Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
I wasn't going to make a thread of this game but this first boss has been irking me. Hoping someone has some information as to what his RNG is based on, because right now I am just taking guesses. The ways I see to change his pattern are: pausing the game, delaying entry into the stage, delaying the hits on the boss, and moving to different points around the map. This seems like plenty of options for getting the boss to behave how you want, but it is seemingly difficult to get a result where you can actually hit the boss.
It would be so much easier to have some kind of information at least showing if my inputs are getting a desired result or not. This recent optimization has made him a particular pain in the butt.
I will also throw down the most recent WIP up to the first boss, where it desyncs cause I was trying to manipulate his behavior.
User movie #638079605471419285
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
After watching this run I went back and looked at the list of runs from 2007 and there are actually quite a lot of still unobsoleted runs from that year, and even some runs from previous years. Could be a good place to look if someone likes to improve outdated runs.
Anyway, this was very enjoyable to watch, especially with the resource management. I'm sure it wasn't easy to balance that with also trying to keep an eye on that lag counter. Also I'm always in favor of zips, and it seems like you used them in as many places as possible (to the untrained eye at least).
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
Well I guess since no one has given feedback on this yet, I'll make a post!
I'm always in favor of more shmups, we dont get enough of them. I imagine mostly because it usually revolves more around playarounds than actual optimizations. That isn't to say there is no optimization though of course. Killing bosses, lag management, etc all have to be balanced along with creating superhuman playarounds. This TAS does all of those things very well, and I'm hoping this submission inspires more shmups TASes in the future (more thunder force!?).
Voted yes.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
When you asked in discord for a short game, I kind of suggested this as a MEME game, but I'm glad to see you took it seriously. Deep down I knew the game's mechanics weren't as simple as they looked because I knew jumping was faster.
I like what I saw, and since it's my suggestion I have to vote yes right? (I did)
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
Exonym wrote:
I have a very vague memory of a platformer where you choose some kind of form beforehand. I want to say it was on gameboy, in my mind I see it in black and white but I can't say for sure. I think I remember seeing a speedrun of it on speedrun.com, because I remember the runner choosing a faster form before the levels or something, but I wasn't able to find anything. I remember it looking interesting from a TAS perspective but I dont have it in ANY of my notes. One thing that doesn't help is that I remember basically nothing, it's just a hazy thing that popped back in my head yesterday and driving me nuts that I can't remember what it was.
Well I have found the game after arduous searching. Turns out it was a wonderswan game! Here is the game I was talking about:
Link to video
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
I have a very vague memory of a platformer where you choose some kind of form beforehand. I want to say it was on gameboy, in my mind I see it in black and white but I can't say for sure. I think I remember seeing a speedrun of it on speedrun.com, because I remember the runner choosing a faster form before the levels or something, but I wasn't able to find anything. I remember it looking interesting from a TAS perspective but I dont have it in ANY of my notes. One thing that doesn't help is that I remember basically nothing, it's just a hazy thing that popped back in my head yesterday and driving me nuts that I can't remember what it was.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
WIP of stage 3
Link to video
I've learned several things while working on this stage. First of all, any projectile that hits you in this stage will freeze you into place and you have to keep pressing punch or jump to break out. It takes quite a while so we want to always avoid this.
The projectiles that spawn from turrets when the music changes can be avoided by timing your grappling hook to pull you up on the same frame it would collide with you. The projectile disappears and you dont get frozen or take damage. This is better than the alternative of waiting for the projectile to move out of the way or ducking underneath them.
The enemy projectiles do not work in this same way unfortunately, so they must be ducked under, unless there's some other way of dealing with them. No batarangs to be picked up in any quick locations until near the end of the stage unfortunately.
The batarang can be used to skip a damage boost I previously had on the rocks before the boss room. This alone was not worth the detour to pick up the batarang but since we start so far from the boss, we can get a few early hits on. However, the batarang seems to deal less damage than just a normal punch so I'm still unclear on whether or not the pickup was totally justified. I'll probably do some more testing later on to determine if it is. The laser projectiles on the boss can be avoided by grappling in the same way as the orbs from earlier in the level, so there's probably something I can do further regarding that and may cut out he batarang entirely, but we'll see.
Subpixel positioning when pulling yourself up through a grappling ledge matters, having a lower subpixel position means you land back to the ground a frame earlier and can start moving faster.
Pressing the opposite direction on the same frame as jumping will turn the sprite around during the jump but not lose any speed so long as you do not press another direction. This makes for some cooler looking jump sections even though it does absolutely nothing.
Also there's some weird glitch in the second level (which I didn't post here) where if you punch while in the air just before hitting a ceiling it will glitch out and softlock the game. It's super easy to do and in fact happened to me several times while just playing the game. Not sure how this got past testing but so it goes..
