Posts for byrz


Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Most of you probably saw this already, but we have released the improved TAS for SMB1 Expert through Master. Link to video Check out the workbench thread here. On other news, we have been brainstorming and playing around with the idea for SMB1 competition mode 4-player co-op all bananas individual level TASes. Sounds confusing? Basically, SMB1 has a competition mode for up to 4 players where each player plays the same floor at the same time split screen style, trying to beat other players by finishing the fastest or collecting the most bananas (each banana can only be collected by one player). "Co-op all bananas" means that instead of competing against each other, the players would try to cooperatively collect every banana in a level and finish the level as fast as possible. This concept is extremely interesting and we can't wait to see what kind of crazy recordings we can create. To make things even more interesting, we are planning to synchronize the finish of every player to take place on the same frame. Just imagine 4 monkeys approaching the goal from 4 different directions to finish at the same time.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Actually I've had the idea for max unique bananas TAS for SMB1 Expert-Master for quite a while now. If I would make a TAS like that, it would mainly be a playaround with no intend on optimization. There are several floors with no bananas so it would be quite boring just copy the inputs from the regular TAS, I'd probably have to do some kind of stunts instead to keep it interesting.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
PAL version is definitely the superior version in SMB2 when it comes down to how fast it can be theoretically beaten. The first SMB2 TAS was played with NTSC version because the main player base of that game doesn't have the PAL version so they were more interested in seeing an NTSC TAS. I'd also like to point out this difference is not in SMB1 where both version are pretty much identical, so it's most likely an oversight that happened at some point when localizing SMB2 for PAL.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
eternaljwh wrote:
The comments on levels got a bit repetitive about how there's an unexpected strat [usually also not described] that saved unexpected time.
Sorry about that, I'm not very creative at writing about all of the duller floors. Most of the floors I commented like that are short floors that the whole SMB community has thought are clearly maxed out and the improvement truly caught us off guard.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
I apologize for not updating any info here. I've been TASing Monkey Ball games pretty much the whole time since the release of both TASes last spring. As CyclopsDragon mentioned, I've done Beginner and Advanced runs for both SMB1 and SMB2 NTSC. You may wonder why didn't I submit them to TASVideos? Well, at the time I felt like it would just be very spammy to submit 6 TASes in such a short time so I kind of forgot about it. All of them are already outdated: a 1 frame improvement was found for an extra stage that appears in both SMB1 categories, a frame was improved in the first level of Beginner in SMB2 and CyclopsDragon has pulled off a faster strat in Advanced Extra floor 1 Auto Doors. Though CyclopsDragon already linked a playlist (please check it out as well) with them, I'll just embed them, anyway. I hope you enjoy. Link to video Link to video Link to video Link to video Also, since we are remaking SMB1 Master TAS from the scratch, you may be wondering if the first one is bad quality or rushed or anything. I can guarantee you that for the time it was made, it was considered very high quality and every single floor was best known time for the game and we did work hard for it. I have simply gained more experience and patience so that's why we are managing to save so much frames the second time around. Huge shoutouts to CyclopsDragon for joining the project, he's a very creative TASer.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
RachelB wrote:
Does this happen in latest version? 4.0 is ancient.
I went ahead and tried with Dolphin 4.0-4250. It does happen, first frame I found it happen is input frame #1569 in SMB1 NTSC. I tested doing multiple separate recordings from scratch and it always happens at the same frame. I can also confirm that specific frame is not a glitch frame with 4.0.2, so the frames happen at different times on different emulator versions.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Anty-Lemon wrote:
Perhaps it's related to this? Also, honestly, two frames lost every once in a while isn't nearly as bad as you talk it up to be. :P You could easily go back and hex edit those two frames once the bug is fixed
It might be a similar issue, good to know it is known. Monkey Ball TASing is a highly optimized frame war, I have redone a whole TAS to save 3 frames. So, 2 frames is a big amount for me. The pausing also makes it annoying to TAS when I have to do the pause trick every time I redo the part, my TASing style includes redoing split second parts up to a thousand times, no joking. I will try hex editing afterwards if the glitch get fixed but Monkey Ball usually desyncs if you hex edit an earlier part so I don't have my hopes high up.
Post subject: Weird input glitch when recording
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
I TAS Super Monkey Ball 1 and 2 with dolphin 4.0.2 64-bit. I have found a really annoying and frustrating glitch while recording .dtm-files. Basically, this only happens when I'm recording inputs, the emulator forces specific input frames to do no input to the game at all. It seems somewhat consistent: it happens at the same frame on every separate .dtm-file for the same game. This is extremely frustrating in Monkey Ball, it either forces you to slow down for a frame (making optimizing impossible) or use my way "around" it: pausing the game and let the glitch frame happen in the pause menu. This costs 2 frames and looks nasty on the final TAS, but I (or the whole community) haven't found any other way around it. Of course, if I'm lucky enough, the glitch frame can happen during level transition, when it would not be an issue. It happens the same way to every Monkey Ball TASer, 32-bit or 64-bit. The glitch does not happen at all outside of recording inputs, so it's not an issue in the game. Consistent glitch frames I have found that happen every time: SMB1 NTSC: frame 5249 SMB2 NTSC: frame 13653 Example of two SMB2 TASes where I had to pause to get around the glitch, around 3:41 on both videos: https://www.youtube.com/watch?v=Th1MQDnxe3I https://www.youtube.com/watch?v=Fr_GmvSzlmE I just want to know if there is a way to get rid of this annoying glitch, the whole Monkey Ball community is just sad because of it.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
sack_bot wrote:
voting meh due to sub optimal character
Feel free to hex edit at offset 000017A8, if you input right you pick Meemee and no input leaves you with Aiai. :)
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Weatherton wrote:
yes vote! I still want to see a high score run for both of these though :) Plus warpless...
