Posts for zidanax


1 2
14 15
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
I didn't notice before that you can still kind of tell where the ground is in Level 8: when the will o' the wisp goes offscreen, the ground changes to pitch black, while the corridors still have a cave-like background over them. Still, a map would probably help. @JXQ:I definitely don't want to disable BG 1, because it has that cave-like background. Disabling BG 2 doesn't seem to do anything noticeable. Disabling BG 3 just gets rid of the piece of parchment or whatever it is at the top of the screen. No noticeable effect when BG 4 is disabled. Just in case, I tried disabling sprites, doesn't help. Disabling transparency doesn't help, nor does disabling graphics clipping. @stickyman05: I messed around some with your suggestion, and I got the same result: Peter Pan will only move forward if hit while facing backwards. I suppose sliding backwards might work too, but I'm not sure why I would want to do that here. However, I've managed to get Peter Pan hit facing backwards only 2 frames slower than if he was running forward, so I was able to save some frames going up to the next pit. I'll play around with it some more and see if I can squeeze any more frames out. Thanks for the suggestion, stickyman05! At some point, I'll put up a 2nd version of the level 6 WIP with the new strategy. I might also play around some more with the boss fight and see if I can find any way to save more time. This is because the second time the balloons came out in my run, they came out differently than in hero of the day's run. Perhaps I could do some sort of luck manipulation of how the balloons come out to save even more time? Anyways, because of this further work on level 6 (and because spring term just started here), I probably won't release a level 7 WIP today.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
@hero of the day: OK, thanks. Yeah, I was going to plan the route before really TASing the level. But I think I'll attempt to composite a map of the level. @stickyman05: I'll mess around with your idea and see if I could get something like that to work that saves time.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
@hero of the day: tomorrow, I'll probably do level 7, which should be a cakewalk, since it's just an autoscrolling level. But then I have to do level 8, which I'm definitely dreading trying to TAS. Since I'll be going faster than the will o' the wisp, I'll essentially be TASing blind. I could investigate the level by waiting for the will o' the wisp and going through the level. Maybe draw a rough map of the level too while I'm at it. But this is a TAS, so I want to be precise, and I don't trust my mind's eye or a hand-drawn map to be sufficiently precise. I suppose I could go through the stage and make screenshots of the level sections when they're lit, making sure to disable sprites and BG#3 (which contains the "parchment" piece) when I take screenshots. Then I could try to composite the images into a map of the level. Did you try something like this? Do you have any other suggestions for how to go about TASing this level without losing my sanity?
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
OK, a WIP of level 6. -Not as complicated to TAS as level 5, so it's mostly small optimizations here. -Occasionally saved frames due to avoiding going up steep slopes. Note that if the slope isn't quite as steep, you won't lose speed. -Saved 18 frames overall during the boss fight. The first hit was done 18 frames faster, and the 2nd hit 8 frames faster. However, the balloons then came out such that I couldn't get in the third hit until 8 frames later than in hero of the day's run, thus neutralizing the time gained with the 2nd hit. -Some route changes resulting in small time savings. -Small time savings going down the switchbacks, or whatever you call them. -And as usual, small savings due to minimizing ground attacks and jumps. Overall, 63 frames saved from level 5 fadeout to level 6 fadeout. EDIT: Whoops, the WIP is here
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Original, and it's kind of cool that you turn something as mundane as a controller test into a music maker!
