Either my connection was down at the time (acmlm.cjb.net is hosted on my computer) or port 2 is blocked, but here they are anyway:
(both use the V1.1 ROM and will desync in V1.0)
Old: http://dehacked.2y.net/microstorage.php/info/2564/Lemmings-4.smv
New: http://dehacked.2y.net/microstorage.php/info/2565/Lemmings-b1.smv (now with level 4 done)
Level 4 should be optimal, I just had to make sure only one lemming could cross the hole (as soon as possible) and would also be the last to fall down.
5 and 6 are simple enough, so I'll probably see what I can do about 7 tonight ...
Hm, this one uses Lemmings (U) (V1.0).smc, while I used Lemmings (U) (V1.1) [!].smc ... not sure what's different between them if anything, but the timing is a few frames off (desync) if I use the wrong ROM ...
Anyway, comparing with my run now:
(end of each level, when it fades out completely)
lvl -- mine - yours ---- diff
01 -- 2563 - 2536 --> +27 (gain: +27)
Small gain from digging the minimum height, while I was 1 pixel too high.
02 -- 4357 - 4324 --> +33 (gain: +6)
03 -- 6384 - 6461 --> -77 (gain: -110)
Setting the 99 drop and first blocker was too slow.
04 -- 9429 - 9663 --> -234 (gain: -161)
Bad timing for when the mining ended, all the other lemmings had to walk back and forth the whole way before falling.
05 -- 12566 - 12783 --> -217 (gain: +17)
06 -- 14642 - 14947 --> -205 (gain: +12)
07 -- 19032 - 19071 --> -39 (gain: +166)
Good improvement there, yet it does look like it could be even better, with the 2 lemmings you missed ...
08 -- 21454 - 21494 --> -40 (gain : -1)
09 -- 24219 - 24208 --> +11 (gain: -51)
All the messing around in my run must have cost more time than I thought here ...
10 -- 25568 - 25641 --> -73 (gain: -84)
Slow start again.
11 -- 29677 - 30003 --> -326 (gain: -253)
Faster digging from the right (using multiple lemmings) but you didn't go all the way with it, which must be how you lost 4 seconds ... I used the spare bashers on the left side instead, but I'm not sure which one is faster.
12 -- 33112 - 33058 --> +54 (gain: +380)
Possibly more lag by not killing off the last lemmings, but using bombers to clear the way at the end was apparently much faster (and smarter) than builders.
13 -- xxxxx - 36942
Mine stopped there so I can't compare, but I'm sure you were a little faster there as well, by setting the diggers in a way the last lemmings can simply walk down (I made all of them dig).
Overall, this is hardly an improvement on mine yet, although some parts were definitely better ... also remember you can hold B to move the cursor faster, which really helps sometimes.
I'm working on my own improvement now as well to see if I could do better, and redid the first 3 levels so far:
http://acmlm.cjb.net:2/emu/snes/snes9x/smv1/Lemmings-b1.smv
Comparing this new one with yours:
01 -- 2536 - 2398 --> +138 (gain: +138)
A new trick I found: 10 lemmings together dig faster than one!
02 -- 4324 - 4187 --> +137 (gain: -1)
03 -- 6461 - 6115 --> +346 (gain: +209)
The drop rate and first blocker were optimized better.
Finding the optimal drop timing for level 4 won't be easy ...
Comparing the times with my run:
1029 - 1027 --> +2
2507 - 2502 --> +5
3050 - 3049 --> +1
4296 - 4296 --> ±0
4873 - 4872 --> +1
The loading time does seem to vary, while everything was actually very close so far.
I think it starts lagging after about 50 lemmings, and I know explosions can cause lag too, but I don't think simple animations would ...
That's what I did too, I gave it right before the first one touched the floor (and let the rest die), and did the same for level 10 ...
But I did do a few minor mistakes as well:
- No fast cursor movement until level 10
- I kept too many lemmings on screen a few times (causing lag) or set the drop rate to 99 too soon, like on level 11 and 12
As for messing around, while I did overdo it a few times (mostly level 4), it's also not going to be possible for most of the later levels when they won't give so many spares to waste (or even any), so it won't be like this all the way through. I wouldn't play the same songs over and over either (it's loud too), but at least it passes time when there's nothing to do for a long time (like level 13 where I stopped) ...
