United Statesman i remember playing trying to play flash games on the library computers at school to pass the time until the librarians came around and made us stop because it was "eating up to much bandwidth". You kids don't know how good you have it now a days...
What emulator? When you say you can't replicate it, did you run the same exact tas inputs and you don't get the same results or did you try making a new tas and didn't get the same result?
yeah i think "stopwatch" time is always truncated, not rounded. But it seems a bit late to change policy now. The game only really needs to be timed down to the hundredth of a second anyhow....
oh i see. I didn't do much work looking at different setups to give you the right subpixel at the flagpole. I was more working on finding frame rules and setups in 8-2 that will yield an optimal BBG shot.
My strat at the stairs is:
- Do a backwards jump onto the stairs
- Before you land, hold down+right to keep you from turning around
- Wait for bullet to be in the middle of the column
- While still holding down+right do a full jump to the column
- Right before landing release down
- Jump after ~1-2 frames on the column
forgot to mention... I'd recommend practicing with the sock folder rom to learn BBG. Sub pixel values ...080 to ...270 yield BBG. Using that rom you can tell whether your jumping too early or too late.
No going over the pipe faster doesn't guarantee that you get a faster shot. Different speeds over the pipe will yeild different bullet shot times but faster pipe doesn't equal earlier shot.You just seem to be practicing on a bullet shot that's somewhere between "optimal" and "unoptimal". I made a thread on this a while ago.
Do you have the rom loaded before you load the lua? I think thats what you get if you try and do it the other way around.
No you need to get to 8-2 on the correct framerule and the 8.17 offset accounts for the time saved by doing the two FPGs. If you don't want to do any FPGs then you would need to start two framerules earlier to account for the two framerules lost in your run by not attempting them.
mmm.... I don't use bizhawk. Nes>Advanced Settings>Region Override?
Never mind, you said it was running at 60 fps. Yeah i have no idea dude. See if changing the core works: NES>Core> NesHawk
Otherwise Try downloading FCEUX http://www.fceux.com/web/download.html
I don't have anything better to do with myself so I decided to look at the BBG setups a little closer.
From what I've seen, the two things that determine what kind of bullet bill shot you get is dependent on the framerule you get to 8-2, the enemies you kill in 8-2, and how long you slow down to get over the pipe in 8-2. So Looking at the original setup, I recorded what kind of bullet bill shot you got based on how long you slow down for the pipe. To do this, I did a bunch of attempts and looked at what frame Mario's back hand passed over the last bullet bill launcher. Mario moves more than one pixel per frame at full speed so, at least from what I can tell from my recording, there are three pixels in the x-direction that he could be in the first frame his hand clears the last bullet bill launcher. This seems to be a good predictor on what kind of shot you get but there still seems to be a bit of randomness due to his exact position and probably some subframe shenanigans.
So starting from RTA start and without any FPG or fast 4-2, Mario typically passes over the last bullet launcher in one of these seven frames:
Time% of attempts 3:22.216_4% 3:22.233_15% 3:22.250_15% 3:22.266_26% 3:22.283_21% 3:22.300_14% 3:22.316_5%
After ~200 attempts, the percentage that I hit that particular frame is shown above.
This is what I found for the original setup (waiting 1 framerule on start with no FPG or fast 4-2):
Framerule 529 on Pellison ROM TimeShotBB Shot Time 3:22.216_Un-optimal_3:22.882 3:22.233_Un-optimal_3:22.749 or 3:22.882 3:22.250_Optimal_3:22.499 3:22.266_Optimal/very rare Un-optimal_3:22.499 or 3:22.766(rare) 3:22.283_Un-optimal_3:22.766 3:22.300_Un-optimal or rare no shot_3:22.766 or no shot 3:22.316_Un-optimal_3:22.632 or 3:22.882
So for this framerule you only have two frames where it's possible to get optimal shot (bullet shoots at 3:22.499 always). The rest give you an un-optimal shot that shoots anywhere between 3:22.749 to 3:22.882. For my attempts, I got an optimal shot ~40% of the time. Its commonly reported that its 50/50 which I assume is probably true since I oddly didn't get 3:22.250 enough to make percentage distribution look even.