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
OK, here is a WIP of level 5. Some time savers: -Instead of only gathering enough pixie dust to fly up the first pit, I gather enough pixie dust to fly up to just before the 2nd icicle. I also manage to fly under the first icicle. -I managed to jump past the 2nd icicle into the water with somewhat greater precision. -Instead of killing the 2nd fish, I go around it. -Most interesting time saver from this level, even though it doesn't save the most time: I found a new trick that lets me get past two of the icicles in the latter part of this stage faster. I was thinking of how when you crouch, you're lower to the ground, so it takes longer for an icicle to hit you. I messed around with crouches a bit, and found that if Peter Pan is running, and you switch to pressing down-right and Y instead of right and Y, Peter Pan will go into a crouch, but he will "slide" a bit while in that crouching position. There is a 2-3 frame window where if you do that "crouch slide" while running under a falling icicle, you will not get hit by the icicle. Your forward movement is slightly slower during the execution of this trick, but it's much faster than waiting for an icicle and jumping over it. BTW, I tried this trick with the icicle over the water, but I can't get it to work, probably because I'm closer to the icicle on the Y axis in that case. Then there's another icicle where it starts falling right when you fall off the ledge leading up to it, so that by the time you've hit the ground, so has the icicle. Besides, I need to be left of the icicle to get into a run early enough for crouch-sliding under the next icicle. -I optimized the gathering of pixie dust at the second collection point somewhat, taking advantage of the fact that when you first switch out of the state of flying, Peter Pan falls very slowly. And when Peter Pan is in a state of free fall, even if it's very slow, he gathers pixie dust faster than if he's hovering. In total, an improvement of 294 frames from level 4 fadeout to level 5 fadeout. The WIP is here EDIT: I decided not to get hit in this stage, because there's no autoscrolling section or boss. In other words, there's no "downtime". Getting hit twice would save me about 30 frames in the score tally screen, but then I lose time when I get hit. Each time I get hit, there's about 30 frames where I have no control of my character, so it's better just to not get hit.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
OK, so it's a novelty run, and the idea isn't original, but IMHO, the run is entertaining to watch.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
hero of the day wrote:
I had a feeling that the flying/running part in the beginning was improvable, nice to see how much can actually be saved. I don't understand how 74 frames were saved at the end of the level though. Almost 10 seconds ahead of my run now, that is awesome.
*compares the runs again* For a little while just after the screen starts scrolling to the right, you appear to have let go of the forward button, slowing you down somewhat, then you have to accelerate back up again. Still, I don't think that would have made that big of a difference. *Looks further*... A ROM difference, maybe? I say that because for some reason, my run, measuring from when the boss dies, starts scrolling exactly 70 frames earlier than yours. I suppose it still counts, since we know the ROM I'm using actually appeared on cartridges (e.g., mine:) ), but we don't know about the other ROM (probably a beta).
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Fun to see those complicated puzzles get solved so quickly. Really catchy music, too, that helps to keep things entertaining.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Level 4 has been completed Some timesavers: -The start trick was executed 1 frame sooner. -I take the time to collect as much pixie dust as I can at the first collection point, then fly all the way to the second collection point. Flying is slightly faster than running, and given the distance between the two collection points, this more than makes up for the time spent collecting pixie dust. -A new trick I stumbled upon: normally, Peter Pan can't make a running jump off of the platforms in the first part of this level. But for some reason, if you land at the left of a platform from a jump, you can barely manage a running jump. This is used after the second pixie dust collection point. -The 3rd hit on the boss was done 7 frames sooner. -After the boss dies, I got Peter Pan into his end-of-level pose 74 frames sooner. -29 frames saved from the tally trick. In total, 252 frames saved from Level 3 fadeout to Level 4 fadeout! The WIP is here
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
...Unbelievable precision, insane glitch abuse, and an amazing eye for entertainment. I strongly agree with the other people who think this run deserves a star. Way to go, Scumtron!
