Posts for VanillaCoke

1 2
7 8 9 10 11 12
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Haven't really worked on this due to Kirby:RTDL coming out. The plan is: 1) Find the address(es) for the RNG and find all actions that cause it to change (if by a set value, which I assume it is, find what each action's value is). 2) Edit and freeze the values and see how it affects Captain Jelly fight. 3) Narrow this down to either A) Confirm 306 is the best result or B) Find the best result 4) Repeat 2 & 3 for the rest of the bosses. If B is true for #3: 5) Either edit the Globotron fight for the best result in the Captain Jelly fight, or find another fast means of RNG manipulation 6) Finish the Captain Jelly fight while keeping the RNG as the main focus for the next boss. Hopefully I'll get this all done by Monday. Classwork and RTDL are eating up my time xD
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Hrm, yeah I suppose you're correct. I'm used to it due to familiarity, but it would be ackward for someone seeing the game for the first time. Alright, thank you for your insight and help! I appreciate it.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Seems like a bad choice in game selection <_<.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I'll retract what I said about Meta Knight. I was experimenting with Shuffle Loop and I think that would totally be abusable with a TAS. Basically just jumping out of it (although I'm not sure if I was jumping in the middle or at the end, kind of off today) and doing another shuffle loop lets MK move up really really fast. I'm not sure what causes lives to not add back during Quick-Leave. It might be having anything but a full health bar.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Yeah, sorry about that. I can TAS for hours on end, but I've been procrastinating on purpose pretty hard when it comes to finding the Dehacked upload link xD. I'm going to upload it now (I didn't actually get to search tonight, I had a bunch of classwork to finish that's due tomorrow). http://dehacked.2y.net/microstorage.php/info/1844540252/Sonic%20Colors%205.dsm Here's the link to Nitsuja's lua script (Frames for any current level are counted in blue): http://pastebin.com/gz975ErQ I compare the differences by the blue shock-wave created by the first minion launched by the boss. It's easily recognizable.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
CoolKirby wrote:
What about playing with 4 players for the whole game? Then they could break blocks and defeat enemies that are ahead and everyone can go through the levels faster.
The thing is that there will likely be times when it would be faster to use Quick-Leave and then Quick-Join, warping to Kirby or a combination of the two, than just progress with that character normally. I know in NSMBW there were times when me and some buddies would use "The Bubble" due to the mini power-up. It was faster for them to just bubble and let one of us go through certain parts in levels. There were still times when it was good to have all four players with differing abilities, but it's just slower always having four players :/. I'd imagine that something similar would happen in a TAS of this game with Quick-Join/Leave and/or warping to Kirby. Besides using extra Kirbys, I see Dedede being extremely useful in a run. He can use the hammer throw infinitely, which is a pretty hard hitting move. Meta Knight could have some moments, he seems like he moves upward faster (not counting any Kirby abilities). Waddle Dee has an interesting flying mechanic, but it requires charging it first. I can see a lot of different ideas on how to approach a level with so many different combinations for P2-P4. Although I'm definitely seeing the game being mostly 4 Kirbys with an occasional Dedede appearance for bosses.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
First, thank you for watching! As for the version differences... well, yes and no. Sweet Mountain 2 was about halfway finished when I was still using the Euro version. Since SM 1&2 sync up effortlessly between Euro and USA, it was easy to test that out. Same results (16 frames faster), same method works to manipulate the RNG to reproduce that 16 frame improvement. I've also managed to produce the four variations on every setup (USA on 0.9.6, 0.9.7 and Euro on 0.9.6 and 0.9.7) this morning before going to class with resetting (always makes Captain Jelly attack at 320). So version differences (Both game and emulator) are ruled out. I still need to test if Globotron changes the rng. Rather than going through the hassle of editing the dsm, I'll just see what the results are by resetting before Captain Jelly, beating Globotron and differing times and record the results. I was really tired and could have given myself some results by recording Captain Jelly's attacks on each on of the real-time tests. I didn't think about it at the time until I was on my way to class. More importantly, I'll hopefully find that RNG. I know that it moves during those four instances: Exiting a level, Restarting once before exiting a level, Exiting a hub and entering a hub. And it moves by a set value for each of them (that's why doing everything after beating Globotron in real time will produce the 306 variant). Maybe I can find that first, then I can watch the run to determine what else shifts the RNG or confirm the Globotron idea.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Paused wrote:
Thanks for the detailed break down when you caught the question. Always interested in that stuff for Sonic games.
