Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Very nice, great to see an improvement that uses the latest discoveries in this game.
Riyan wrote:
Thing is you only get three lives, so even if it was possible you wouldn't save as much time compared to the next two stages.
I assume there is a reason the RTA-runs only death warp in stages 3 and 4. However, since there are continues in this game, it's not obvious how the number of lives would matter for this decision. It seems like the time it would take e.g. to fall into a pit at the beginning of stage 3 and force a game over could be weighed up by additional skips. Are there no suitable places for death warps in stages 1 and 2?
Finally, I'd like to add a few comments about this trick. I think it's correct that Chambers was the person who figured out the game mechanics behind the skips and worked on finding setups for doing them in RTA-runs. However, I believe 'badbrakes' was the first* person to report getting a death warp (https://forum.speeddemosarchive.com/post/journey_to_silius_21.html# ). I'm not familiar with the history between badbrakes' initial discovery and Chambers finding the explanation and setups, but I'm pretty sure it was a "known unknown" among speedrunners of the game, so Chambers knew what he was looking for. Not that this diminishes Chambers' work in any way, but thought it was still worth mentioning.
* To my understanding, the requirements for this trick are quite specific, but it's still a trick you can stumble upon by accident when playing normally (I think this is what badbrakes did). So it's not too unlikely to imagine that others have done it in the past too, just never reporting it publically.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I hope there is still time for replacing the movie file. I watched the TAS again and noticed there was an unexplained change in how the two secret rooms were solved (one at the end of the pond, the other in the right Landing). It turns out the route at the end of the pond was 4 frames faster. Implementing that route in the secret room in the right Landing worked for the 4 frames saved, but they were then immediately lost due to stair mechanics in the hallway (it seems like you can only move on stairs on specific frames, determined by some global counter, which did not pass a cycle from these 4 frames). Even though there was no overall time saved, I still think it's better to include the faster route in the secret room in the final movie:
https://tasvideos.org/UserFiles/Info/637927384474000109
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I've found a minor route optimization that can be used at the start of the last two stages. I've updated the TAS for a 42 frame improvement. Should I make a new submission or is it possible to make a minor update to the existing submission, even if it's after publication?
The improvement was found during console attempts, which I'm now done with. I'm no longer playing the game and have no intention getting back to it any time soon, so I expect this to be one-time thing.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
deuxhero, you're of course right. It saw western releases on many different platforms. Reading it again, I can see that it wasn't clear that I had "Nintendo systems" in mind when writing it. I have therefore updated the comments.
mklip, I completely agree with you. The entertainment value of this TAS is most likely rock-bottom for anyone not already interested in the game. When I pick a game to TAS, I couldn't care less about how entertaining the end result is though (but I still try to play around whenever the game allows me to do so). In this case, I recently played through the game casually, then started to ask myself a few questions about the inner workings of it and eventually got interested in how fast the game could be beaten. Entertainment was never a factor in my decision.
For those of us that don't care much about the entertainment aspect of TASing, I guess it's unfortunately not possible to submit directly for the Vault though. So there will always be this mismatch in expectations when verifiers look for entertainment in movies that weren't submitted for consideration of being "objectively" entertaining in the first place.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
No reason really. It's just the latest version I had on my computer. But I'll change to 2.4.2 and I have already checked that the movie syncs. 2.4.2 still doesn't show the lag frames though, so I assume it's the game that deals with lag different from many other games so Bizhawk can't recognize them.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
This is a game I have done a console speedrun of and it should be a short and straight-forward TAS. While I know it's going to be pretty unspectacular and very close to the real-time speedrun., I'd still like there to be a TAS of it, so I plan on making one. Current knowledge about the routes and game mechanics that I will base the TAS on are found here:
https://kb.speeddemosarchive.com/Ikari_warriors_2
If anyone should be sitting on any useful information not mentioned in the guide above, please share. (The guide doesn't yet explain the game mechanics behind the "Galango tunnels" at the end of each level, but I have figured that out and will update the guide)
For now, I have completed the first level:
http://tasvideos.org/userfiles/info/64873528446356604
Finally, probably not the right place to ask, but if anyone knows how to make TAStudio display lag frames in this game, that would be very helpful. I'm using Bizhawk 2.3.1 and all frames are shown as valid, despite the game running at 20 fps and lag being a thing in this game. (and no, I don't have "hide lag frames" selected in the menu)
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
With the RTA timing method (based on SDA's "character control" timing defintion), the TAS time is more like 4:35. Before the "jump to door" trick was found, I got the time down to 5:02, which was ~10s slower than your previous TAS. So the new run is roughly equivalent, just with the jump trick added.
