1 2 3 4 5 6 7
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Got my scholarship refund last week and I've been distracted by actually having extra money for the first time in months. I'm going to get back to work on this starting next week. A week off was nice as I had been working on this almost every day since Christmas Eve.
Skilled player (1307)
Joined: 9/7/2007
Posts: 1354
Location: U.S.
Any updates?
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I've done 3 seperate runs of Globotron and can't get lower than 0'12"93. I tested it before after beating Captain Jelly in my last WIP and got 0'12"91. If I don't get it after the next two tries, I'm going to just leave it as 0'12"93. I've even tried not moving at all except to avoid getting hit and to attack first frame possible. Still 0'12"93. March 8th Same thing again. If I don't get it my next try, I'm moving on. It's been nearly a month since I've had any real progress.
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I think it was just a fluke >_>. Moving along. I'll post my WIP when I finish SM1 as there isn't anything "new" in the Globotron fight.
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Not sure if I posted about this earlier or not. I tested it out though. Tested something out that I've been wondering about as a possible improvement to TR2. Holding left and jumping after being on certain slopes will give a higher that ground max x-velo (not a lot though, like 64448). I was thinking maybe there would be an improvement on TR2 after the first loop by holding left and jumping before hitting the water (landing on the water before x-velo would decay below 61440). I did that, but x-velo would be reset as soon as right would be pressed. So I decided to test out if jumping on the water at all would reset x-velocity (by freezing x-velo at 61440) and it does. However I did find a trick that might be useful later on to speed up running on water sections if Sonic is under 54528 when reaching the water. Anyways, on to SM1.
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
So I've gone back to running Globotron LOL. I'm on spring break this week, so hopefully I can make some good progress during the week. Hopefully I'll get through SM2 by Sunday (I don't think I will, but that's the goal). Looking forward to seeing the improvements in SM1 though, as I had a hell of a time optimizing to "perfect" last time. It should be a lot easier this time around with lua.
Joined: 5/9/2005
Posts: 748
How's it going?
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I've been slacking :(. SM1 is about halfway done, but I haven't touched it in the past week. I've gotta focus and get this run done xD. I feel really bad that about slacking at all when Nitsuja did all that work making the Lua script. So it's definitely going to get done, just been going at a snail's pace lately.
Skilled player (1307)
Joined: 9/7/2007
Posts: 1354
Location: U.S.
Dont worry! Its better to keep working on it for a long time then to rush it. That will make an amazing movie. ;)
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I'm still taking this at a snail's pace, but once this semester is over in five weeks, I'm going to dedicate all the time I need toward finishing this :D. In the mean time, I'm going to finish up SM1 either this week or next (three papers due this week! -_-) and upload the dsm so there's some shown progress. Already a full second faster about 30% of the way in (restarted it).
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
End of the semester is May 18th for me. As soon as the semester is over, I'm back to working on this full time. Starting with Globotron redux. EDIT Taking my last final today and I'm going to start this TASing overtime :D
Joined: 5/9/2005
Posts: 748
VanillaCoke wrote:
Taking my last final today and I'm going to start this TASing overtime :D
How's it going?
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
There hasn't been real progress. I've doing a test run to get back in the rhythm again. There's a lot of little things that I've had to relearn, like conserving momentum, skipping hitstun when killing an enemy, breaking max speed when jumping off slopes (albeit slightly). I believe I have gotten it all down now. I'm gonna mess around for the rest of today and then I'll be back and at it. 5/26 So I'm going to just attempt to splice the Globotron fight from my previous run on Desmume 9.6. It was a lot flashier than what I currently have. I've tried to improve this for a long time now (off and on) and I just can't. I'll post my .dsm and upload the fight to youtube once it's done (copy + pasting certainly won't work due to the game having random frames of lag that always differ, but I can mimic and accommodate for those lag frames).
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Link to video I'll be posting the .dsm after SM1 is finished. Sorry this isn't "new" per-say. I figured since I haven't been able to replicate an improvement in 50 or so attempts since February, I might as well just go with the original. I wish I'd of kept the '12"91 xD, but I figured I could reproduce it at a later time. Good luck to whomever attempts to obsolete this in the future! I spent about an hour and finally found a good sweetspot that led to only 1 frame desyncing in the copy+paste. All in all, one lag frame was eliminated rather than any being added. Then I spent the past hour or so uploading it to youtube. I'm about an hour late posting this too haha. I'd of gotten more accomplished tonight, but I was busy watching the heat vs bulls :(.
Skilled player (1307)
Joined: 9/7/2007
Posts: 1354
Location: U.S.
Theres something ive been meaning to ask sice you started. How come you have been using the European version of Sonic Colors? Nice job on Globotron!
Joined: 5/9/2005
Posts: 748
That is because we can spell correctly, silly American.[/Troll] Great to see the 'new' video VanillaCoke.