That's it for now, I might not post another update until submission but I just wanted to document some of the tricks found so far.
edit: So it turns out I was right, using no batarangs is faster even though it takes much longer to get to the boss initially, simply because the regular fist does 2x the damage of the projectile attacks. Using some tricks to keep the boss in a loop ends the fight around 70 frames faster than using batarangs. It's kind of a shame since I wanted to find a way to make them useful, but I think the updated version will not collect the batarang powerup. It only saves like 15 frames on the rock section, and takes much longer than that to grab the powerup.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
That way of skipping bosses reminds me a lot of Zook Man ZX4 (GBA) during DTC6, where we had to coax enemies into a position so that we can kill them just as the boss spawns, tricking the game into thinking it was the boss that died.
I'm not sure how similar the trick actually is, but this is the first thing it reminded me of when I saw it.
It's a cool trick that is very hard to find, and I will of course vote yes.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
Well I said it was somewhat edit friendly, but after improving the first stage by 9 frames up to the boss I learned that since we have a value of speed between 0 and 1 while accelerating there are some subpixel shenanigans going on that makes certain jumps desync. I'm not really good at pinpointing subpixels on the xposition via RAM, so I just had to manually edit the stage at these points.
Also the boss fight desyncs and based on his first attack, it seems random when he will appear after the first hit. My 9 frames saved in stage 1 may be all negated if I can't figure out a way to manipulate this fight to a better pattern.
Here's the 9 frame improvement file anyway, I killed/hit some extra enemies for entertainment: http://tasvideos.org/userfiles/info/72954535641103783
I'm not really good with figuring out RNG via RAM watch tbh, so I'm not sure about it other than trial and error.
Edit: Well I inserted a frame in order to sync the movement back up and somehow it resynced the boss fight, at the cost of 1 frame. I also killed another enemy in the level just because I could, and didnt result in any desyncs. Here's the new movie of 8f improvement to stage 1:
Link to videoEdit 2: So the subpixel position of X and Y were right next to the values of the x and y position, how about that.
Xsubpos: 0010
Ysubpos: 0018
Subpixel count increments by 32 every time you accelerate or decelerate, to a max of 256. If it reaches 256 you gain a pixel, if it reaches lower than 0 you lose a pixel.
Subpixel incremental values: 1.0 -> 1.125 -> 1.250 -> 1.375 -> 1.50 -> 1.625 -> 1.750 -> 1.875 -> 2.0
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
This game seemed pretty cool for a game boy game, although a bit slow paced but very simple to work on and somewhat edit friendly.
WIP of the first stage:
Link to video
Some things to note about the game that I've found so far:
- Xspeed: 0014
- Xsub: 0013
- Xpos: 0011
- Yspeed: 001C
- Ysub: 001B
- Ypos: 0019
- Taking damage gives you a nice boost of about 3 xspeed for 5 frames. Managing these will be one of the key elements to this TAS. I tried pretty hard to find some way to preserve this speed after the damage ends, but to no avail.
- Acceleration on the ground is instantly at 1 xspeed, whereas in the air it has to build up to get to 1. Therefore damage boosts along the ground are faster than damage boosts in the air (from enemies).
- walljumps are necessary but should probably be minimized considering how long they stop your movement speed on the wall. In general whenever you can grapple to the ceiling it's your best bet for scaling vertical sections since you can pull yourself up rather fast
- Every time you land on the ground your momentum is halted for 1 frame. Have not found a way to bypass this, so minimizing your number of jumps is also something to consider.
- Somehow by doing a max jump and just barely landing on a ledge I was able to go behind the background at a certain point in the level. I'm not sure what causes this really but I think it looks cool.
- There's a relatively quick batarang pickup near the middle of this first level. Every one you collect gives you 5 projectile attacks that replace your fist when toggled on. This could possibly save time on the first hit on the first boss, since you're so far away from him. I haven't tested it, but it would have to save more time than picking up a batarang powerup. It seems you do not keep your projectile count to the next level though.
Edit 7/20/2021: So I just noticed you do not want to ever just walk off a ledge as that briefly sets your speed back where you have to accelerate again to reach the 1 px/f speed. Instead the better option is to jump over the ledge, thus keeping your 1px/f the entire length.
That's about it so far, nothing really major discovered yet. The game is coded pretty well so I'm not sure there will be anything major, at this point it all comes down to just optimizing the above findings.
I might continue on with this if people find it interesting, but it is a bit of a slow paced game.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
I had planned on creating the run at some point in the future, but that's just speculation. I'm not currently working on it, I mainly only TAS during fall/winter when I have time from work. So this is probably acceptable for a first publication if I had to say, especially considering I never publicly uploaded my WIPs etc.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
arkiandruski wrote:
Exonym, that slope trick can be used throughout the run, right? Do you have an estimate for how much time it might save?