Before closing out, I'd like to write down a few memory addresses. Not that they're particularly difficult to find, but I don't think most of them have been mentioned before and might therefore save someone a few minutes of searching.
$59 - Character's x-position on the screen
$6E - The screen's x-position
$BC - Goldmoon's HP (I think you've already mentioned this in one of your previous submission comments)
$35B - x-position of Goldmoon's hitbox
$380 - Protection from evil timer
$383 - Global timer that's reset at the start of every attempt. I believe this provides entropy to the RNG. It also seems to directly control the upwinds in 2-2. As far as I can tell, the upwinds are therefore not possible to manipulate and the stop you had to do was just a consequence of the time it took to arrive there.
$3A9 - Temple timer (room 4-1). Not directly relevant. However, it's easy to confirm that it's not possible to speed up the ending sequence (at least not by any ordinary means).
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Like in my reply for the previous submission, I need to start by saying that my contribution has been to provide ideas here in the forum. I haven't actively worked on this TAS.
Very nice job. Awesome that you managed again to get a clean fire gauntlet and the room after as well!
Everything in this run makes sense in my eyes. I did quick spot checks in a few rooms that raised question marks after watching. However, I couldn't find any improvements that didn't impact e.g. the health. As far as I can see, this run includes all known time savers and appears globally optimized, based on today's knowledge. But again, a thorough understanding of the hit mechanics would probably open the door to a few more frames to be saved.
Arc, I finally got the RTA run I was looking for yesterday (4:45). That means I will stop playing this game. The chances of me finding anything new from now on are therefore pretty low. Since I'm not aware of anyone else actively playing this game, you can probably catch your breath from this game for a while now. :-)
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I've found a new time saver. By jumping to a door and holding up or down (whatever is applicable for that particular door), the door transition will be sped up. It's like the game starts registering the input for initiating the door transition during the jump even before the character is within the door's hitbox (just an unconfirmed guess from my side though). The earlier you jump, the better. So you should try to land barely inside the door's hitbox.
I've only done some spot checks, but for the doors I've looked at, the time save was always 35 frames if done optimally. There are a lot doors in this game... The time could therefore drop up to somewhere in the range of 1000 frames (depending on what RNG the new run encounters and assuming this works on all doors).
I'm focusing on console speedruns for this game, so no plans from my side to improve the TAS.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Just posting this for documentation:
https://www.twitch.tv/ktwo123/clip/AbstruseSilkyPlumageSMOrc?filter=clips&range=7d&sort=time
It's the first time I've had no arrows in 2-3, including save state practice. I don't see this changing anything for the current TAS, but the knowledge that it's possible to get no arrows in this room could be useful if someone looks at an improvement in the future.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I'm going to start by saying that I haven't actively worked on this improvement. So I hope it doesn't look like I'm replying to my own submission here. I appreciate being added as co-author, even though my contributions imo were quite humble.
The bulk of the run is the same as the previous one. The overall route and strats are still the fastest with today's knowledge. And with the inclusion of the improvements that were discussed in the wake of the previous run, it now also contains all known time-savers with today's knowledge. It's nice to be able to say that, so well done for following through with the improvement.
Hit detection and RNG in this game aren't fully understood and judging by the run comments, it doesn't sound like that has changed with the new submission. Many of the rooms are clearly individually improveable, but were done unoptimally because of previous rooms or to set up the right conditions in following rooms. A better understanding of especially how the hit detection works would improve the possibilities of linking together more rooms (or even the whole game) in the optimization. If there are any frames left to be unlocked, I think that would be the key to open the door to them.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I've also been playing (RTA) this game fairly actively the last couple of days. I've had to modify the strats a bit in order to avoid relying too heaivly on RNG. With the modified strats, sub-5 (SDA-timing) isn't going to be feasible, but it has still allowed the time to drop by several seconds compared to the old strats.
