Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
KusogeMan wrote:
1 I literally just turbo A in those sections, i really didn't bother
2 I learned that this mechanic existed only on Cream Story lol, but yeah fair enough
3 I also didn't know that as well
4 I'm aware but this limits CPU speed, the real reason for equipping cards is for Emerl's partner behavior, I strongly disagree here
5 Mistake really, the RNG in this game is very set i believe it's like set when starting the game as no actions seem to change the drops for 3 to 4 fights later, never tested a whole story later though
6 No, there was a tremendous lack of cooperation in the fights i forego that, but what really would save a lot fo time in the first 4 chapters would be Walltech instead of guard which gives more meter earlier, but trades HP for tht extra meter so it can't be abused all the time, all Stories before Amy would have multiple faster section by using the best possible Walltech setups which is: get hit by a bouncing move, wall tech for extra meter, and cut the wallbounce jump short like Emerl does in the later parts of the TAS. Obivously limited by health, but here and there would speed up all fights with some 1 or 2 faster instakills.
7 Lol.
I appreciate your thoughts about this TAS, and might add as well, explore a lot of the walltech cut short to improve mostly the early parts of this TAS. I would gladly watch your WIPs to comment in some fights if you're willing to share videos with inputs or moviefiles. I sincerely hope you're able to get the Ultimate Air Action to kill Rouge and Eggman and the earliest Knux or Cream Heals as well.
I'll be looking forward for your runs as well.
4. Ahh, so it was not just to speed up Emerl fights, but also for the CPU? Understandable.
6. Ahh okay. They definitely are very uncooperative, so that does make sense. And yeah, during Sonic's story I was wondering if you were going to bounce off the wall to fill the meter quicker, especially during the Gamma fights, and I was a bit puzzled when you didn't.
I'm unsure how I will upload the movies to user files, but I can definitely upload videos of each story. Progress will be slow since I've been super busy the past few months and have had little time to work on TASes, but I'll do my best! Good luck to you as well!
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
This run was really good! I've also been working on a TAS for all stories (on JP instead of US), and there are some things I noticed while watching that may have cost you a lot of time. They are really just minor nitpicks, but over the course of an almost 3 hour run the timelosses do add up quite a bit. I haven't gotten to Knuckles' story and beyond, therefore I'm not as knowledgeable about those, so I'm going to focus on the first three stories.
1. For inputs on textboxes, I noticed you have A being pressed on every other frame. This in itself is not bad, however I noticed a lot of textboxes were staying onscreen for an extra frame, which loses a lot of time over the run. My recommendation is you should go through and try to press A on the first possible frames that you can advance/close textboxes. Scrolling textboxes takes an extra frame, (for some really dumb reason) but on the card screens and at the end of other textboxes you can press A on the first possible frame that the text is done being printed to move on. And for textboxes that only print one line of text, there's always a delay between hitting A to speed up the text and hitting A again to advance the textbox.
2. On the overworld map, sometimes when you changed areas, you lost some time to not skipping the text that shows you what the area is named.
3. On the overworld map, you can hold B to move the cursor faster, but I saw for most of the map movement early on, you only do it a couple times. Doing it every time would probably save a few minutes over the entire run.
4. In Sonic's story, you put in the Tails' run card for Emerl. I believe this may have lost time, as I found out that you can continually turn Emerl left and right in battle to move very quickly without changing the run card. The only issues with this trick are it doesn't work in midair, and you have to time the turns right otherwise Emerl does his little dash and most of the time you don't want that to happen. I could be wrong but testing will need to be done.
5. In Tails' story, after the third Robot fight, you delayed exiting the card screen. I'll let this one slide, as one can assume you may have done that to manipulate RNG.
6. Throughout the run, I think with some AI manipulation, you could probably get more meter fills from blocking the enemy's special attacks when their meters are full which might help with faster kills.
7. Tails is too whiny in the run (just kidding ;D)
The overall RNG looks like it's pretty good, but I think with these improvements it could potentially be better. Or worse, cause RNG sucks in general.
The strategies you put into the fights are really clever! I also like how you used the Emerl skills to speed up the fights. I completely understand if you wouldn't want to improve this, since literally any improvements (in and out of battle) result in different RNG and means you have to do EVERY fight ALL OVER again, and that's really irritating. Other than that, this run was really fun to watch, I give it a Yes vote, and I wish you luck if you do decide to improve this!