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
OK, here is an attempt at level 3. This WIP saves 109 frames from Level 2 fadeout to Level 3 fadeout. -A new strategy in the beginning part of the stage so that I don't get hit and even save some time. This way I can take advantage of the two later time-savers involving hits and not die even though I don't have the golden sword. -Other small changes; I try to play with greater precision. This level was interesting, since it travels vertically instead of horizontally, which makes it harder to optimize. One thing is that it's somewhat harder to find good start and end points for dividing the level into "segments" to use to compare times. Do you guys see anything that looks "off"? The WIP is here
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
OK, here's my WIP up to the fadeout of the second level. Improvements: -Managed to skip the second story sequence one frame faster. -Started running one frame faster in the second level (The point I used to measure this was the first frame the character moves two units forward instead of one) -This level in some ways didn't seem to have as much to optimize as level 1: no flying sequences, no trying to go over a lake or up several ledges. There was one thing, however -- since this level has a multitude of places where you have to jump, various ways to move through the level thanks to the branches, and of course, enemies to stand in your way as you leap through the branches, I was able to save a decent amount of time as a result of attempting to find a route that minimized the number of jumps and ground attacks, which cause small delays in forward movement. Much messing around with that... I'm certainly much more familiar with this level now. I was able to save 22 frames in the main portion of the level measuring up to the semiarbitrarily chosen X position indicated by 7E05AF=2512 (appears to be for a scrolling layer, but good enough...) through this kind of optimization, and I was able to fit in some air attacks on enemies... if done right, those don't cause any delay. -The last section of the level was interesting to try to optimize. Measuring from after you jump over the pit between two branches, it seems like getting to the boss takes a bare minimum of three jumps. I was able to find a route through that section with only those three required jumps with no attacks whatsoever, yet this didn't end up giving me the best time I could manage when experimenting with various ways to go through this section. My times seemed strangely incosistent with what I know of how the character's movement works. My guess is that it has to do with lag. At one point in this section, there are four enemies on screen, assuming you don't kill them. My best attempt had a one frame delay apart from the neccessary jump delays, due to a fireball hitting a pirate. I'm guessing the reason that attempt worked best anyways is because it kept the number of enemies onscreen down. If it was indeed lag, I suppose I'll have to watch out for lag in the future when I notice lots of onscreen enemies. BTW, I didn't notice strange behavior like this in stage 1, probably because there is, in my run, just one brief point there with three alive enemies onscreen. -Exact same number of frames per hit on this boss as in hero of the day's run. However, I took advantage of the acorns to minimize my life to use the score tallying trick. Yes, that means I lose the Golden Sword, although there are places to get the sword back in later stages. Hopefully, losing the Golden Sword I had in level 2 won't make me lose frames later, but if it appears later that I should have kept the Golden Sword I had here, I'll go back to this fight. Anyways, I also managed to, on the last hit, position myself so that I was next to the southeast hole, and yet not get hit by the boss emerging, or by falling acorns. So I was able to kill the boss and also be very close to the bottom-right edge of the screen. I had Peter Pan already running once the screen starts to scroll right. I saved 11 frames in getting to the point where Peter Pan gets into his end-of-the-level pose. Quite frankly, I'm not sure exactly why, because while hero of the day did exit the boss section in a different way, it looks like he was running at the bottom right of the screen when the screen started scrolling to the right. -32 frames saved in the score tallying screen, if I'm counting correctly. -In total, 84 frames saved measuring from level 1 fadeout to level 2 fadeout. Level 2 WIP
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
ZeXr0 wrote:
I just watch it, very nice, just one little thing, in the middle of the first level you seem to bump in the wall and it would be faster to jump earlier to same maybe a couple of frame
I could have started a running jump 2 frames earlier, but that doesn't really matter. Thank you very much anyways, because I experimented with releasing the B button while ascending to the first ledge, and managed to save 15 more frames (!). I also experimented with releasing the B button early during the climb to the last ledge, but it didn't make any difference, and in fact would cause a 1 frame delay compared with my last attempt if I didn't do it exactly right. I also managed to save 3 frames during the flying segment, but unfortunately, the boss didn't go into his vulnerable position until 2 frames later, which was what happened in hero of the day's run. So I lost 2 frames on the first hit of the boss. Oh well, 1 frame saved in total going from the beginning of the flying segment to the boss, and the flying section looks, IMO, a little nicer. However, I was able to save some more time with another trick that I should be able to apply in some other stages. I was thinking how in some games, the score takes more time to "tally" if the character has, say, more time left on the clock at the end of the stage. After many rerecords (and much cursing at the boss), I was able to orchestrate things so that the boss reduced my life to the minimum, and yet I still did the last two hits in the same number of frames as last time. As a result, I spend 31 less frames(!) tallying the score before the screen starts to fade out. Now my time to the frame where level 1 starts to fade out is 4693 frames, 122 frames over hero of the day's run. The WIP is here EDIT: Since the ending music of stage 1 doesn't seem to get cut off by tallying the score quicker, I was curious as to whether or not the next sequence would actually start quicker as a result of the trick. I made copies of both WIPs of my run and compared how long they took to start fading into the next screen when no input is given. It looks like my newer WIP does start the next sequence exactly 47 frames sooner (the total amount my newer WIP improved over the old one), so my trick with tallying the score quicker does appear to work.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
