Least amount of time needed to beat levels

[Dullstar]

(NeoLemmix equivalent)

We are now back on required, then....

Since I plan on doing all versions and maybe a few clones, let's organize these by game.  What's the least amount of time needed to beat official levels?
This is for the Original Game, Ohno more Lemmings!  and Holiday lemmings.
The clock apparently is slower in DOS, and physics may vary from port to port, so we'll do a separate one for every version.
Please mention a version of the game to get it added.
I won't start a chart until someone does one in that version.
I have Linux DOSbox, so I'm going to start with DOS.

Format is time used.

About once every month the results will be released in a spreadsheet and you can view them. Two are provided - one in the MS Excel format, and another that's in an open document format (for those of us who don't like proprietary formats).  EDIT:  May be able to provide PDF instead.

--------Contents--------
01:  DOS/Lemmix
02:  Gameboy
03:  SNES

--------01--------
DOS/Lemmix
Fun:
1. 0:20
2. 0:20
3. 0:21
4. 0:36
5. 0:37
6. 0:22
7. 0:48
8. 0:24
9. 0:29
10. 0:12
11. 0:31
12. 0:30
13.
14. 1:08
15.
16.
17. 0:25
18. 0:48
19. 1:22
20. 0:36
21. 1:20
22.
23.
24. 0:26
25. 0:36
26. 0:42
27. 0:31
28. 1:55
29. 0:59
30. 0:17

Tricky
1.0:54
3.1:06
4.1:19
5.0:54
6.0:39
11. 0:37
15. 0:15
16. 1:24
17. 0:31
18. 0:33
19. 2:23

Mayhem
1. 
2. 
3. 
4. 
5. 
6. 
7. 
8. 
9. 
10.
11.
12. 0:10

Tame
1.  0:19
2.  0:24
3.  0:51
4.  0:14
5.  0:21
6.  0:23
7.  0:25
8.  0:21
9.  0:23
10. 0:42
11. 0:23
12. 0:33
13.
14. 0:44
15. 1:02
16. 0:25
17. 0:11
18. 0:35
19. 0:27
20. 0:27

Crazy
1.  0:20
2.  1:19
3.  0:38
4.  2:26
5.  1:03
6.  1:56
7.  1:34
8.  1:31
--------02--------
Gameboy
Fun

1.   0:36 -
2.   0:34 -
3.   1:41 -
4.   1:34 -
5.   1:23 -
6.   0:54 -
7.   1:17 -
8.   0:53 -
9.   0:38 -
10. 1:45 -
11. 1:48 -
12. 1:11 -
13. 3:01 -
14 . . .

-------03-------
SNES
Most should be improvable, as the results on this appear to by max. percentage.
Fun 01: 0'35" 100%
Fun 02: 0'25" 100%
Fun 03: 0'42" 97%
Fun 04: 0'44" 100%
Fun 05: 0'51" 100%
Fun 06: 0'37" 96%
Fun 07: 1'06" 100%
Fun 08: 0'34" 99%
Fun 09: 0'40" 100%
Fun 10: 0'19" 95%
Fun 11: 1'04" 100%
Fun 12: 0'46" 100%
Fun 13: 0'54" 100%
Fun 14: 2'00" 100%
Fun 15: 2'50" 100%
Fun 16: 1'54" 100%
Fun 17: 0'56" 100%
Fun 18: 1'00" 92%
Fun 19: 2'00" 100%
Fun 20: 1'03" 100%
Fun 21:
Fun 22:
Fun 23:
Fun 24:
Fun 25:
Fun 26:
Fun 27:
Fun 28:
Fun 29:
Fun 30:

[namida]

This is basically just a clone of the "Completeable under one minute" topic, IMO. A more interesting one may be "Quickest time saving the maximum amount of lemmings" or something like that.

[ccexplore]

This is basically just a clone of the "Completeable under one minute" topic, IMO. A more interesting one may be "Quickest time saving the maximum amount of lemmings" or something like that.

Well, this does generalize the "under one minute" concept so that all levels can be included, such as those that simply can't be done under one minute due to the sheer amount of walking alone.  Or the levels that already only have one minute, it could still be interesting to see what it takes to get the quickest time possible.  I was actually thinking of proposing this thread when the "under one minute" thread was winding down, until ClamSpammer and you guys started coming up with all these new challenge threads.