One final tip: If you don't already know this, (I assume you do since at a quick glance you're dealing an optimal amount of damage) in the US version you can find the health for the fighters in IWRAM at RAM addresses 1CFC for player 1, 1DF8 for player 2, 1EF4 for player 3 and 1FF0 for player 4, and with those you could bruteforce inputs/RNG to deal as much damage as you can in as little time as possible. If you put them into RAM watch, the addresses you want to put in if you view them as 1 byte is 1CFF for p1, 1DFB for p2, 1EF7 for p3 and 1FF3 for p4.
i may also steal some of your strats for my run (i'll make sure to credit you) ;)
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
I believe the rerecord count would be 18,176. The reason the rerecord count on the file is so low is because of three factors:
1. I created the improved run from a brand new movie file. This was before I found there was a “Splice Movie” option in Gens (which I found out AFTER I had already submitted the movie) which would have saved a bunch of time actually making it.
2. I had used a program to copy the inputs into this new movie file. I had done this because I didn’t want to copy all the inputs by hand into this movie file so I went with a more convenient option.
3. I did most of my optimizations for the levels in Bizhawk, again because I got lazy and still wasn’t used to the Gens TASing style, and found TAStudio to be more convenient. After I had optimized the levels I wanted to, I copied the inputs from the movie I had in Bizhawk to Gens, and for a couple of the levels made adjustments due to the differences in emulation.
If we included the count from my Bizhawk movie then the count would be 19,856, but since those weren't actually done in Gens, I am not going to count them.
I do apologize for the inconvenience that this may or may not have caused. I got super lazy and didn't put in the effort in Gens and did a lot of the work on another emulator.
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
Alright, so rodonic01 might have figured out why this might be happening. The reason I was having issues with the BIN/CUE dump I have is because I have all the tracks dumped as seperate .bin files, and apparently Gens doesn't like dealing with a cue and multiple bin files. Perhaps that could be your issue as well Thunder? If that is indeed the case then your best bet would be to use a tool like CHDMAN to compile all of the bins into one bin file and then make a new cue file for it.
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
ThunderAxe31 wrote:
I'm failing to sync this submission, I get stuck to this screen. what am I doing wrong? I'm using a confirmed dump of the game and the exact bios as indicated in submission.
Are you using a BIN/CUE dump? I've tried using BIN/CUE on Gens multiple times and it never works for me (I get the same result you're getting), and the only dump I've gotten to work on it was an ISO/CUE.
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
Honestly quite incredible run! I enjoyed this a lot and have given a yes vote!
However, I spotted one area where some time could potentially be saved. In Lava Reef 1, you opt to go for the monitor clip using the fire shield monitor that the signpost bounces out of the ground. Is there a reason you didn’t go for the monitorless method? Does it affect RNG later in the run, or was it just something you never thought about doing?
For anyone who doesn’t know about this “monitorless” method, during the second time the hand comes up, you can jump at a specific point and if you hit the hand on the same frame that it smacks the ground, you can be in a weird state where if you roll while in its hitbox, you can actually clip right into the floor and still be able to get the final hit in. If you do this in an area where there are no monitors present, the signpost will hit the ground once and you’ll actually save roughly 2 seconds over the old monitor clip method. This trick is extremely difficult to pull off, but it is definitely doable, as I have been able to pull it off on a few occasions.
I believe this was initially found by RTA speedrunner eandis, and his video on this trick can be found here: https://www.youtube.com/watch?v=NRTK4ZJIoKI
All in all, excellent run, SP! I look forward to whatever you have in store for us in the future!
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
Another day, another optimization found!
This time, it's in Landslide Limbo! Instead of going around the boulders before the boss, let's see if we can squeeze between them! This trick was shown to me by SpectralPlatypus, and it saved 15 more frames! The updated movie can be found here until the submission file is updated (again :v): https://tasvideos.org/UserFiles/Info/638237285397274783
EDIT:
Another 14 frames saved in Landslide Limbo! I think we're hitting the limit of what we can do, but hopefully we can find more improvements someday.
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
I'm probably a bit late to doing this, but I figured I'd better make a post here in case people are wondering about all the edits and why an improvement was submitted so soon after the first publication.