I've started on a TAS, using Hook (2648), and have finished level 1 so far. I didn't think I would find much to optimize at all in level 1, but I saved more frames than I expected. Measuring from reset to when the level 1 scoring screen (or whatever you call it) starts to fade out, I've saved 75 frames over hero of the day's TAS. I don't think any of the savings so far are due to ROM differences, because my TAS so far doesn't desync with Hook (29252). Some things that save time: 1. It turns out that attacks done on the ground while running result in 1 frame of no movement on the X-axis during the attack. A jump started from a run results in 2 frames total of no movement on the X-axis during the jump, same for letting Peter Pan run off a ledge. All of these things will, other than those delays, move Peter Pan forward at the same rate as running. I used this knowledge to try to shave off a few frames. 2. I was able to advance through the stuff before level 1 a grand total of two frames faster :). 3. I managed to execute the pause trick at the beginning of level 1 a few frames earlier. 4. Trying to go up the ledges after the lake while running, resulting in going across the ledges faster, and in the process managing to skip one of the ledges. 5. A different strategy in the section where you fly. Since Peter Pan gathers pixie dust slightly faster if he isn't already floating, I just had Peter Pan jumping while the pixie dust was collected. I tried to do it so that once Peter Pan had gathered enough pixie dust, he would be positioned at least roughly at the upper-right corner of the area where the game will give Peter Pan pixie dust. I gathered just enough pixie dust to let Peter Pan land on the platform after the one hero of the day landed on. This is because flying is even faster than running, and it turns out that if you land very close to the left edge of that next platform, you still have enough space to get a running jump down most of the multiple ledges that come before the boss. 6. I managed to to hit the boss 2 frames sooner as he comes onscreen. @hero of the day: Do you see anything that might be improvable?
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Hello, I'm going to take look at Hook and see if I can manage to find any ways to improve on hero of the day's TAS. One thing though: I'm wondering if I can change the ROM from Hook (U) (29252) to Hook (U) (2648) if I manage to make something worthy of a submission. Rationale: A while ago, Overload picked up a cart of the US version of Hook off eBay. When he dumped it, the ROM image was Hook (U) (2648), rather than Hook (U) (29252), which was the version of Hook in NSRT at that time. Correct me if I'm wrong, Nach, but I believe Nach thought that that was strange, because Hook (U) (2648) looked very similar to the European version. I got a cart of the US version of Hook and dumped it, and the resulting ROM image was also Hook (U) (2648). Just to be sure that there wasn't something funky with the European version, I got a copy of that version too, and dumped it. The resulting ROM image matched up with what was in NSRT. So Hook (U) (2648) is now considered the correct US version in both NSRT and No-Intro. Hook (U) (29252) is now labeled as a beta version in both NSRT and No-Intro, not that I'm sure how certain we are of that. The point of all this is that Hook (U) (2648) is more likely to be the "correct" US version.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Another WIP of the first level, now 11223 frames, an improvement of 65 frames. I didn't seem to be seeing improvements with getting the sandals first overall Maybe I'm doing something wrong, but the time gained going to and back from Horo didn't seem to be quite enough to make up for the time spent gathering coins. Any improvement in overall time by getting the sandals first would probably need really kick-ass luck manipulation of how the enemies appear in order to gather coins quickly. EDIT: oh, hello Gocha, I see that you decided to try the US version. It's finals week here anyway, so I'm going to stop TASing for now. I'd love to see if you can improve on my level 1 time for the US version. EDIT #2: I just remembered that you said you would do a 2-player run if it saves time. You could probably do a faster-enough boss fight to beat my time.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
@Gocha: A well-optimized 2-player boss fight would almost certainly be faster than a well-optimized 1-player boss fight, due to many more opportunities to do four damage at a time by hitting an enemy twice in the same frame(would there even be any such oppurtunities in a 1-player run outside of Boss 1?). Also, I know of at least one time-saving trick that's only possible with 2 players (the double jump). There may be other 2-player-exclusive tricks that I'm forgetting. You can, as you probably already know, look at DarkKobold/Yeoj's runs for some examples of 2-player time-saving tricks. I'm not going to attempt a two-player run, since this is only my first TAS
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
In that case, I will continue working on my run of the US version, since it sounds like Gocha prefers making a TAS of the Japanese version. First, I'll try to optimize level 1 some more. Seeing Gocha's run reminded me that there's also the possibility of getting the sandals first. I'd actually tried that method before in a previous attempt, but by the time I'd reached the tourist agency, I was only about 50 frames ahead of JuanValdez's run, and I attributed that to some other optimizations. But now that I know a little more about moving through town in a fast manner (e.g., jump diagonally instead of walk diagonally), I'll try the method again and see if it saves some time. Besides, I also noticed that the patterns of enemies can somehow be manipulated, so I may be able to gather coins for the sandals a little quicker than before, given numerous rerecords.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
... That was pretty good, considering that you said there's more optimization to be done. It looks like the level 1 part of your testrun is a few hundred frames faster than mine. I don't know how much ROM differences contributes to that, but if the people here are OK with a TAS of the Japanese version, then I will just step aside and let you make a complete TAS.