I actually got to Fun 13, back in August last year, then I think I stopped there because the next few levels were going to be long and maybe not as interesting ...
A full run would be pretty nice to see, but there's 125 levels (4 sets of 30 + "Sunsoft" set of 5) and quite a few are only harder versions of previous levels, so it'd last about 2 hours ... then you'd also have to find ways to make it entertaining all the way through, mostly during waiting periods, like how I did music and messed around with spare lemmings (without wasting time).
The bonus at the end of each level can be skipped, without having to avoid coins ... that'd save about 2 minutes total:
NES Panic Restaurant in 14:26
That was one of my first runs (November 2004, not submitted), played mostly at 25% speed, so it can still be improved more ...
This run looks good overall, or at least I didn't see other obvious possible improvements (not sure how well the lag can be avoided), but the time lost on bonuses is a lot.
Solomon's Key 2 has been done, too (and rejected) ...
- Static (not really a game, but that's the point)
- test cartridges
- Donkey Kong Jr. Math
- Pacman
Yours desyncs for me, once you start the first mission (while mine runs fine here) ... I have the same ROM, so it's either the emulator version (I got v17 and v19, same result in both) or settings.
My time on the second mission was 2:06 (5 bullets, 3 alarms, 65%), if you need to compare ... nice improvement on the first one already (although I can't view it yet), so good luck for the rest!
I stopped in the 3rd mission near the end when I couldn't reach an alarm trigger in time (and couldn't avoid the alarm without waiting for a while), but I found a bug there that I haven't been able to reproduce after ... if you enter an elevator exactly when the alarm times out, it'll stay on forever (no fail), which helps a lot since you don't have to worry about it anymore.
Other odd things I've seen involve Sam turning away from a safe while unlocking it, guards disappearing (too many or running too far?), and Sam's sprite going over a guard's while entering an elevator ...
You missed a 36 somewhere, but I wasn't quite right either ...
Cards: 16*16 (I counted 16*15 earlier, then remembered they're not on the same set)
Going first: 2 (I forgot about this too)
1+1: 36
3+5: 9
2+4: 9
3+3: 36
1+1: 36
1+3: 18
5+5: 36
1+2: 18
16*16*2*36*9*9*36*36*18*36*18 = 22 568 879 259 648
Rising Tempest wrote:
If 1P is black, you can win in two moves...
Right, but good luck getting the CPU to do the 2 required (bad) moves ... but I'm surprised that 3 turn one worked, the CPU did pretty much suicide there.
I'm sure it could be done in 30 seconds on console, if you try to redo the same moves I did and get lucky with the CPU (same random choices) ... but that's really only because the odds to get everything to work aren't that low for such a short game. I won in 8 turns (CPU got 7) and some of them weren't random, so it could be done within a few dozen tries at most (just not as fast).
(edited: this was before seeing the above post, looks like it's even easier)
For comparison, to reproduce my 30 second Monopoly run on console (same cards and rolls), the odds would be about 1 out of a few trillion (10 579 162 152 960, divided by the few slight variations that could work) ... low enough to be considered impossible, unless everyone in the world decides to try it at the same time for a few days nonstop ...
I did this a year and half ago:
http://acmlm.cjb.net:2/emu/nes/fceu/fcs/Chessmaster,%20The%20%28U%29.fcm
18.95 seconds (1137 frames) would be the time to beat.
I won in 8 turns and manipulated luck a bit, but there might still be a faster way ...
That's at the lowest difficulty (default) too, but the waiting times for CPU turns would be too long otherwise ... Remaking a "legendary game" would only work in 2 player mode and wouldn't be much more than a replay that could easily be done on console.
Hm, it works here, but my connection was down for a while earlier (acmlm.cjb.net is hosted on my computer) ... try again now, I'll upload it somewhere if it still doesn't work for you.
Most of the planning for this run is to find out which weapons I should use, but that's usually not too hard to figure out ...
My 100% run will do this:
- Do all optional corridors
- Beat all minibosses
- Max out every weapon (but not extras and all shops, not sure about enemy erasers)
- Get all score bonuses (if the last one is reachable, not sure how many points it takes)
I also try to score as much as possible without wasting time, especially in the corridors where I shoot everything (almost) and even destroy as many projectiles as I can (which means I have to let enemies shoot them), so there's a lot of action at times ... I'm guessing it'll get even better as I progress and get through the harder levels with more powerful weapons.