So after the first submission was published, and I returned and was able to start working on stuff again, I figured I wanted to play around in SegaSonic for the lolz and see what else I could find that could potentially be an improvement. The day after I got back, rodonic messaged me with a 14 frame improvement in the first part of the game, which we added to the original movie and sent to ThunderAxe (the judge) to update the movie file. The next day, me, rodonic, and SpectralPlatypus had the following breakthroughs:
I discovered a faster way to fall off of ledges, which saved over half a second in Icy Isle and contributed to another timesave in Landslide Limbo.
Spectral showed me a faster method of reaching the second set of rafts quicker in Wild Water Ways, which also saved about half a second.
The week before, Spectral had theorized that maybe we could skip the vines in Landslide Limbo using intentional deaths and respawning another player on the other side. I toyed with this in one-player mode, and found out it was viable for singleplayer runs, not just multiplayer. This saved another second and a half, and brought us past the 10:31 barrier into the 10:30 region. Like I mention in the submission text, there is a variant of this that is over a second faster that involves spawning the second player in after the first player dies. However, we agreed this did not fit the "one-player" branch that we were going for, so we skipped out on that.
These last three optimizations not only helped the TAS with saving time, but it's also helped the RTA SSTH speedrunning community save time as well! After we found these, we put them into the movie and had the input file updated again. So far, we haven't found anymore improvements (as far as I know) and we are still working on theorizing potential timesaves. If we do find any, I will either make a new post here or wait till the next submission.
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
Hopper262 wrote:
I'm not familiar with this game, but I'm surprised that avoiding (nearly) every ring doesn't cost time overall. For instance, in Icy Isles, do those extra jumps to dodge rings really not cost you any velocity or momentum? At only a 1-frame penalty, I'd have expected a few rings would end up being too much in the way for a branchless TAS to avoid.
Sorry for the late response! I was away at the time when you posted this and was unable to respond, and I just now got around to being able to.
I have not done extensive tests of this, but just from doing a few attempts at running the game I don't believe that jumping costs any momentum at all, as it seems like removing jump inputs from some areas or adding jump inputs doesn't seem to really change the outcome of the run.
Maybe taking some paths around the obstacles that are jumped over could save time? I can probably try experimenting with that, but for now I would say jumping doesn't cost much time, if any.
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
Hey Eduardo! I just finished watching this, and I just wanted to say that since (as far as I can tell) this was your first attempt at this, I would say this is a pretty decent template for this type of run! Like Spikestuff said, this could definitely be improved, and I wanted to share a couple things you could use to improve upon this with.
First, I noticed that you don't really take advantage of the physics to build up a ton of speed. The physics are so horribly broken in such a way that jumping won't take any momentum from you while in the air, and combining this with jumping on downward slopes can lead to gaining some insane speed. You would also want to avoid bonking on ledges as much as you can. This may require you to go back through certain sections to redo jumps in order to avoid bonking on that one ledge further into the level. You can also immediately come to a stop when landing by turning the opposite direction of where you're headed, then pressing the direction you ARE heading when you land, which will allow you to come to a dead stop and immediately be able to spindash. This would definitely speed up areas where you need to stop and spindash in order to get your speed back.
Second, I would probably change some of the levels that you obtained 50 rings in. Green Hill 1, Green Hill 2, and Spring Yard 2 are good candidates for getting 50 rings, but I would probably change the other three levels to be levels that you can get 50 rings in without losing too much time. Especially in Marble Zone, where you went for the ring monitors in the secret rooms. I would also recommend routing a level so that you don't get more than 50 rings, since getting more than 50 can lead to losing time since you're spending precious frames getting all those rings.
Third, lag management. This may be a more advanced thing, but there are certain things you could do to reduce lag in certain areas. When you're going at high speeds, jump as often as you can, and as soon as you can. Not only can you build up speed by jumping on downward slopes, but it also reduces the amount of lag you generate. At points like Spring Yard 3 and Star Light 3, there's lag that you will not be able to do anything about, and there's also lag that generates when you move. In the latter scenario, your best bet is to only move if absolutely necessary, otherwise just stand and wait.