Post subject: Legend of the Mystical Ninja WIP
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Hello. I noticed that, looking at DarKobold/Yeoj's attempts at a Legend of the Mystical Ninja TAS, there are certainly improvements to be made over JuanValdez's TAS. I decided I should try doing this game as my first TAS because it's a game I enjoy and hasn't been highly optimized throughout the game by the more experienced TASers here. I've finished level 1 up to the point where the tourist agency starts fading out, and thought I should put it up to see if anybody's got comments or suggestions. Level 1 TAS Some notes: This run saves 568 frames over JuanValdez's run of the first level (unless I'm counting frames incorrectly), and 246 frames over DarkKobold/Yeoj's aborted second attempt at a TAS, in the first level. Things that save time compared to JuanValdez's run: -Getting through the intro screens quicker. -Going through various sections of level 1 faster, with many thanks to DarKobold/Yeoj for their trick of jumping through barrel corners. -Faster scrolling when talking to people. (many rerecords associated with this!) -Faster scrolling through the Horo dungeon. NOTE: If you're wondering why I jump over some of the enemies in Horo, it's because it turns out that jumping over an enemy without attacking it is 1 frame faster than attacking it in midair. So I only kill the enemies I need for a pipe upgrade. -Faster boss fight. Many thanks to DarkKobold/Yeoj and JuanValdez for setting a standard to try to beat.
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
comicalflop wrote:
I have been getting consistent comments about SMK freezing. I'm not 100% sure if this has been addressed or not, but enough people have talked to me about it that it seems like a problem that needs to be looked at. Does anyone know why this occurs? wrong snes9x used? Is it something to do with a bad ROM dump? system based?
Huffers wrote:
Koopa Beach 1 is taking a while as I'm experimenting with a couple of different routes/strategies as well as different characters.
Use Toad! Toad Toad Toad!
freezing? I haven't had any problems with that using Snes9x 1.43+ v9 with a good US Super Mario Kart ROM NSRT output of ROM:
---------------------Internal ROM Info----------------------
       File: Super Mario Kart (U).SFC
       Name: SUPER MARIO KART       Company: Nintendo
     Header: None                      Bank: HiROM
Interleaved: No                        SRAM: 16 Kb
       Type: DSP-1 + Batt               ROM: 4 Mb
    Country: USA                      Video: NTSC
  ROM Speed: 120ns (FastROM)       Revision: 1.0
   Checksum: Good 0xEB44              CRC32: CD80DB86
--------------------------Database--------------------------
   Name: Super Mario Kart
Country: USA                    Revision: 1.0
 Port 1: Gamepad                  Port 2: Gamepad
Genre 1: Racing                  Genre 2: Battle
Experienced Forum User, Published Author, Player (215)
Joined: 2/12/2006
Posts: 373
Location: Oregon
Not that it really matters, but the exact dump this video was done with, according to GoodNES, is Ninja Gaiden (U) [b5].
1 2
14 15