But this is going to be a long run, at least twice as long as a minimal run (I'd say a hour and half) ... I'm now in the middle of corridor 13 (area 3), after 23 minutes:
http://acmlm.cjb.net:2/emu/nes/fceu/fcs/guardianlegend-2.fcmTilus wrote:
Run desynchs very early in the opening stage. (Yes, I have the correct ROM - checksum matches and everything.)
This was recorded with version 0.98.15, so it might desync in other versions (I don't have 0.98.16 to test on) ...
That was one of my old favorites on NES ...
The NES version had a 2 player mode too, but if you can manipulate luck in the Genesis version, that's definitely better.
I tried it myself on NES late last year and got to 71 in 3 minutes:
http://acmlm.cjb.net:2/emu/nes/fceu/fcs/Klax%20%28U%29.fcm
The "You must catch __ tiles" waves go so fast when you throw blocks, since the same one counts each time you get it back, and they all accumulate as you get more.
By 100% run, do you mean getting all the powerups and weapons, or just going through all 21 corridors (only 11 are required)?
I also never played through the whole game before, so planning everything is harder, but that's what FAQ's/walkthroughs are for ...
The game does seem to have good potential for a run, though, so I did some progress on it tonight:
http://acmlm.cjb.net:2/emu/nes/fceu/fcs/guardianlegend-1.fcm
(completed corridor 1)
I got halfway through in the NES version last year, and there was another thread about this game here:
http://tasvideos.org/forum/viewtopic.php?t=2081
About a hour for the completed run sounds right, at least for the NES version ... some differences between each (NES, SNES, Genesis) could change things, so the planning wouldn't be the same for SNES.
Either way, here's what I noticed for now:
- I didn't recharge ring power, since most monsters die fast enough already and full power only lasts 1-2 minutes ... I just kept it for bosses, but did use it while leveling on a monster that took 12 hits (it was about 70% full by then). I'm just not sure the time it takes to recharge and switch rings (and go back several times) is shorter than what you save by killing in less hits, but then I haven't tested it ...
- You bought an armor and shield, while I only went for swords, is it worth surviving a few more hits? I never got hit except against bosses (refills HP when defeated) and just before gaining a level (refills HP on NES), just about anything would kill me in about 3 hits. Also, spending more means it'll take longer to get enough gold for the Broad Sword ...
- Since you'd have to spend time walking back and forth for leveling anyway, wouldn't it be faster to kill the monsters in the way before getting the ring (to make it shorter after), or do they really take too long for that?
(this was before watching WIP 2, which I'll do now)
I made this a few months ago when there was some mini contest to score as much as possible in the first autoscrolling level (up to the boss):
http://acmlm.cjb.net:2/emu/nes/fceu/fcs/guardianlegendscore.fcm
It looked good, but that's really just a start since it doesn't even get to the actual dungeon part ...
I had this game on console too, I remember getting pretty far in Rescue and Puzzle and even scoring 400 million in Rescue (it took hours and the run still did more in 6 minutes, haha) ... but that was around 1997-1998 when it was new, I didn't play it much since then and wasn't even that good in Time Trial, at least until I decided to start this run.
I recorded one just now where I scored 15 million in realtime:
http://acmlm.cjb.net:2/emu/n64/mupen64/movies/tetrisphere-realtime.zip
It's much easier to follow and might help understanding the tool-assisted one better (just think of it as what would happen if I could think/move 10x faster and never miss).
trazz wrote:
I could be completely wrong, but, it may be that the best way to get a high score is to get the x20 multiplier, pass as many levels as quickly as possible and then do the last level in one gigantic overtime combo.
(This is just a wild guess but it's based on the fact that Acmim's score in level n+1 is always much larger than his score in level n. I don't know anything about how the scoring system works but I must wonder if level 5 grants even more points.)