Finally, I think the special stage routes could be improved. I think the routes you did use are pretty decent considering the astronomically bad physics in the special stages, but you may want to develop a more straightforward path towards the emeralds. Remember that every frame counts!
A little tip I would give if you're starting out on making TASes is to start small. Go on SRC (speedrun.com) or Youtube and look up the real-time world record run of the game you want to TAS. Study it, take notes on what could have been improved, use it as a reference of what you should do, and try to save as much time as you can over it. Once you make a run or two (or three!) and get some experience under your belt, you can try following the tool-assisted world record, and do everything within your power to attempt to beat it! And don't be afraid to experiment! Maybe you can find a faster route by changing what path to take, or find an area where better execution could save time. You can also use inputs from the tool-assisted run if you are struggling to replicate a trick. Just make sure to give credit in your submission if you do! This is how I started out making TASes, and I found it to be a good way to start out for newcomers.
I think that's all I have to say for now. Good luck on improving this!
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
Paused wrote:
Wow; you really didn't hang around with this. Not to sound like a broken record, but you're not giving me chance not to; another obvious yes vote. Especially having now compared your Chaos Ring run to Upthorns run showed a small handful of areas for improvement, which this run has already remedied and surpassed.
Nothing really to add to the description this time either. On the topic of possible improvements you mention, the only things that potentially comes to mind is switching Knux for Espio in levels where glides are not needed, that some of the real time ILs do, and possibly co-op but I've no idea how that effects the 'Call' and 'Leader system' in general. But both could be argued to different goal choice/categories territory anyway, even if they do prove to be faster.
Congrats on the runs! I've seen bits of your stuff on the discord the few times I think to pop on there, but you sure made one heck of an introduction here.
Thank you very much! Yeah, after finishing like three levels, I knew I needed to improve my Good Ending run again, but I'm planning on taking a break from Chaotix for a little bit, since I've been working on it for a good while now.
I may not decide to try switching to Espio in levels where Knuckles isn't needed. Even though my main focus is optimizing for IGT, I also want to try to optimize for real-time wherever possible, making it so I don't need to go back to the Combi Catcher again. But I will definitely have to tinker around with that at some point.
Experienced Forum User, Published Author, Active player
(310)
Joined: 8/29/2022
Posts: 15
Paused wrote:
First off; great summary of the "leader system." I've in the past been curious about the basics of the strange beast that is the Chaotix base mechanics, and my googl-ing left me still a little in the dark. Whilst I'm sure there is far more to it, I feel like my understanding has been increased ten fold by that brief outline of yours alone.
Some minor additions to your comments; only really relevant in TAS circumstances, but the negative ring trick would also be effected by ending the stage with 00 milliseconds on the clock, as that gives you a secret bonus too. And I can't recall the details, but pretty sure the time of day effects all the bosses, not just the AA one? Nothing major, just little stuff I believe; like the 'level roulette' order during the Metal fight. (Not that you'd end up doing that one any time other than dawn, but, eh.)
I get Knux is used for his glide, but Espio used for any particular reason? I realise the partner characters abilities won't be used in single player, (Though if co-op is ever done, I did wonder if Mightys wall jump would be useful over throwing.) but curious about the innate stats they have. They all have varying speeds and 'weights' don't they?
Anyway, loved the run; easy yes vote. Glad you submitted it to the site. Seem to recall Upthorn saying on here, was looking forward to seeing the old run obsoleted. Hope it does the job!
Thank you! I'm glad you enjoyed it!
I agree that there might be more to the leader system than I explained, but I just gave a simple explanation of it from what I had gathered by tinkering around with the physics.
To answer your question, I went with Knux and Espio just because I was more used to them and their traits. And yes, the other characters do have other abilities (Charmy being the prime example ;)), different weights and different speeds, and I've been thinking that maybe using Mighty as the partner would be faster, since he's one of the fastest characters in the game, and I could use the leader system to go a lot faster in levels that are more flat like Speed Slider, but I will need to do some testing before I make any conclusion on that.
Also, thank you for the additions! I've been reading over the submission text again and have noticed some errors I made, and have been fixing them as I find them. I hadn't seen anything about the 00 milliseconds affecting the secret bonus, and hadn't come across it in my runs of the game, but it's good to know for future runs! I've edited the text to include these, as well as an explanation of the character's abilities.