You're surely welcome. I'd have to write that information up when I submit this run at some point. Now it's fulfilled two purposes xD. I just have to add in one other part about the different directions while exploding (up, left/right and down), which makes a total of 23 different ways. 17 of those being combos of stage 3 explosions, but 9 are stage 2's that aren't used in the run. Actually, I don't think I used over 3 of the stage 3 combos with Stage 1 directions. I know I did Up - Stage 1 -> Left/Right - Stage 3 and Left/Right Stage 1 -> Left/Right Stage 3. Maybe it's only two lol. Rather than being 18 combos, it's 17 because Down - Stages 1 & 2 give the same result. I did a test run of one of the special stages today (second time doing so) and I just can't help but find the stages extremely boring. The test run isn't optimized besides trying out some different tricks and even then they wouldn't show up on Youtube due to 30 fps playback. So I can't help but think that they shouldn't be included. Nega-Mother Wisp is also pretty boring in my opinion (not to spoil the game, but it's kind of a throwback to SA2's final boss). EDIT I'm not sure why, but Captain Jelly attacks 16 frames earlier in this run. Input or restarting doesn't change it either. Delaying the stage start or beating the previous level by any amount of frames has no effect either. There is some kind of RNG because the only thing that makes a difference is resetting the game. This makes Captain Jelly behave like before, attacking 16 frames later. EDIT 2 I've managed to reproduce the 16 frame improvement, although it's quite time consuming. The process is restarting once on another stage, exiting, then exiting sweet mountain, then reentering it and entering Captain Jelly. It does require each step in there and in that specific order (two restarts ruins it). I've even managed to make it slower (18 and 28 frames slower). I'm not really sure everything that has an effect on the RNG. Dying in a level and then exiting a level, exiting a level without dying, exiting and entering a hub (each moving the rng, probably) all make an impact. Yet none of those actions has been performed more or less in this run in comparison to 0.9.6. I even ran SM1 and SM2 in real time from the point the hub was entered in the movie and still had the 16 frame improvement. Since Globotron was exactly the same (Attacking on frame 318 of the fight) in both the old WIP and the new WIP, but Captain Jelly is different (16 frames faster) and nothing after entering Sweet Mountain's Hub had an effect, then I'm left with only one assumption. Killing Globotron faster than the old WIP changed the RNG. The transitions from everything after the fight to Sweet Mountain's hub are optimal, and idling in hubs doesn't effect the RNG even if one WIP of the two hadn't been. I'll also attempt to find the RNG's address too. Since I know what actions can force it to change, I'll eventually be able to find it/them (I hope it's only one lol). EDIT 3 Well, those are the only four results I've come across (306, 320, 322 and 332). I've been measuring by when the first minion hits the ground and produces a shockwave. I tried to RNG Globotron, but he always attacks on frame 318. Even fighting him directly after fighting him (which is my theory about the RNG changing), he still attacks at 318. So, I guess he's not effected by the RNG. I can't help but feel that those four differences for Cappie are really odd. It's a 16 frame difference, then a 2 frame difference, and then a 10 frame difference. There's no semblance of a pattern at all.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
It gets really good after the running is unlocked. The other abilities are pretty interesting too, I just wish the game was a little bit faster without the running (Why wouldn't the devs make running the FIRST ability?) but it's still a good game to run regardless. I'm voting yes. for a Barbie game >_>.
Post subject: Video dump of "Main Screen" in DeSmuME?
