Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Yet another timesave from the RTA community, which I'll put here for reference. Robster can clip into the ground and then fail to do the next jump. This clip is on the rematch and I'm not sure how it works. Here's the clip: https://clips.twitch.tv/HardStupidMageCoolStoryBro
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Paused wrote:
Obvious yes vote.
Ever look at non-special stage abuse routes for the levels you grabbed an emerald in out of curiosity?
Thanks! Some levels had a big enough difference that I just needed a rough TAS for timing. But I had to fully TAS both routes for some levels like Turquoise Hill 1 and Gigalopolis 2 since the times between the routes were so close. In the levels with special stages, I didn't find any new no special stage routes over the known speedrunning routes.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
I attempted to find the original page that I looked at for the FM sound information but had no luck. However the TAS syncs both with and without FM sound since FM sound wasn't supported with this game, so I'm happy to turn it off if that's more legitimate.
I found this footage of the BIOS: https://www.youtube.com/watch?v=VpVOgXqx9Vo
It seems to be closer to the Japanese BIOS. But when I try to boot Wonder Boy in Monster World with the Japanese region setting it just skips the BIOS and goes straight to the games inbuilt sega screen. Essentially I was using the US bios because the PAL game works fine on an official US Master System and plays at 60fps. Then the main reason for mentioning the Korean release was to justify playing at 60fps.
If the Japanese region (no BIOS) would be considered better then I'll edit the TAS to use it
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Sorry I think I've had some issues with my button files. I'm going to cancel this for now and sort it out over the next few days. Sorry for the inconvenience
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
This is really embarassing but I accidentally uploaded an old button file. This one is a few frames faster http://tasvideos.org/userfiles/info/44116687937909380
Also, the desert pause exploit doesn't work on Genesis as far as I can tell
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Wow, awesome work. One of my favourites was falcon BtP with the taunts, also doing fox BtP basically backwards was really cool to see. Link was a massive highlight too and the bumper boosts with purin were great. The submission text is very informative without dragging on and I like that you completely finished and optimized a US run. However, the most impressive thing in my opinion was the amazing level of research and optimization that went into this. Fantastic work guys!
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Nice work, the new finds are really cool and the Cortex fight was great. The publisher should set the speed/entertainment tradeoff flag since SFX was kept on for this.
I'm a bit confused about the choice of using J so I decided to have a look into it. You seem to get one less blue warp cutscene on J, this is for the first level entry. But also, J loses time on the title screen. I've heard of some physics differences with jump height between the two but as far as I can tell, the physics are the same, unless I'm missing something. So it seems like SFX on is faster on J (one less blue warp cancels time lost from title screen) and SFX off is faster on U. If you have more info regarding your choice I'd be interested to hear it
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
I’m personally on the side of obsoleting the NTSC U TAS with this, BUT I think the correct decision should be to have separate publications. Here’s why
To the people who want obseletion:
I agree with you, but it’s important to remember that even if obseletion was decided by a judge, Happylee could still cancel the submission and we’d have no PAL TAS at all. I mean, if you put that much time into the NTSC U TAS and specifically said you didn’t want it obsoleted it would be conceivable that you might pull out the PAL submission. I’m not sure if Happylee would do this or not, but either way we should respect the opinions of the author of both TASes.
To the people who want rejection of PAL:
It makes sense that some people might notice the pitch of the music and other differences, but if it’s explained well enough in the publication it should be fine. Also, only about half of people who played this game had the US version, there are still 8 million people who had the PAL version and find it valid.
Anyway, if this run was rejected it would reflect terribly on TASvideos. I remember finding a PAL TAS of SMB on youtube some time ago which was faster than the TASvideos publication. To the average viewer this looks really bad and makes TASvideos look like it has unoptimized TASes and it looks like TASvideos isn’t a very reputable TAS site. So the fastest time of SMB should definitely be published in some way.
About the separate publication:
This is probably the way to keep most people happy. Most people will just click the fastest time anyway, which is good for those who want PAL but if they think something’s a bit off, they will probably realize that it’s because they watched the PAL run (as long as the publication is very descriptive about PAL vs NTSC)
Really from the start, the SMB TAS should have been done on PAL. Many communities play the faster version even if it’s not on NTSC U. In my opinion, the “TASvideos rule” for NTSC U is useful for people who can’t figure out which version is fastest, so they’re just safe to TAS on NTSC U. The rules should be taken as a guideline to make sure we don’t publish crap, but each game should be treated separately. The main entertainment for SMB comes from the extreme optimization and we shouldn’t pass up that opportunity just because people had different versions.