A few observations that I've made during my attempts and some light looking at the code:
* It's possible to jump through the enemy in 1-2 without getting touched (I assume that's the same improvement you have found)
* The enemy in 1-5 will sometimes not shoot and allow you to run straight through it (maybe the same that you already found?). This happens much less frequently than getting through 1-2 without getting hit though.
* It's possible to run straight through the troll in 2-1. I think you already knew this, but the RNG just wasn't favorable in this TAS. The probability of this working appears to be roughly the same as what I described above for 1-2.
* The enemy in 2-6 will sometimes not shoot and allow you to run straight through it. The probability of this working appears to be roughly the same as what I described above for 1-5.
* I've now noticed that many more of the door locations change than I had previously observed. I haven't been able to find the door locations in the RAM, but it seems to have something to do with the screen position ($6E) and the character's position on screen ($59). Once a door has been loaded, it's locked to the screen position. Change the screen position's value and the door will move along with it. My best guess is that the different door locations occur when the character gets hit and bounces around on the screen.
* As far as I can tell, the governing RNG-value for enemy fights is whether the global timer in $383 is even or uneven. I'm sure there is a lot more to say on this topic, but at least I haven't seen any indication of the character stats playing a role in getting hit or not.
* Depending on the enemy and whether you get "body" or "weapon" damage, the game attributes a pre-defined number of hits ($64) and looks up the damage of each hit in a table (either 1 or 2). Body damage generally only incur 1 attack per frame, meaning a damage of 1 or 2. Getting hit by a warrior sword incurs 2 attacks (total damage of 2 or 4) etc. The fires in the gauntlet are the hardest hitters and can deal up to the maximum 5 attacks. So there is a potential loss of 10 health within a single frame.
* I haven't tried to go into detail of when the game decides there is a body hit or not. However, I've noticed that both the warrior in 1-2 and the troll in 2-1 are passed through when your speed is doubled due to both the screen scrolling and the character position on screen moving. Since the timer in $383 isn't rolling every movement frame, I guess it's possible to skip over the hitboxes if everything is aligned and you're moving at double speed. In the TAS, you also manage to pass through other enemies without getting hit, but always by introducing pauses or quickly turning around first. Not sure how that fits in. It's not something that makes it more reliable in RTA though, so I'm not planning to look at it further. But I can say that the other warriors and trolls have never allowed me to pass through them without getting hit by just running or jumping into them.
* Maybe it's just my imagination, but I have the impression that jumping through enemies makes it slightly easier to get through. Jumping loses a movement frame when landing, so should normally be avoided. But since it feels more reliable than just running into an enemy, it's still how I deal with most enemies. I've only been able to get through the warrior in 1-2 without getting hit by jumping. Never running. So there must be something in the hit detection that changes between running and jumping.
I wish I could have been more conclusive in some of my points above, but at least you know what I know now.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I haven't spent a lot of time speedrunning this game, but I remember noticing very early on that the position of the exit in 2-9 varied from time to time. That was the only room I noticed this in. Now, when I've picked up the game again, I quickly noticed that the same thing happens in 1-3. My feeling is therefore that this is not something that goes on in every room. But if you decide to look at it more closely and find out more about it, I'd be interested to hear.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Yeah, forget my comment about the last select input. I tried it out properly now and it's clearly not possible to enter it one frame earlier.
True, attacking is pretty slow, so it might be difficult to end up with an overall improvement by doing the menuing in 2-7 instead of in 2-8. I haven't thought more about it, because I tried something else and managed to improve 2-8 and 2-9 by 27 frames (http://tasvideos.org/userfiles/info/62315251918382424). With that improvement, I don't really see how entering the menu in 2-7 could save time.
Whenever you get out of the menu, the window position ($6E) is readjusted on a per modulo 16 basis. The character position on the screen ($59) is fixed here, which means you can teleport a few pixels by entering the menu on the right frame. The rest was just getting through as quickly as possible without losing any health. I got this result with barely any testing, but since it felt pretty good, I didn't pursue. There are quite a few possibilities left to explore based on the "menu teleporation" trick, so it's possible a few more frames could be saved.
Nice that the attack turned out to be slightly faster in 3-11!
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Apologies for the untimely reply. I just started looking at some of the details today.