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
The main screen option works pretty well for choosing which screen to display in DeSmuME for Sonic Colors (Minus boss fights, but both screens are handy there). The screen being used switches depending on position on the map. Sometimes it's top, sometimes it's bottom, and switches multiple times during any given level. I was wondering if there was a way to dump what is being displayed as the main screen only or if I'll have to manually crop it myself. I've got no knowledge of how lua works and I imagine that would be one option. However, I don't want to trouble someone else to make a script and I don't have the free time to teach myself with college, so that's out of the question. I'm crossing my fingers for another option that I'm overlooking though lol.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
SM2 Finished. I'll get the .dsm uploaded once Captain Jelly is finished (sometime tomorrow). Out of all the levels, this is my least favorite one. The layout rarely is even for more than a second. It's full up of ups and downs, and it's a pain to optimize. Link to video
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Finished TR2 7 frames faster. Time is now 37"20. Crossing my fingers that the rest of the WIP will sync without a hitch. Sweet Mountain isn't nearly as laggy as Tropical Resort (I think because of the backgrounds. SM's are static). EDIT Woooooooooooooo, the rest synced up perfect :D. I can get back to work ^_^.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I doubt someone would be running a game for two and a half years without any peep about it <_<.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Sync effort is going pretty well. TR2 has to be manually finished due to an improvement I thought of, but I'm almost done with it. FFR, all future .dsm's are going to be the US version, not the Euro versions anymore. Thanks Nitsuja for writing up the lua script to cover all three bases.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
http://dehacked.2y.net/microstorage.php/info/1864544706/Kirby_Mass_Attack.dsm It actually ends five or six frames before the star is controllable, but no extra input is necessary.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Took a little break and decided to screw around with this today. Here is what I learned in four hours of messing around. Double tap running seems like it's faster than anything else, even with a single Kirby. Double tap running can be achieved in six frames. Frame 1: Tap wait four frames Frame 6: Tap (and hold as necessary) A Kirby can be throw every five frames. Each individual Kirby has a 30 frame cooldown timer before being throw again, unless hitting the ground. The cooldown addresses for each Kirby are dynamic and change each map change. I didn't bother much with finding memory addresses. A neat trick is to knock a Kirby into the ground. He'll bounce up and can be knocked again during this bounce (due to hitting the ground). I'm sure this will be useful, but really situational. Entering a door or canon with one Kirby will suck the rest of the Kirbys into it. I don't think the "cloud" will be used except when it's absolutely necessary. It seems like it's the slowest method of moving a group of Kirbys around. I figured this might be a fun side project, but I can see otherwise. I would love to see someone work on this though since it'll make for an entertaining run. If anyone wants my .dsm, they're free to have it. It goes right into the first level and ends input the frame Kirby is controllable (or rather, the Star). Just leave a reply and I'll upload it. I think I'll run this if no else has after I'm finished with Sonic Colors. There will definitely be a lot of attention needed for route planning with fruit, but I don't think this game would be that hard.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
To be honest, I wouldn't mind seeing "streamlined" versions of runs with no loading or cutscenes. But if you get into that, then what happens in games where there are interactive cutscenes that are a part of gameplay (RE4 comes to mind)? You know what will ideally happen in those cutscenes, but they ARE gameplay. I guess the issue of instant-loading is actually Pandora's Box.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
pirate_sephiroth wrote:
This hardware thing is just the same as Left+Right input. Some may quickly jump and yell 'THAT'S NOT POSSIBLE WITH AN ORIGINAL CONTROLLER". Well it becomes possible after months of use, after the dpad center digs a depression in the board. You'll get all 4 directions pressed at the same time but that's enough in most cases. Are those defective controllers allowed in SDA? I don't think so.
Not exactly a good comparison. The controller is completely obvious if using Left + Right together for an advantage in certain games. There's no way to pass that off as anything but that. Yet faster loading could easily be passed off as a "fresh" last gen gamecube with a "fresh" disc. Anyways, I know this isn't SDA. Being that Metroid Prime 1/2 runs will likely never sync up correctly due to hardware differences, I actually wouldn't mind seeing them ran with instant loading. It'd be something completely different than anything else that is out there for them. As long as the game is emulated properly enough that there isn't any emulator exclusive bugs, abuse or tricks. If realistic loading times get included later in Dolphin, why not just have separate categories for the different runs that use instant-loading effects? There are tons of games with different branches as it is <_<.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I haven't thoroughly tested the US version since DeSmuME updated to 0.9.7, so I decided to give it a whirl today. There's definitely not the big pseudo-lag difference there used to be between the US and Euro versions. It's more like a trade-off. The US version still has more lag to it by a few frames (Although the WIP syncs perfectly until TR2). The beginning has one less lag frame in the US rather than the Euro. Then the US version lags for three more frames. They even out at around frame 8800 on both runs, then the US version finally gains one more (US 4, Euro 3). I've had to edit the dsm seven times so far, but it seems like I might actually be able to convert over to the US version now. If everything stays smooth with syncing, then I'll only stick with Euro if times are worse in total with the US.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
DarkKobold wrote:
pirate_sephiroth wrote:
Everything was said in the first replies... loading times don't belong to gameplay and consoles don't use always a single brand or model of optical drive and EVEN if they did, it would be pretty impossible to emulate perfectly all action (and faults) performed by the hardware.