The score speedup was mostly from coming up with better strategies, the very fast scoring at the end could have happened on level 1 (after hitting 20x) or 2 ... I'm thinking it would go like this, with a score going over 1 billion:
1- Max out the bonus multiplier as fast as possible, mixing the 2 methods (gravity and fuse combos) instead of doing them individually
2- Clear 9 power (shining) blocks as early as possible to max out the combo points per block (10000 and 900)
3- Boost the combo score as much as possible for about half of the level (to build it up enough), by dropping blocks on vanishing combos for bonus points
4- Finish the level with a lot of gravity combos for huge bonuses (a few million each, depending on the combo score)
5- Repeat steps 2 to 4 for each level, the bonus multiplier doesn't reset
6- Try to finish level 3 in the last few seconds (getting this to work perfectly without wasting time or rushing to finish would be hard to plan)
7- Do steps 2 to 4 during overtime for level 4, like this run did at the end
Sliding also tends to go faster when I'm not dropping blocks, and they can only be dropped about every 10 frames.
XIF wrote:
about that glitch where you can move blocks across impassable places: could it possible be abused further by using a bot with more strength (like Stomp) or is it more practical with the average bots? Or perhaps bots with faster movements?
I think all this changes is how fast you move when only holding the directional pad, all of them can do a full 30-60 steps per second when using frame advance.
Xoria wrote:
Anywho, I can't help but wonder just how much faster this run would be if items were used. I mean, I know items weren't used, and there might be a TON of ways to break the puzzles down, especially if items were used.
I hope to see more runs for this game. It's such a great game to play, and just mind boggling to watch.
Using items would obviously clear the levels faster, but there's either no bonus or not enough to be worth it, and I can't do anything (other than moving around) until the explosion ends, so it's really a waste.
As for other modes, I'd have to try VS CPU enough by myself to see how I'd go about it, but I think Puzzle can be cleared within 10 minutes of game time, depending on waiting times (combos, etc.) in some levels.
I haven't tried FractalFusion's yet, but it seemed like it'd work ... the CPU would buy one of the orange properties before landing on Chance and Community Chest, then would be missing $7 to pay for repairs.
I tested the one where Ollie trades all his money away, it's about half a second slower than this submission (setting the money to trade takes too long), but at least it ends about 10 seconds faster.
* Calculate the first roll with no delay (0+0+0), then 0+0+1, 0+0+2, etc. until it gets the dice I want
* Calculate the second roll, same thing
* Third, 4th, and so on
* Once it gets through the whole thing, it counts the total delay and shows what it found
* After it goes through all delay combinations for one roll (0+0+0 to x+x+x, 5-10 is good) or the current total delay is higher than the best found, it continues with the previous roll
I said 70 and 42 frames in my previous one, but that's without counting the better dice manipulation, and I counted the end delay wrong (7+42, so 49) ...
Here's the luck manipulation delays, if I counted them right:
Old:
- Cards: 0+69 frames
- 1st turn: 0+6, 0+0, 2
- 2nd turn: 4, 0+9, 0+4
- End: 7+42
Total = 143
hero of the day's:
- Cards: 0+23
- 1st turn: 0+6, 0+0, 2
- 2nd turn: 4, 0+9, 0+4
- End: 5+58
Total = 111
(same as above, but with "Player 1" and an earlier (and slower) winning timing at the end)
New:
- Cards: 8+5
- 1st turn: 4+0, 0+6, 2
- 2nd turn: 4, 1+3, 0+2
- End: 29+7
Total = 71
With how much of an improvement this new one did, it'll be harder to do better, but I'm sure a few more frames can still be gained ...
Since each roll being affected in so many ways (3 timings + previous turns), that leaves too many combinations to test to find the best one.
Nice, I never realized just the 2 avenues with houses were already enough to make the CPU player go bankrupt ... but most of the improvement came from the much shorter wait (if any?) at the end to manipulate the CPU's last double turn, you were only 18 frames ahead when building the houses.
I'll have to try that second strategy to see if I can get it to work, although it'd only save time after stopping the recording (but still good, like a few seconds) ...
There's Mario Picross on SNES, too (Picross NP Vol. 1 (J) (NP).smc, up to 8), and Oekaki Logic (1 and 2) with the same gameplay ...
But I also think filling up every single image within seconds each would get repetitive long before you're done, since about 90% of the run would be the images being revealed after you solve each ...
It's funny for a short time though, since it goes so much faster than anyone would ever be able to do (even knowing the solution) ... I tried it just once before, and did a 10x10 in 2 seconds.