Very nice job coming up with the new strats! I've done RTA speedruns of this game and I could have sworn I tested going left in 1-2. However, since it's clearly faster, I guess it's just something I kept telling myself and never bothered double-checking. The "hold person" strat is a very good find. I remember trying out all the spells to see what could be useful, but didn't think of combining them in this way. I estimate this could save around 10 seconds. My pb is 5:11 (SDA-timing), so I'm pretty sure sub-5 is possible (although I'm guessing not by much).
With the current public knowledge about this game, it's difficult to evaluate how optimized most parts of the run are. As far as I know, the outcome of most enemy encounters are random (whether to get hit or not, HP loss, knockback and which different action(s) that are best suited in each situation to get past an enemy). The room-by-room comments make it sound like each room (or couple of rooms) was optimized in an ad hoc approach. Arc, if you have additional insight in how to manipulate the enemies other than through "running up to an enemy and just try everything", then at least I would be most interested to learn about it.
I still have a few comments about the run itself:
1. Your final menuing is unoptimized. You:
17972 R
17973 [nothing]
17974 D
17975 [nothing]
17976 D
17977 A
this can be done 2 frames faster:
17972 D
17973 R
17974 D
17975 A
In the same final menu, you can input the final select one frame earlier. The earlier select doesn't trigger the ending earlier, but my understanding of tasvideos timing is until last input that triggers the ending. That means the movie file should end after the last select has been input.
By the way, you might want to consider editing your comment about Randil's final menuing being sloppy. ;-)
2. You pressed down at 8949. You could have pressed down 4 frames earlier (it would only result in 3 frames faster menuing though). Was that to manipulate something or just a mistake?
3. This is an open point. I don't have a definitive improvement to suggest, jut a question about the optimization. Going into the menu can make it easier to get through enemies. In 2-7, you could have gone into the menu right before the last enemy instead of doing the menu trip in 2-8. This would save you quite a bit of time in 2-7 (at least 16 frames). I understand that the RNG in 2-8 would be different and you wouldn't have "Protection against Evil" in 2-9, but there are lots of possibilities so maybe one of them would still end up faster than what you did.
4. Just a minor correction. The spider in 3-11 is just a normal enemy, so it can be attacked (obviously by a low attack). It's equally fast to using the "hold person" spell, so it doesn't change anything for the run. It's just in case you want to edit the comments for this room.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Thanks for all the information posted so far. While I'm very interested to learn more about what's going on, I don't have the motivation right now to contribute much (or even follow in detail what you have found). It's temporary though and I'll take a look at it when the time comes.
However, I think it's unlikely I'll work on an improvement for implementing the warp to 8 (and let's not forget despawning one of the eyes of the final boss!). Don't let that stop you from publishing what you have found in this game though (scripts etc). Personally I was hoping you would feel the urge to make the improvement yourself (along with finishing the R.C. Pro-Am II TAS!), but that's clearly something that should come from own motiviation and not because someone else wanting it done...
From RTA-testing, I've gotten the second warp about 4 times now, I think. A rough estimate is that I've done a few hundred attempts of "blind" testing to get this result. By "blind" I mean that I've just mashed like I normally would to shoot the warp, but without any specific mashing pattern or otherwise looking for specific enemy patterns etc to be present. When posting about this finding, other players have reported also getting the second warp, but it had only been shared in various chats. Anyways, this was just to say that it doesn't seem like the conditions need to be "needle-in-a-haystack" specific for this to occur.
I would be surprised if it turns out to be a difficult decision on what to buy. My impression of anti-gravity is that it's pretty slow without boosters. I would expect that buying double-strength boosters after 8 and 13 is the fastest. But it's not something I've tested, so it's just a gut feeling.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I've managed to record an emulator movie showing the warp from 1 to 8:
http://tasvideos.org/userfiles/info/58524046396026308
The second warp object value is stored in $325. Some of the other properties can easily be found by applying an offset to the addresses in FatRatKnight's post higher up. I'm primarily trying to trace back in the code why it spawns (frame 2609). I'm struggling a bit though, so any help analyzing the mechanics behind would be much appreciated. Firing at a steady pace to ensure the object adresses are kept full seems to be a key. The resulting bullet pattern also appears to matter. I haven't been able to identify what determines if it's a correct bullet pattern or not though.