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Back when I tested it on 9.6, euro version had less lag frames than the us version. The beginning credits were also faster to get past, but I believe I read somewhere that-that is a negligent point. It's the weird kind of lag I talked about earlier in the year. Every input but jump works during that frame. There's a lot more of those in the US version when I tested the two of them in 9.6 Also the european version is in English, if someone wants to pause and actually read during the scenes they're able to do so. I tested it out to see if it was possible to switch regions and watch the playthrough in Spanish, French etc. Unfortunately it desyncs :/.
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I find it weird that sometimes excess x-velo can easily be stored by spinning (pressing down)/sliding and holding back, and other times it doesn't. However I found that jumping for a frame and holding back does just as good of a job, so I'll be doing that from now on. Spent yesterday figuring out the best method of using the first set of Jelly in SM1. Just to make sure, I optimized using both the first jelly and the second (I did this in my previous run, but I'm a lot better this go-around so I had to do it again). Second Jelly is still the better choice. Edit Calling it a night for tonight. Tons of tiny little improvements. So far I've saved 24 frames in the first 8 seconds in the level with all the tricks I learned while making this second run. Glad I got a Nitsuja script that works when nothing else does hahaha (RAM Watch addresses are frozen). The ghosts are infinitely useful. Can't over state how useful this script has been. I even saved time getting hit by the Orbinaut (3 frames). Last run I didn't realize that x-velo before being hit effected your x-velo after the hit (figured it was fixed like in most games). Not leaving any stone unturned this time xD. 5/30 Decided against taking the route in my previous run with Red Burst. I tested an unoptimized run against the previous run and I've lost 26 ms (milliseconds) on my 86 ms improvement at the slope after the popcorn shooter (didn't check frame count). I believe I'm going to be able to cut that 26 ms lost by a lot due to having an unoptimized speed boost off a rail (5812 lua value, unoptimized) and now I've gotten 6186 (lua value). I've also decided to test and find out if it would be faster to just grind the rail normally rather than landing on the end of the slope rail for the speed boost. It eats up a lot of time for the speed not to be conserved very long (I've tried a lot of different methods to save it through the loop, but none of the worked unfortunately. I noticed Nitsuja's WIP for Rush also lost the big speed in a loop too.. so if he couldn't save it either, I doubt there is a way to do so). Edit Figured out a new trick that might require me restarting the run ;_;. Basically it's a trick with boost that slows down decay on regular jumps. Instead there being a huge decay the second frame after jumping, this drastically slows decay for 9 frames (Lua value jumping and holding back brings speed down by around 100. This cuts it down to 6 per frame!) How it's done: Boosting two frames before hitting the ground, then jump + holding forward a frame, then back for 9 frames. I'm 95% sure I'll have to restart from TR2 (Can't be used on TR1). I'm going to go back tonight and look at just how much time can be saved using it. update (sorry everyone in IRC xD) I messed around with this some more and found that it's only useful for "small jumps" (<15 frames) rather than big ones (16+ frames). There's only one section that benefits from this (Frames 8469 - 8498) and might give a frame improvement. Other than that, there only large jumps or sloped jumps that this wouldn't apply to TR2. I'm going to check SM1, won't update this post until tomorrow night. 5/31 Five frames faster than my WIP from two days ago with this new trick. Unknowingly, I've been using this trick (although not at it's "full" potential) in the run. I've been using a system of going faster (when it returns with the best results) of in this sequence; 1: A + > 2: B + > 3-6: <- 7: Y + <8> 9-10: <- This left me with leftover slow decay frames. But even checking back, there isn't but one part that this can be used on in TR2. I'm interested to see how this effects using slopes for "higher than ground cap" x-velo. Such as landing on frame 10 of the technique, sliding from frame 11 until hitting 61440 speed, then having some leftover slow decay frames which would delay the use of boosting to also slow decay. Unfortunately, this would only be applicable to constant slopes and some curves, but not loops (which result in the best x-velo boost in this group mention). This should also be really beneficial for curved rails (gives the best x-velo boost period), but there won't be any use of this until possibly SC1, as there aren't any rails in TR2, SM1 or 2 that I can reach from under them. Update There is too much lag to perform the slowed decay on the ending slope of TR2. I tested it out after the enemies were gone, and optimized the test. Too many frames are wasted for it to be useful since the spring has to be hit. The jump has to be performed later and leads to only an extra frame of lessened decay. The spring is hit in the same amount of time at the same y coordinate (which is all that matters). I managed to actually do a decent job of syncing the improvement to the rest of the run, however the section at Frame 9515 ends up having more lag that ruins the improvement due to inputs being dropped (tried to correct this manually). I tested to see how this trick works with boosters. They override each other, as in whatever is performed second is what is going to be left. I should have SM1 done sometime Friday or Saturday. I've been taking it kind of slow due to NBA Playoffs and now the Finals xD.