In the past week or so someone found another framerule that gives you an optimal shot consistently. To do this, you wait 11 framrules on start and do the same 8-2 setup as the original framerule (stomp no enemies until the two koopas right after the pipe jump).
This is what I got for the new framerule: Framerule 539 on Pellison ROM Time_ShotShot Time 3:22.216Optimal3:22.433 3:22.233Optimal3:22.433 or 3:22.483 or 3:22.549 3:22.250Optimal3:22.549 3:22.266Optimal3:22.549 3:22.283Optimal3:22.483 3:22.300Optimal or No Shot_3:22.483 or no shot 3:22.316Optimal or No Shot_3:22.483 or no shot
So this framerule is a lot better than the original. The only thing that can trip you up is that if you're slow then there's a chance you get no shot or if you get one of the 3:22.549 shots and are very late bouncing off the bullet bill to end the level. I'm not sure what the cutoff time is between saving one and two framerules but I've only saved one framerule on some of those 3:22.549 shots a couple times.
So I believe that I found a BBG setup that gives you an optimal shot consistently one framerule after the new setup above. This would mean waiting 12 framerules on start. Having BBG setups on two adjacent framerules is useful because it allows you to lose a framerule on your way to 8-2 and still gives you the opportunity to do BBG. The video below is how to do the setup. You just need to stomp two additional koopas in the same manner shown in the video.
These are my results when testing this framerule: Framerule 540 on Pellison ROM Time_ShotShot Time 3:22.216Optimal3:22.483 3:22.233Optimal3:22.483 3:22.250Optimal3:22.483 3:22.266Optimal3:22.483 3:22.283Optimal or No Shot_3:22.483 or no shot 3:22.300Optimal3:22.483 3:22.316Optimal3:22.483
So could somebody tryout this setup to verify that it works for them too or if any of this makes any sense to anyone?
edit: So discussing with andrewg this morning, this would be the WR strategy:
- Wait -8.17 on the title screen
- Do the 1-1 FPG and the 4-1 FPG
- attempt the new fast 4-2. 4a) If you get fast 4-2, then you line up with 8-2 BBG framerule that mav/kosmic found 4b) if you get slow 4-2, then you'll be one framerule behind and line up with the BBG framerule that I found
I'm also trying to find another BBG setup on another adjacent framerule. Having 3 in a row will allow you to do the above strategy plus if you don't get fast 4-2 and get bad judges you can still do a BBG and make up a framerule by doing 8-3 FPG.
Yeah it's difficult. You want to keep your speed up as much as possible when you go up for the grab and make sure you're still facing backwards. If you hold left too long in order to brush up against the block, you won't get p speed when you exit the power up animation.
It's because your using the PAL rom. The fceux timer counts the frames then converts it to seconds using the ntsc frame rate. Not sure why you want to use the PAL rom. All runs submitted here need to use the ntsc rom.
My thoughts: I don't think the point of streaming was to prevent anyone from submitting a spiced run, TAS, etc. but was a way to verify that someone had put in the effort and has the skill to get a good time. When you stream attempts you have actual proof that you know how to play the game due to the hundreds or thousands of attempts it takes being shown publicly. It's more of looking at the bulk of someone's work as a whole vs. just meticulously inspecting one run to prove that someone isn't cheating. It's very apparent if someone streamed a bunch of horrible attempts, can't do a single pipe jump or whatever, and at the end played a 4:58 TAS. No one who's going to submit a TASed run is going to also make and stream thousands of other TASs of failed sub 5 attempts.
To offer a compromise and I think someone has brought this up before: Instead of streaming attempts we could have the option of just recording our attempts and uploading them to youtube for verification. There could me a minimum amount of attempts required (like 2 hours worth or something idk...). In the end there's little difference I can think of between streaming all your attempts or just recording them in terms of proving that you know how to play the game. Streaming is just live. In fact there's nothing that prevents someone from recording all their attempts and then streaming them back at a later time with how the current rules are. No one could tell the difference.
I don't think it's easy as people think for the mods to spot a standalone spliced run that's submitted. Particularly if it's done on emulator save states. If it's done properly it should be indistinguishable from a legitimate run.
Whatever the rule is, just make it objective. It's a slippery slope if its handled on a case by case basis where the mods would have to decide on their own if someone is trustworthy or not. I feel like it would put people who aren't vocal in the "community" at a disadvantage.