But it doesn’t make sense to get rid of years of progress on NTSC U and most importantly, since the spread of opinions is fairly even, we should just go with what the author wants.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
I've been wanting to see a well optimised PAL run for a long time and I wasn't disappointed. Yes vote. Nice new tricks and cool glitch at the end of 8-3 with the flag. Good playaround too.
For the people that aren't familiar with the PAL, NTSC terminology, if both versions get a published TAS it might be a good idea if they were labelled as US version and European version.
Also it would be cool to see a sentence or 2 in the publication description about how PAL usually loses a lot of time over NTSC. For example Sonic, Metroid and nearly every SMS game take 20% longer on PAL. But some games like Mario adjust the physics to compensate, which in this case causes more glitches. This would hopefully avoid confusion of people thinking that PAL is faster for every game, where in reality its usually at a huge disadvantage for most games.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Nice work on this and great explanations to accompany it. Highlights for me were the boss with the music syncing, the boss with the visual glitches. I liked some of those clever skips like the one with using the enemy to gain height.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Alright that sounds good. All the RNG seems to be optimal. Thanks again
Only thing that's bothering me a little is the playaround got broken at the end of toy factory boss, but it's not really an issue since it wouldn't lose any time. Unfortunately I didn't spot this until after 10k rerecords resyncing the crab.
So that resync should be fine
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Anyway, here's the resync http://tasvideos.org/userfiles/info/40105171216213950
If possible, I'd really like the original v1.12.2 button file to still be available. The TAS was optimized for that version and it's possible that the resync has sub optimal lag reduction since my approach was just to push back the inputs for lag frames rather than go back and try to reduce eliminate it with different movement, especially in the last level. Any non-lag related things should be optimal though.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Been following these WIPs for a while and it's great to see this submission. Definitely a yes vote from me. It seems fairly well optimized, I like how you almost outrun the screen in some parts and I like some of the routing like jumping over the loop in the first level. Ending input early is cool, but it makes me wonder if any earlier is possible.
I'd have to agree with zoboner about the viewport issue, this seems to be an issue with a lot of GG games. While I would have preferred a TAS of the SMS version I can understand why the GG has it's appeal. Since GG was more popular than SMS in North America and it was the GG version included in sonic mega collection plus, more people will be familiar with the GG version which is probably for the best. I got used to the viewport eventually.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
@ Alyosha: Thanks for starting this resync off. You shouldn't have to resync the whole thing though, so I don't mind taking over from here. Currently synced up to the crab boss.
Just curious to make sure, which levels exactly did it lose time to lag? Also did you have to manipulate the tornados in level 2?
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
https://youtu.be/jAo5LARuYfM
Found this while preparing for this TAS, lots of info in the description. Pretty sure 1.0 is the way to go now. Basically it's a massive zip off the second platform in Labyrinth 1 for the GG version, should save 10-20 seconds in the TAS I think. I can probably get higher with it, haven't investigated it fully.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
mike89 wrote:
So a YouTube user by the name of Trapstar uploaded a bunch of old videos from TSC, including... SprintGod's original Sonic 1 GG videos! The two relevant videos for the strategies we never fully ascertained are:
Jungle 1Labyrinth 1
According to Sprint earlier in this topic the Labyrinth 1 strat is only possible on v1.1, which disables the vertical zipping glitch that makes Labyrinth 2 faster. So which one saves more time should determine the version used.
Does anyone know what these strats are? The main one I'm insterested in is the Labyrinth one so that I can decide whether to use 1.0 or 1.1. If anyone knows what the jungle one is that would be helpful too but I'm assuming it's in the published TAS anyway, unlike the Labyrinth 1 strat since the TAS used 1.0.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
This is really cool, I didn't realise that this was a big priority. It'll be interesting to see if the pause button can be used in any of these. Honestly about half of this list is on my list of possible future TASes so hopefully I can help to deal with this over time for any that are left over.