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Lost what I was typing by hitting F5 LOL. This'll be short. I'm going to redo TR2 from about 16 seconds in. I decided to do it because I don't want my submission to be trumped with someone simply utilizing the decay trick in parts where I didn't :/ Shouldn't take too long though xD. SM1 is looking pretty good, hopefully it'll sync up without much fuss. Edit Testing out a new route in TR2. 6/9 Alright, the route is pretty much TAS-only and saves about 40 frames. There is going to be another route change, but I'm not going to go into any details as I want it to be a surprise. I was messing around and found something awesome that's only applicable to TR1 and TR2. After I upload the finished TR2, I'll go into more details :P
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
I'm not sure of the mechanics behind it, but I've managed to produce, then replicate and improve a glitch that kept the breakable walls for loading. This will save 6 frames due to there being no 3 frames of hitstun (which freezes x-velo to 0 for the duration) from breaking the wall, 3 frames less decay on x-velocity, and 3 frames of actual movement. Will have this done late tonight or early tomorrow. This won't disappoint! :D
ALAKTORN
He/Him
Player (99)
Joined: 10/19/2009
Posts: 2527
Location: Italy
I'm not sure what you mean, but it sounds sweet!
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Going to take a little longer due to an optimization I made in the second route change. In the end, the optimization is 36 frames faster than it was last night. There is only one more part that I'm going to test and then the rest of the level is pretty much going to be the same as before, so I can "copy" the inputs. Will have this new version of TR2 completed before I go to sleep, about 2/3's done with the level as of right now. I'll edit in the YouTube upload after I finish.
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
Finished with a time of '37"32. Uploading to YouTube now. I'm too tired to upload the .dsm tonight, but I certainly will tomorrow. Overall, that's a '5"25 second improvement. I'll also explain the mechanics (from my understanding and execution) behind the zipline glitch tomorrow. You'd think going this fast would of lead to the level being easier or shorter to complete, but it actually took the longest. It was refreshing taking two different route changes though :D. Link to video
Active player (458)
Joined: 12/24/2010
Posts: 297
Location: CT, USA
http://dehacked.2y.net/microstorage.php/info/1213309895/Sonic%20Colors%20Test.dsm Sorry it's been almost a week, I was planning on uploading this later on last Saturday but I've been slacking off this week haha (Been playing DNF). So from my understanding of how this zipline glitch works, the first time it takes the difference of your x-velocity when boosting backwards and adds the difference from it and your x-velocity the next frame, then frame after that it makes that value into speed. So for example (this isn't the exact figures from the movie), 41952 is the value that boosting sets your x-velocity when going right. Boosting left makes it temporarily -41952. The difference is 98304. Add that to the speed value of 41952. Adding that with 98304 makes the value 140256. That becomes the new value for speed. When performing the glitch a second time, the difference is added to the speed value again. So 140256 + 98304 = 238560. Formula: (X + -Y) + Z = S X represents x-velo right Y represents x-velo when boosting left Z represents speed S represents New speed The glitch can't be used right after getting on a zipline. Y-velo has to be 0 and depending on the x-velo prior to getting on the zipline, it takes 1-3 frames until it becomes 0. Last note I'd like to add is that this formula completely break when hitting a zipline with an excess x-velocity (65792 for example). The results are completely different and there's a LESS of a boost in the speed value. Nitsuja gave a great description of how speed and x-velocity works. Basically Speed value is possible speed. If it is higher than x-velocity, then upon landing x-velocity will become whatever Speed's value is. X-velocity is "Actual speed" and that is how fast you're moving each frame. So that's why moving so fast doesn't occur until after landing from the zipline. The reason I didn't building up any extra speed, or even actually use, the second zipline is for a plethora of reasons. For one, x-velocity goes back down to regular values of 41952 (boost) or 40960 (non-boost). Too much time is lost setting it up. A lot of time is lost due to having to having to use the launch at the end of the zipline to get the launch. There's waaay too much speed (There's an oxymoron for a Sonic game) is gained going onto the rail. It makes it impossible to land atop the loop, wasting more time. It's still possible to glitch the breakable wall into not loading. However x-velocity has to be dropped down quickly in order to make it under the wall with spikes on top of it. Then to top it all off, all the extra speed and x-velo is lost after reaching the loop. Overall it's not worth using the second zipline at all. Incase anyone is wondering, the ziplines aren't worth doing this glitch on Tropical Resort 1 for a big reason. Every area where there is a zipline, there is something that makes the boost in speed lost shortly after. There is one of those walls that transports your downward (the mouth thing) at the first zipline in the level. Then for the second zipline, homing attacking up to the second zipline loses all of the gained speed. Going down early would be a possibility if it weren't for the need to hit ground to get x-velo to copy speed's value. Using it on the third zipline actually has two reasons. The curved rail forces x-velo down, losing all of the extra speed. The other big reason is that even if you skip the rail until it's evened out again, the breakable walls lose all of the speed+x-velo. Anyways, I'm going to spend today trying to sync up the rest of the run with this improvement.
Skilled player (1307)
Joined: 9/7/2007
Posts: 1354
Location: U.S.
"Amazing!" work VanillaCoke! Cant wait to see the final product.
1 2 3 4 5 6 7