I would also be interested in why it gets the value corresponding to the warp (#$0A). I've traced it back a few steps, but I eventually get stuck on RAM-addresses whose purpose I have trouble finding. As far as I can tell, if something spawns, it's always going to be the second warp though. So it's more out of curiosity than an actual need for manipulating something.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I've managed to warp from planet 1 to planet 8! Link to the video and the powerpak save state is found here: https://forum.speeddemosarchive.com/post/solar_jetman_nes_7.html
If this can be repeated, it would save several minutes off the current TAS. I have poked around a bit in the code, but I'm not sure what to look for next. Unfortunately, I haven't been able to reproduce this, so there is no emulator movie to dissect. My best guess is that a random enemy somehow got attributed the warp value. Maybe lag or some kind of overflow are involved, similar to the eye despawn described above?
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Nice to see a TAS of this category. Good find of the new clip jumping!
While watching the movie, a few improvement ideas came to mind. I decided to play around with them a bit to see if any time could be saved. Most of my ideas didn't produce faster solutions. This was only a spot check in a few places, but I wasn't able to shave frames off the solutions that were used in this run either. I still have a few comments/questions though:
* At frame 17911 (room 22), you should have jumped over the blue flame without quenching it first. That would have saved 20 frames.
* In room 29, it's possible to get the gargoyle to start moving to the right (iso left in this run). I think that could speed up the jumps to the exit. Unfortunately, I don't know the game mechanics behind this. It's only based on (a lot of) RTA experience.
* Not a time save at this point, but I'm still mentioning it. At the end of room 31, it would be faster to use a fire ball against the last demon head. However, all fireballs along the route would cost more to collect than the time saved in room 31. At least from what I can see.
* How much slower is the current RTA-route in room 48? I assume you have already tested it, so it's just out of curiosity on my part.
* Finally a very, very minor comment. There are lots of "garbage" inputs (inputs that don't do anything) in the movie file. Unless these serve some purpose not explained in the comments, it's good practice to keep the input file as clean as possible.
From an optimization point of view, I think the time loss in room 22 should be corrected. It would also be nice to get clarity of the possibility to save time in room 29. Other than that, this appears to be optimized in my eyes, both in terms of solutions and execution.
If you're up for it, I hope you consider looking into the best ending run as well. There is a little time to save in that run.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
I'm not providing a temp encode for this one. The video is next to unwatchable in 30 fps and I haven't looked into how to get youtube to accept this type of footage at full frame rate. If someone else knows the trick, feel free to post an encode.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
In level 5, couldn't you have started the jump over the last skull (at 1:20) from more to the left? I haven't tried it out myself though, because I haven't set up bizhawk for c64.
Experienced Forum User, Published Author, Player
(243)
Joined: 8/10/2008
Posts: 113
Thanks for the feedback, Jigwally.
Unless I'm missing something obvious, there is nothing in that algorithm that favors one parity of $2A over the other.
Going into the loop:
$23 even -> $2A will change parity in every iteration
$23 odd -> $2A will stay with the same parity indefinitely
The loop is visited twice every frame, with the only difference being that $23 has been incremented in-between. (apologies if you already wrote this with different words in your reply)
I don't know if there is a good way to count the number of times a loop like this is run. But even if you were able to somehow control the number of iterations, the starting values still determine what you'll end up with and I don't see how you would be able to control that. To me, it just goes back to that the pass locations should be evenly distributed.
You say that you observed more odd than even values of $2A. I've checked and written down the value of $2A frame-by-frame at various points, with and without user input. In the end, I always end up with a more or less even distribution. If you do other checks and find a different result, I'd be interested to have a look as well.
Since this is not a popular game to speedrun, it's of course possible that the number of console attempts are too few to conclude on a distribution other than 50/50 and that it's currently just a streak of (particularly bad) luck that has resulted in one location being more common. I guess time will tell...
I have found one interesting thing though. The transition time between level 2 and 3 is impacted by player input. You should keep holding up, left or right (or a combination thereof) when you get near the end to finish level 2 asap. Obviously trivial on console and an easy time save (up to 1.5 secs compared to just standing still), if you're aware of it. The current TAS by zakem66 could also save 7 frames that way. I haven't looked at any other level transition. The ending of level 2 is slightly different from the others though, so not sure if something similar can be applied elsewhere.