I understand it would be really amazing to see high score run, high score warpless, time warpless, max banana% warpless etc. But considering a project like this takes 2 months for a 4-5 TASer group, I don't currently have the extra time for that. :/
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
ALAKTORN wrote:
have you checked whether it’s an issue about lag frames? you should post the movie file for the improved levels, someone may be able to make it sync
Apparently only thing I have left from the improvement is a replay in the memory card. :/ However, I do have a movie file including both improvements for SMB1: http://puu.sh/7DvNK.dtm It's up to EX9 and I can't get that level to sync. If this can be synced, then I might have motivation to give a try to include the improvement in SMB2 too (would have to redo the improvement).
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Sorry for the confusion. I tried hexing the improved level, that caused me to having to redo nearly every stage after that. At the time every other TASer had already been done with the project, and I couldn't do it alone. Monkey Ball games don't have RNG in the engine but for some reason, entering a stage at a different frame usually gives different output with the exact same input. Sometimes some stages might sync.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
thatguy wrote:
How hex-friendly is this game, just out of interest?
Not very. If you hex edit a past stage, 30-40% of the stages after that will desync.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
stootab wrote:
Any reason for gongon?
We chose Gongon mainly to shock the Monkey Ball community. Other than that, there's no difference.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
The submission page says "DTM file starts from a savestate". It doesn't start from a savestate, I don't know why it says that.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Both of the TASes are near finishing, so we decided to make a little combined preview video. :) http://www.youtube.com/watch?v=iF1hw5UzKRA
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
We tested about the endings of these games. In SMB1 you have to press A to get past the congratulations scene before credits, but no input is required after that, it will finish credits and name entry on its own. In SMB2 without input you will get stuck in the name entry screen so it seems we have to finish the final cutscene, skip the credits and manually finish the name entry as the last input. Next thing I gotta figure out is the encoding when the TASes are finished, my audio dump desyncs with my frame dump and the AV-sync version just crashes when I start encoding...
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
In my opinion, the credits in these games are just long and boring, they wouldn't really be entertaining at all after the first 30 seconds, no matter how optimized they would be. Would't look much different than non-TAS play of those. And just to confirm: so it is required to finish the name entry screen in addition to credits no matter what?
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Glad to hear people like this stuff. The progress is going really well, SMB2 should be done within couple weeks and SMB1 within few more. :) I have a question in mind regarding the TASVideos timing: Is it required to complete the name entry after credits or is it fine to just stop the movie when getting to the credits?
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
ALAKTORN wrote:
that WIP looks great to me I assume you guys are aware of the WRs and are beating them all? from YT or Cyberscore
Of course, I've made sure that all the times in the TASes are the best times ever gotten in the games, also beating all the old TAS times in addition to beating just the human records. :)
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
WIP for SMB1 Expert TAS: http://www.youtube.com/watch?v=9ZUU6T66LbU SMB1 progress is on full blast and SMB2 has been stuck in E36 for few days now, I hope we can get that rolling again soon. :)
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
So, I found out that E6 desyncs between 32-bit and 64-bit dolphin. This shouldn't be a problem tho because 32-bit users can just use save states to get past desyncing parts. Oh and btw, I got 51.03 in E7, finally a time I'm happy with. :)
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
THC98 wrote:
Are you sure that the Dolphin revision you guys are using doesn't desync in SMB1? And if all the settings (graphics backend, audio output) are the same for all of you?
We have made sure the version and settings are right. As long as it syncs for some people I guess it's manageable. We can just transfer the save states to those who work on the TAS but don't sync.
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Now SMB1 seems to have some really weird desync issues. :/ Our dtm desynced on E6 for barhunga, and we were able to fix that by hex editing an extra empty input frame to E2 - E5 transition. However, that time he desynced on E5 couple times before getting through it randomly. Both, the original and the modified dtm file worked perfectly for me, though. The original dtm also desyncs for keylie on E6, different way than for barhunga. I hope this issue doesn't make the game ultimately unsubmittable in TASvideos. SMB2 project file (haven't tested SMB1 yet) in the other hand syncs for me and jcool perfectly (not for everyone tho), so I don't think the problem is about my dolphin/computer. I'll throw the SMB1 dtm here if someone else wants to check if it syncs or not: http://puu.sh/6CZji.dtm EDIT: The SMB1 project did sync perfectly for Jcool, so there is some hope it will sync for someone else too. :)
Experienced Forum User, Published Author, Active player (386)
Joined: 10/17/2012
Posts: 39
Location: Finland
Finally I was able to hex keylie's E2 26.45 into our project file. :) It did actually desync in the end of the level making it a frame slower, I had to manually redo the ending. The biggest surprise for me was that E6 didn't desync even if it was entered at one frame earlier. That's the first time that level didn't desync for me when past levels are improved. o_o And yeah, I got the correct iso anyway, it's exactly the same but at least I don't need to get warning messages about wrong checksums anymore.