If we can't do something 100% perfectly, don't do it at all?
"It is futile to do with more things that which can be done with fewer". In Metroid Prime (which seems to be the only project mentioned in this thread that is being effected right now) runs over at SDA, what is the consensus about first gen. gamecubes versus last gen.? Seems like the last one released that has the lowest laser resistance is at the biggest advantage. If it is indeed fair game to use any gamecube in those runs, then why is there problem with having no load times at all? I mean, all I'm saying is that if their stance is "Well too bad" for those with launch gamecubes, then why not adopt the same "Well too bad" attitude with having no load times? Also... I'm not a hardware modification (or laser for that matter) expert, but isn't it theoretically possible for someone to modify or swap their gamecube's laser to have a resistance of 2 or 1 ohm? I know there's a "no hardware modifications" policy at SDA, but what's stopping someone from changing that 66 ohm down to a 55 ohm? That person could blame their "fast" load times on someone else's just being "slow". There should be an "all or one" rule. If any gamecube is allowed in these runs, then I think the instant-loading Dolphin should be in the clear. But if there were a standardized model to prevent a laser swap and/or balance the playing field, then that's what the emulator should be based on. I'm tired, I hope I got my point across xD.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I looked around and I couldn't find this thing in specific. I was wondering if there is a TAS of someone doing the warp in meteo in cockpit view. I managed to do three rings as a kid, but I eventually gave up on doing it because it's really hard. It's one of those "Now I can die happy" kinda things lol.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I really really wanna see Kirby Mass Attack get the TAS treatment.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
So in the end, I'm going to have to restart SM2 again. I need about two more frames. While the regular sonic section is two frames faster in this WIP, everything else is wrong. The coordinates where Burst is attained, activated, speed value are completely wrong. I'm going to have to figure out what value speed needs to be to get right of the balloon in 12 frames (Too high, go too far. Too low, I'll hit the thing after going upward). I never got around to answering you Paused (Suuuuuper Sorry!) Here's the run-down of how it does work. You can use Burst's explosion the frame after transforming. So 1: X, 2: B/A/Y. There are three levels of explosion. These levels differ by how many frames a button is held (B/A/Y/X). The frames required for each stage are at: Press X at Frame 1000: Frames 1001-1010: would be Stage 1 Frames 1011-1030 would be Stage 2 Frames 1031 and onward would be Stage 3. Stages 1-2 will trigger the next frame after the button is released. There's an exploit in Stage 3 that allow Stage 1 & 2 to be used during it's down time (Takes 15 frames for the Stage 3 Explosion to trigger). Using a button aside from the one just used for Stage 3, the frame afterward any button can be pressed/held and activate a Stage 1 (or 2... Although it's certainly not worth charging up a stage 2). This Stage 1 explosion has no decay on it at all. A Stage 3 Explosion also delays reverting back to Sonic by 31 frames if used on the very last frame of Burst. Charging makes you fall slower, which is useful. X-Velocity takes longer to build in the air than on the ground, so it's not wise to charge while airborne without having a reason for it. Burst has exclusive interactions with certain objects. In this run, you'll see it in Sweet Mountain 2. The Hot Air Balloon will rise upward with Red Burst, but sinks with Sonic. There is also a Ferris Wheel kinda object on Tropical Resort. That won't be used in this run (it might in a 100% run, if one were ever made) Transforming into burst takes whatever your speed value (not x-velo), and changes that into your current x-velo. So if transforming while in mid-air at a dead stop (x-velo is 0), but speed was 61440, then it would change x-velo into 61440 the next frame from X being pressed. It's unique to Burst as I know that Cyan Laser, Drill and Orange Rocket don't behave this way. There may be a similar exploit with Void to a lesser extent (it has a "cap"... hard to explain off the top of my head). Sorry for being extremely late Paused! :( Update 2 Found a place to use a trick I learned during my first run to save a lot of time here in SM2 (25~ frames faster than my old WIP). In total, I'm 50 frames faster after 14 seconds into the level. The trick is getting the speed value below 3072*16, which allows using boost to immediately go in the opposite direction while traveling down a wall. Rather than slow down normally, it's much better to boost in the opposite direction while running to slow down. There is an 8 frame cooldown for boost, so that has to be accounted for when using boost as an accelerated method of decay.