The one specific one that I could fully commit to is Sonic 1 GG since I've had experience with the SMS TAS. Also I'd love to see if Jungle boss is skippable and if there are any more underflows on the GG version. It'd be good to update it with the more current goal real time minus bonus screens rather than just game time. So I'm happy to do Sonic 1 GG unless anyone else desperately wants to.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Redhotbr wrote:
Honestly playing a slower variant of any% is very arbitrary, as it's slower than the fastest any%.
I understand that it seems arbitrary and I'm not usually a fan of middle categories myself. I think the best way to look at it is all levels and bosses (not including secret levels) rather than a set of glitch restrictions. As for zip-zagging use it's hard to choose what to do. Some people love it, some people hate it. Both choices have validity in my opinion.
As for the optimization, most points in that message can be addressed by reading the submission text (specifically the reason why I used damage abuse in midnight run).
The video does manage to get up a hill while under the wall that I couldn't manage to get up (not through lack of trying). But this in fact cost me no time in the TAS. If I did go back and save that small amount of time, I would lose it on the deep trouble global cycles. You'll notice that I would have to wait at the first wheel if I got there any sooner. There is no way around this since the level order has to be very specific for mask requirements in N. Tropy.
It's important to remember that comparing level times visually can be misleading, there are 2 main reasons for this.
First, the lag can vary significantly in this game (which is why it's not resync friendly). I know this from redoing levels in different orders for global cycles. Sometimes, just doing the levels in a different order with the exact same optimization and lag reduction methods can make the final times differ (by 1-2 seconds in platforming levels and up to 1 second in non-platforming in extreme cases). While the game does sometimes adjust movement speed for lag, it's not perfect. So the only way to properly tell is to watch the internal global timer rather than the frame count. Of course, I properly reduced lag where possible but these differences are still present regardless. While the video posted was apparently faster, it is unknown how much of this time was due to lag differences (it could even be up to a whole second).
Secondly and specifically to pura levels, you do not have a certain amount of time to stay under the wall, it is more like a certain number of jumps. You'll notice that when comparing my TAS to the alternate video, my jumps are much more frequent. This in general is more optimal since I can get all of the jumps that I can while under there and get under again faster rather than wasting my time spacing them out under the wall. I would definitely agree that I could spend more time under the wall but in general it would be slower due to less jumps. This would definitely make the movement seem visually off even if it is in fact optimized.
Overall, I'll accept that I could have climbed up one or two slopes while under the wall and didn't quite manage to figure it out unfortunately. But I hope these couple of minor oversights don't reflect on the overall optimization of the run since there were only one or two of these in the run (just in pura levels), they lose no time over a full game run due to global cycles and comparing levels by frame count or by visually watching is an inaccurate comparison to make, especially in the context of a whole run.
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
Awesome stuff keep up the great work! Was cool seeing some of the loops jumped over and stuff. The boss looks pretty technical too.
I just want to ask about how building up speed works? Sometimes when falling or jumping over/under a wall you press right up against it (for example at 0:52, 0:54, 1:02). This seems slower since you have to accelerate up from 0 speed. Sometimes you do pass the wall at full speed (e.g. 1:04). I hope that makes sense, I guess I'm trying to say it's like what's in this link but sideways instead http://tasvideos.org/GameResources/CommonTricks.html#JumpingOffLedges
Of course this doesn't apply to every game, I wanted to mention it just in case
Experienced Forum User, Published Author, Expert player
(2525)
Joined: 11/26/2015
Posts: 183
Location: Australia
I decided to do a testrun/LOTAD of Crash 1 with zip-zagging. I think that it would be extremely hard to optimize since the screen scrolling speed is essentially unbounded unlike Crash 3. You’d have to TAS each level roughly 80 times with different amount of speed to find out the optimal amount of speed to build up. It is impossible to predict in advance due to different speeds causing different lag. I don’t think an optimized TAS of this game with zip-zagging will ever be possible.
I wanted to leave this here as a reference in case someone wants to try messing around with zip-zagging in this game but I note that the non zip-zagging sections are especially unoptimized. Even the zip-zagging levels should be easily beatable.
Encode: https://www.youtube.com/watch?v=YelWu0mHpjs (There’s a poorly commentated version linked in the description)
Button File: http://tasvideos.org/userfiles/info/39782140193476681
Played on Bizhawk 1.11.9