This trick can be used pretty much anywhere there's a slope your speed jumps to 5 (giving you the downhill dash animation), but not every stage has them. As seen in my video you can even cancel your dash mid-slope so that you start the dash again, which refreshes your 1 frame window to jump. This was useful for positioning and getting the most out of slopes that are in awkward positions.
My particular route over this second money grab stage is around 2.3 seconds of in-game time faster than the submission. (I didn't do any frame comparisons)
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
The only difference between normal and hard is strictly boss hp. However the difference between easy and the other difficulties is less enemies. Therefore, when I was doing test runs, I chose normal because it's the lowest difficulty you can get more enemies for entertainment while also having shorter boss fights.
My concern arises from watching the second money grabbing stage: you take a very unoptimal route. I discovered a trick with slopes where if you jump on the first frame where your speed jumps to 5 you can mantain that speed in the air. You take a high route which has no slopes, and as a result arent getting a lot of movement speed through the stage.
Here's a quick video of the previously mentioned stage I uploaded but never made public:
Link to video
This also shows how you can make the game more entertaining by turning left since it doesnt slow your speed at all if done correctly.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
Okay I actually did some very rudimentary testing and it seems that the mail delivering task would take up too much time. I created a very basic route with moderate routing mistakes and it ended up being around 45s for this mission alone. For some reason in my head this should've been faster. I estimate that a purely optimized route for this would be around 40s or so. I'm pretty confident the bird rescuing mission will take quite a while since it has a cutscene after you complete it that can't be skipped afaik.
So yeah, the RTA route is probably the way to go here unfortunately (I despise the window washing task is the only reason I wanted to test this).
I will probably keep going with optimizing the mail delivery route for a bit just to see if there's not something I missed but now I don't have high hopes for the route.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
I'm aware of the monetary values for each of the missions but have you actually timed the other 2 out and compared them to the window washing mission?
I'm not saying it's a flaw in the TAS itself, that lies elsewhere. I was mostly just curious if this idea was worth looking into.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
When I saw the graphics of this game I didn't have high hopes for it honestly. However it was more entertaining than I thought since optimizing the routes doesn't seem as easy as the game looks.
It was a somewhat short game too, so all-in-all I found it pretty entertaining, especially for a game from 1984 for a system that doesn't have much going for it IMO. I voted yes.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
Wow this is a game I've considered running for a long time, it's nice to see it on the workbench.
However, I also have concerns about routing, just based on eyesight in the second money earning stage. I question why you took a very slow route into the slow buttery substance. Ending this particular stage is only slightly faster than the RTA run, and that runner made a single jumping error, which means if he didn't it likely would have been faster than this TAS. I'm just strictly going by time remaining on the in-game timer.
One theory I've always had and wondered about whether it was actually faster to do the 2 other tasks besides the window washing one, since the window washing one takes around a minute to complete alone. It's just a theory but it's something worth testing. The best bet would be to simply optimize each room, take the frame totals added together and compare them with the window washing room.
Just a hunch but I feel like this is pretty unoptimized, but that's not surprising since it's your first TAS. Also I'm going purely on conjecture, I haven't actually delved into TASing of the game. The lack of game submission text explaining your choices doesn't help.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
Probably the most simplistic platformer that exists, but that doesn't make it bad. Pleasant graphics and listenable music along with its relatively short length make a decent enough movie that I had no trouble watching all the way through.
Voted yes of course
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
While it's obviously not the best music I've ever heard, I have definitely heard worse. I don't think it's as bad as people said.
Anyway, the game is simple but looks like it has some weird momentum which allows me to enjoy the level of play even more. Plus it's really short, I had no problem watching the entire thing even if it is slightly repetitive.
I voted yes.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
I don't see any problem so long as the TAS is faster than the RTA (taking into account timing differences of course).
As for the vault rules, I don't necessarily agree that it should be completely distinguishable from RTA for the category. In my mind I've always viewed vault as just a record keeping of games from the TAS perspective. Moon/Star tiers obviously are the exception and I think the rule should only apply to them. Obviously I didn't make the vault rules, but that's just my take on that particular rule.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
A great example how varying up your playarounds during autscrollers (especially really long ones like this game) can increase the entertainment value of the run significantly.
Great improvements to an apparently overdue game! Yes.
Editor, Experienced Forum User, Published Author, Experienced player
(883)
Joined: 7/20/2011
Posts: 345
FatRatKnight wrote:
The easy thing would simply be to graft Team 7's castle onto Team 4's run, and call it mostly optimal.
At this point this seems like the best option, since it seems no one is willing to actively improve this (myself included). It's still a pretty entertaining run and I think it should be published and people can improve it later.