The only caveat is that it can be kinda a pain to rework the same basic solution idea just to squeeze out the shortest amount of time.  I expect we'll come up with interesting solution ideas that reaches the ballpark of the quickest time, and then others can try to optimize the execution to achieve the quickest time possible with the given solution.

[Dullstar]

namida, this is not a "One minute levels" thread.  This is similar to the least number of skills thread; we are not making a list of levels.  We are figuring out what the least amount of time is for beating the levels.

[Dullstar]

A more interesting one may be "Quickest time saving the maximum amount of lemmings" or something like that.

Good idea, actually.  This one can be for just saving the required percentage, and then someone could make that thread for the maximum percentage.

[Clam]

I'm not convinced that this will be worthwhile. Maybe it's just too soon after the "one minute" thread? I dunno. Some levels (long ones in particular) are just going to get ridiculously complicated. It's a nice idea in theory though.

We definitely need a consistent timing convention - I'd suggest Lemmix, since it avoids complications like pausing for time, and we can share replays easily.

Anyway, your score for Fun 7 is impossible. (edit: it's been changed. That looks a bit more reasonable)

[namida]

The question is, though, is the exact maximum for some levels (eg: Poor Wee Creatures and others that require building a landing space) known?

For example, Pillars of Hercules, I think I heard a while back that the maximum was somewhere around 80%, yet I can easily save 94%... and some of the more knowledgable players may be able to do better yet via glitches (or even just through tricks I overlooked).

Fun 13: 2:14 remaining on Lemmix (46 seconds)

PS: Your introduction post still says  "Format is time remaining", while listing time /taken/ in all levels. You also still have the note about aborting when required amount is saved on Fun 7.

[Clam]

We really can't ever be absolutely sure what the maximum is for each level, but we do have records for each level which are fairly well established.

Here's a replay of Fun 13 completed with 2:15 left in Lemmix. And herein lies the problem: Even on a seemingly cut-and-dried level like this one, it's not so easy to find a solution and say, "This is the fastest possible time for this level. End of story." Just imagine trying to do this on a huge level like Mayhem 30...

[Yawg]

I personally think this is a great concept for a challenge thread. I consider myself a fairly experienced lemmings player, yet I'm often intimidated by some of the challenge threads, which tend to rely heavily on one's ability to fine-tune the release rate, something I've never gotten the hang of.

The reason I like this challenge so much is that ANYONE can contribute. Someone can present an expedient solution, which can then be refined by everyone else. While we may not be able to find the absolute, definitive minimum times for each level, we can probably get pretty good estimates. And lets face it, the timer doesn't even count milliseconds, so our estimates can at least be precise on the order of magnitude in which the game operates. 

I took a crack at Fun 11 and immediately came up with a solution (4:29 left on the clock with the required 83% saved) I would never have thought to try otherwise. To me, that is a sign of a good challenge thread.

[Yawg]

I just realized the ambiguity over max % vs required %...

Which are we going for? Personally I think minimum time for required % makes the most sense, so you aren't wasting time watching every lemming walk home...

Also, I think we should report the times by amount left on the clock, as opposed to time taken. For simplicity's sake.

Here are the numbers I got for fun, saving required %, in Lemmix. I skipped a few out of frustration.

1. 4:40
2. 4:40
3. 4:39
4. 4:22
5. 4:23
6. 4:38
7. 4:12
8. 4:36
9. 4:31
10.4:48
11.4:29
12.4:30
13.
14.4:52
15.
16.
17.4:35
18.4:12
19.3:38
20.4:24
21.6:40
22.
23.
24.4:34
25.4:24
26.4:18
27.4:29
28.3:05
29.7:01
30.4:42

[Dullstar]

To me, it makes more sense to get max. percentage, because it reduces the number of "Shoot, didn't catch the time remaining," incidents.

[Yawg]

Well that's what replay and pause are for. And you'll have that problem either way if your solution leaves blockers (or trapped lemmings) lying around.

[Dullstar]

Lemmix is Windows only, and I haven't been able to get Lemmini to run at all.

Notice that DOSbox has a Linux version.
That's why there are different stats for different versions.  Clock speeds have to be taken into account, along with any physics differences.

[Clam]

I think it should be required %, to be consistent with the least skills thread.