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Final Time - 50"11 (29 frames faster). I'm really happy with the whole burst section. I already was before, but now it actually looks really smooth through-out the whole section. EDIT Two frames faster than my old WIP (due to pre-burst section) in Sweet Mountain 2. I tried to surpass what I did last with bouncing on the bomb-ballons during the burst section. I found it nearly impossible to even match it. This is because the setup is slow, no matter what way it's performed. 1) go to the right of the balloon, then travel upward and fall onto the top right of it (done in the old WIP). T 2) The other option (which is done in this WIP) is to travel up immediately, which is to the left of the balloon. The move right and fall on it. Both of them have their own pro's and con's. Option 1 is initially slower, but makes up for it by being able to exclude some the height. Option 2 gets immediately at it, but enough height has to be achieved in order to maneuver to the sweetspot on the top right. So far, I've matched the results identically with both. It's pretty rare that two things are perfectly equal. I'm leaning towards the first option being the better of the two. If I can't get a better result in either option, I'll just leave the second option in (looks better imo).
Experienced Forum User, Published Author, Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Redoing the last quarter of Sweet Mountain 1. I overlooked a section from my 2010 WIP that was actually faster, back when I didn't have values (couldn't find them back then... not sure why), than what I have now. Bursting with Red Burst (fire wisp) is similar to another game's mechanic, smash brothers. You cannot charge up another burst, until Sonic is moving downward (which is similar to smash in that you can't grab a ledge until moving downward). My old WIP hit the ceiling, which turned out to be faster because of this reason. It's much more fluid and should save at least 10 frames (would be 30+, but it is initially 20 frames slower... Going slower, is faster :D) I'll have the exact figures later once it's done. I won't be uploading the new version to Youtube since it won't be a very noticeable difference. Hopefully this change will permit "some" copying + pasting of the rest of the level.... *crosses fingers*. I believe that I'll have to manually do the whole bit up until stomping to the right of the spikes with sonic sans burst. Hopefully the coordinates should be close enough at that point (due to being ejected from the spikes slightly) that they'll either be identical, or "good enough". EDIT I actually found one other part (the others were done correctly.. not sure what I was thinking during the time for the last two parts) that used the extra bursting glitch when it shouldn't have been used. So this should end up being at least half a second faster in the end. I'll be working on this more than I have been, but not a whole bunch. I'm in dire need of some glasses. My eyes have gotten bad enough that it's hard to focus on any screen for long periods of time due to squinting. I'll most likely set a frame goal for each day. Probably 600 or so. That takes about 30 minutes to 3 hours depending on the section. EDIT 2 All in all, three changes have been made that leads to the burst lasting all the way until the balloon. I figured that I would try and make use of the extra wisp time by taking the higher path. Unfortunately, there is not enough time left to make it all the way up (about 1/3rd). Burst will not revert to normal Sonic if a fully charge burst is used before time runs out, until after the fully charged burst is over. Instead of wasting time optimizing getting a burst wisp, I just edited the value. In my test that was unoptimized, it takes over 150 frames, but I doubt it would close to sub 100. The whole red burst section is extremely optimized as far as my capabilities will allow, so there is no way to reach that section with an extra <100 frames of Burst (300 frames are allocated when first transforming). So it would require resetting that counter by getting yet another Wisp Capsule, which makes the whole upper route out of the question. I'll tally up the savings once it's finished either tonight or tomorrow. EDIT 3 Saved around 40 frames from the exact spot where I'm at. Unfortunately, the psuedo-lag is present for a frame, so I'll have to redo the rest of the level :/. The coordinates were exactly the same, which really got my hopes up lol. It's a little bit different than the previous attempt I have uploaded, but as I said, I won't bother uploading it to YouTube (I will upload the .dsm however). I'm not sure how I'll deal with the psuedo-lag frame, hopefully it won't cost me more than a frame or two. I'm leaving it for tomorrow xD.
1 2
7 8 9 10 11 12