205%29.smv
Experienced Forum User, Skilled player (1640)
Joined: 11/15/2004
Posts: 2200
Location: Killjoy
Arcade TASer of 2010PCE TASer of 2010Arcade TAS of 2010Lucky TAS of 2009Lucky TAS of 2007
zidanax wrote:
Hi everyone, I'm back from the dead! :) Sorry, but I haven't posted to announce that I've completed the game, sims included, in 45 minutes. Rather, I've finally started to TAS ActRaiser again. So... Is anybody else secretly TASing this game? I'd be willing to work with you. In any case, I'm back up to the beginning of Bloodpool again. I managed to shave frames off of Fillmore Act 1 and Act 2 again. In Act 1, I now take a route along the tree branches, which saves about 15 frames, probably because there's a part where I speed up while falling from the trees. In the Sim itself, I now build another road while heading down to the third lair. I built a table with population growth for Ogreslayer's TAS, Sir VG's SDA run, and my previous WIPs. Looking at it, it looks like Fillmore will probably grow enough after I've left that that square will be used (remember that I'm using "cheat roads" to leave Fillmore sooner, so I'll be waiting more in later towns). I lose about 300 frames building that extra road, but there about 720 frames between construction phases, so it should be worth it. In Act 2, I let myself get hit by 2 of the enemies since it saves time due to going through the water faster. Overall, I saved 20 frames in Act 2. I'm now making it my official policy to go through every level at least twice, because I always manage to find another way to save time. Competing against myself seems to work well. I also did some investigation of the game mechanics. It turns out that there are two memory locations (7E08C4 and 7E08BC) that when used together take the guesswork out of finding out how much of a forward boost you will get when jumping, and whether or not you'll get a height boost by using the sword at a certain point in the air. I can't exaggerate how much easier that's made things for me. I also found miscellaneous other memory locations, like the X and Y positions of the angel's arrow, but I'm not sure how useful they'll be yet. It's interesting that I regained interest right now, right at the time gocha is starting to add/enhance lua support in Snes9x 1.51 (I only noticed after restarting my TAS). I'm not sure whether I'll have any use for lua, but if you guys can think of anything, I'm open to suggestions. Anything to automate stuff would be awesome :). Here's the wip: http://dehacked.2y.net/microstorage.php/info/247010340/ActRaiser%20%28U%29%20%28Zidanax%20WIP%205%29.smv
So wait, the test run is 45 minutes? Holy shit! As far as Lua goes, if there is anything calculations that would make understanding those 2 memory address easier, I'll help any way I can. Maybe a script that shows were sim-enemies are while off-screen?
Sage advice from a friend of Jim: So put your tinfoil hat back in the closet, open your eyes to the truth, and realize that the government is in fact causing austismal cancer with it's 9/11 fluoride vaccinations of your water supply.
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Now that I think about it, what we might want to look into (although this may be asking for too much) is whether there's a way to predict how stardust will fall . Trying to manipulate Stardust to fall for maximum damage is a pain. I suppose it might use some sort of RNG, but I don't have much experience with how RNGs work. Maybe then we could have lua somehow calculate how many projectiles would pass through the boss's hitbox. Just throwing this out, even though I doubt we could figure it out. EDIT: or maybe we wouldn't even need to understand the system. Maybe we could simply tell lua to move the hero around and try stardust at various positions and calculate how much damage it did with different attempts. Of course we would want to record how many frames lua spent moving the hero around before casting stardust. Given that, we may want to write the test results to a file. Manually parsing a file may be a bit of a pain, but not as much as trying all those different casts ourselves. I don't think I need help with understaning the two values I mentioned though: Call 7E08C4 "Boost" and 7E08BC "Crest" Boost give the number of frames of 3 positions/frame ( in contrast to the usual 2 positions/frame) movement the hero will do if you jump on that frame. The interaction between Boost and Crest while on the ground: Boost's incrementing behavior seems somewhat inconsistent unless you see how it interacts with Crest, hence this explanation. Before Hero begins moving, Crest and Boost both equal 0. When the hero begins moving, Crest increments by 1 per frame. When Crest reaches 5, Boost is set to 4. Boost then counts down by 1 per frame, while Crest stays at the same value. On the frame after Boost reaches 0, Boost resets to 4 and Crest increments by 1. Once Boost has counted all the way down while Crest=7, everything resets to 0, and the cycle starts again. The main point is that there is a part of the sequence where Boost does not go 4,3,2,1,0,4,3,2,1,0..., but instead stays at 0 for several frames. When Crest equals 4 during a jump, press the attack button. Increases max jump height by 5. This doesn't seem to work with vertical-only jumps. That may sound complicated, but I'm used to it by now, and it really helps.