From the 1-minute thread, my score for Fun 16 is 7:04. And I'm sure you could shave a few seconds off that with a few hours work
(edit: 0:56 in "time used" format)

[Clam]

Well this seems to have died already. I wonder if it might be worthwhile to bring this back, but in a different form:

 - Do the Tame levels only.
 - Set separate records for required percentage and 100% (since all Tame levels can be done with 100%).
 - (Preferably) use Lemmix so we can record replays.

Because none of the Tame levels require huge amounts of building or digging through large obstacles, I think a proper speedrun should be feasible here at least. (Unlike some levels... " title="Angry" class="smiley" />)

Anyone up for this?

[Dullstar]

People seem to like the idea of required percentage better, so I swapped it back onto required percentage.  Most of the records so far are now beatable.

[Dullstar]

Well this seems to have died already. I wonder if it might be worthwhile to bring this back, but in a different form:

 - Do the Tame levels only.
 - Set separate records for required percentage and 100% (since all Tame levels can be done with 100%).
 - (Preferably) use Lemmix so we can record replays.

Because none of the Tame levels require huge amounts of building or digging through large obstacles, I think a proper speedrun should be feasible here at least. (Unlike some levels... " title="Angry" class="smiley" />)

Anyone up for this?

Okay...  this is the original lemmings.   Mainly...  but I added Ohno! and Holiday lemmings.  We'll do what we are doing for the skills thread.

[Clam]

Did you even read my post? I said "Tame levels only". The point being that many of the other levels are so long complex that finding a "speed" solution is likely to become a chore.

[Dullstar]

That was kind of a weird way of saying, it will be more than just Tame!

If you don't want to do the later levels, no one's forcing you to.

[Yawg]

That was kind of a weird way of saying, it will be more than just Tame!

If you don't want to do the later levels, no one's forcing you to.

I agree. Why limit ourselves when this is something we could potentially do with every level (like some of the more popular challenge threads)?

[Dullstar]

Yawg, can you convert the records you posted to Time Used?
Then we would have a Lemmix set.

[Clam]

I still have my doubts as to whether this will be worthwhile, but I'll play along for now. Before we continue, though, we need to sort out a few things:

• Pausing for time and nuke glitch. Are these allowed, given that they don't work in Lemmix?
• To make sure that everyone is consistent in timing, here is what I suggest: Pause the game as soon as the required percentage is met. And by that, I mean the same exact frame that the percentage ticks over to the required amount. This means you probably should save the replay first, and then check the timing. Whatever the clock reads at that moment, subtract that from the total available time to get the elapsed time. (And then take off 2 seconds for the pause trick if we decide to allow it.)
• Save replays. You don't always have to post them, but keep them handy.

[Dullstar]

Glitches aloud even if they don't work in Lemmix.  Truthfully, even though it may be a very small chance, the chance is still there that mechanics are different enough to change the possible times.  Have you seen my edits to the introduction post?  You'll notice that I've started a section for Gameboy.  The reason I have not placed a Lemmix one is that I'm waiting for the records to be changed to Time Used so I can post them.

Lemmix records get their own section.  Amiga gets its own section.  Genesis especially gets its own section.  SNES gets its own section.  Blah blah blah.

By the way, I think the Gameboy Color version might be slightly different mechanics-wise from the Gameboy version, plus a few level differences...

[Yawg]

Here are the times I got for fun, and a couple from tricky. Format is now time taken (as last required % ticks). All done in Lemmix.

FUN
1. 0:20
2. 0:20
3. 0:21
4. 0:18
5. 0:27
6. 0:22
7. 0:48
8. 0:24
9. 0:29
10.0:12
11.0:31
12.0:30
13.
14.1:08
15.
16.
17.0:25
18.0:48
19.1:22
20.0:36
21.1:20
22.
23.
24.0:26
25.0:36
26.0:42
27.0:31
28.1:55
29.0:59
30.0:18

Tricky
1.0:54
3.1:06
4.1:19
5.0:54
6.0:39
11.0:37
15.0:15
16.1:24
17.0:31
18.0:33
19.2:23

[Clam]

Glitches aloud even if they don't work in Lemmix.  Truthfully, even though it may be a very small chance, the chance is still there that mechanics are different enough to change the possible times.

They definitely do make quite a difference between DOS and Lemmix. Pausing for time is good for two seconds per level for a start, but the nuke glitch can make quite a difference too.

Lemmix records get their own section.  Amiga gets its own section.  Genesis especially gets its own section.  SNES gets its own section.  Blah blah blah.