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
Welcome back zidanax. I was hoping you'd come back to work on this, since it NEEDS to be done. Hopefully you can figure out how to make Stardust work at its best, since it can take a ton of damage off and is almost necessary for some bosses for major damage, though granted you can attack all of the bosses faster then I can via TASing (like 6 hits on Marahna Act II boss is a guarantee now). For a boss like the Snow Dragon, I think a combination of several strikes in the right spot plus him sweeping left to right into the magic should make for a quick kill. I've watched WIP5, and it's beautiful. It makes me wanna cry. ^^;; Keep it up, man.
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User
Joined: 7/12/2004
Posts: 524
Location: USA
Glad this is being worked on. The WIP looks good. I'll try to help anyway I can.
Working on: Command and Conquer PSX Nod Campaign
Experienced Forum User, Skilled player (1640)
Joined: 11/15/2004
Posts: 2200
Location: Killjoy
Arcade TASer of 2010PCE TASer of 2010Arcade TAS of 2010Lucky TAS of 2009Lucky TAS of 2007
zidanax wrote:
EDIT: or maybe we wouldn't even need to understand the system. Maybe we could simply tell lua to move the hero around and try stardust at various positions and calculate how much damage it did with different attempts. Of course we would want to record how many frames lua spent moving the hero around before casting stardust. Given that, we may want to write the test results to a file. Manually parsing a file may be a bit of a pain, but not as much as trying all those different casts ourselves.
This is do-able. However, it takes some computing time, and some questions answered - 1. Is there a limit to the amount of damage Startdust can do? 2. Is the RNG frame based or controller input based, or both? 3. Do you care if the player takes damage during that time? 4. Is the # of frames that Stardust occurs a finite number?
Sage advice from a friend of Jim: So put your tinfoil hat back in the closet, open your eyes to the truth, and realize that the government is in fact causing austismal cancer with it's 9/11 fluoride vaccinations of your water supply.
Sir VG
He/Him
Experienced Forum User, Player (35)
Joined: 10/9/2004
Posts: 1910
Location: Floating Tower
DarkKobold wrote:
This is do-able. However, it takes some computing time, and some questions answered - 1. Is there a limit to the amount of damage Startdust can do? 2. Is the RNG frame based or controller input based, or both? 3. Do you care if the player takes damage during that time? 4. Is the # of frames that Stardust occurs a finite number?
From what I know, here's my beliefs: 1. I'm sure there is, and it depends really on boss position. A large boss in the upper right corner (or a well positioned one) can generally take at most, 12 points damage (1/2 life). 4. Yes, the spell only goes for so long, but how long it does is what needs to be determined.
Taking over the world, one game at a time. Currently TASing: Nothing
Experienced Forum User, Player (203)
Joined: 2/12/2006
Posts: 373
Location: Oregon
I don't think it's frame based. If you just stand in one place, save state and cast it, then reload and wait a while before casting, you get the same pattern. Moving around on flat ground doesn't change the pattern, but it can change if you hit the "edge" of an area so that the screen doesn't scroll when you move. I think dying changes the pattern, but that's no help. If nothing else though, changing the hero's position will also put the enemies in different positions relative to the falling stardust, assuming that the stardust tries to fall through the same part of the screen. Also, the stardust often explodes on obstacles other than the ground, so changing the hero's position would change where the stardust explodes due to the stardust hitting obstacles in different places. That may be helpful, since an enemy does not have to have received a direct impact from a piece of stardust; it just has to be within the area of effect. I think screen transitions may change the pattern, but its difficult to tell when the stardust is always hitting obstacles. Note that I was testing this is in the Bloodpool Act 2 level. Starfall patterns seem rather less consistent when fighting the boss. I think it may depend on when you enter the room. Heck, it seems like it maybe is somewhat frame-based when fighting the boss. <-*confused* EDIT: I completed Bloodpool Act 1. I managed to save 22 frames. http://dehacked.2y.net/microstorage.php/info/1107891404/ActRaiser%20%28U%29%20%28Zidanax%20WIP%205%29.smv
Experienced Forum User
Joined: 7/12/2004
Posts: 524
Location: USA
Looks good. You might want to test if it's possible to kill the Act 2 Bloodpool boss with him teleporting once and not twice like I did in the published run. I posted that test run a while back on the top of page 6 where I came close. I was only 1 hit away then he went in his second form. I didn't try a ton, so it might be possible. Stardust seems fairly random... Maybe if I entered the room on a different frame or something I could of got a better Stardust.. Who knows..?
Working on: Command and Conquer PSX Nod Campaign
1 2
6 7 8
20 21