Does that mean DOS and Lemmix have separate sections? I'm fine with that (in fact, I think that would be a good idea), as long as we can use Lemmix to set records for DOS as well. Because they essentially have the same mechanics, I think this should be allowed. Lemmix is just so much easier to use...

[Yawg]

Speaking of how easy Lemmix is to use, I just whipped through Tame and half of Crazy using it  I skipped 13 because I just couldn't come up with a really time-effective approach, and Crazy 8 is my speed-run challenge to you all. I'm fairly sure its the absolute best possible time 

Tame
1.  0:19
2.  0:24
3.  0:51
4.  0:14
5.  0:21
6.  0:23
7.  0:25
8.  0:21
9.  0:23
10. 0:42
11. 0:23
12. 0:33
13.
14. 0:44
15. 1:02
16. 0:25
17. 0:11
18. 0:35
19. 0:27
20. 0:27

Crazy
1.  0:20
2.  1:19
3.  0:38
4.  2:26
5.  1:03
6.  1:56
7.  1:34
8.  1:31

[Clam]

While I won't post any times right now, I can say that some of those times for the Tame levels are improvable. The key is to reduce the time spent falling (and thus not walking towards the exit), without letting any lemmings fall behind the last required lemming.


To add to what I said above, I've found an extreme case of time saving by the nuke glitch. Fun 8 takes a minimum of 24 seconds in Lemmix, but in DOS it can be done in just 5 seconds. And only 2 seconds of that difference is due to the pause trick. I wonder if an even more extreme example exists...

[namida]

...Wild 9?

[namida]

Here are the times I got for fun, and a couple from tricky. Format is now time taken (as last required % ticks). All done in Lemmix.

FUN
1. 0:20
2. 0:20
3. 0:21
4. 0:18
5. 0:27
6. 0:22
7. 0:48
8. 0:24
9. 0:29
10.0:12
11.0:31
12.0:30
13.
14.1:08
15.
16.
17.0:25
18.0:48
19.1:22
20.0:36
21.1:20
22.
23.
24.0:26
25.0:36
26.0:42
27.0:31
28.1:55
29.0:59
30.0:18

I've already beaten Fun 1, taking 0:18. Replay attached. I'll try the rest too.

Fun2: No replay needed IMO. If you wait until the very last possible frame before making the lemming a floater, he gets in JUST before the clock switches to 4:40.
Fun4: Do you have a replay of this one? I don't think it's possible.
Fun30: Beaten by a single frame. The % in this replay hits 66% the frame right before the clock ticks over to 4:42.

[Yawg]

Mis-copied Fun 4. Supposed to be 0:38. Stupid 100% requirement

[namida]

Done in 0:36.

[Clam]

...Wild 9?

Thanks for not rubbing it in.

[Clam]

So here it is again. I pretty much gave up on this back when it started, and I pretty much killed it off by posting my disapproval. (Terrible habit of mine, I'm afraid. ) But now I think it's time to reconsider.

Lately I've been checking out some Tool-Assisted Speedruns (check out Sonic 1 & 2 on Youtube - they're awesome), and I happened upon this video of a SNES Lemmings run. Now, I'm not suggesting that we do a speedrun on an emulator aiming for the quickest time in terms of real time (as they do here) - that would be a huge amount of work - but the original aim of this thread isn't all that different.

I'm willing to give this another try. Who's interested in helping out?

[Clam]

@Dullstar: Are you still OK to maintain the list, or should I start a new thread and keep the list myself?

This challenge should be much more accessible now that we have a proper speedrunning tool available. We're in full control of the game - it's all about strategy, and execution of that strategy is all but trivial. If anyone needs help, I might be able to write up some sort of guide (with things like release rates, how to figure out the number of lemmings needed when the nuke glitch is used, etc).


Since the situation has changed here, I should clarify the timing conventions for DOS at this point:

• Pausing for time is assumed to be used in all cases, to maximum effect (just over 2 seconds).
• Nuke glitch is allowed, and in many cases will lead to a vast improvement in speed.
• The time taken is the time limit for the level minus the time left on the clock on the first frame in which the percentage requirement is met. You can determine this by frame advancing as the lemming exits.

[Clam]

Now it's time to kick off this challenge again.

Here's Mayhem 12 in 10 seconds.

[Dullstar]

I think it should be required %, to be consistent with the least skills thread.

From the 1-minute thread, my score for Fun 16 is 7:04. And I'm sure you could shave a few seconds off that with a few hours work
(edit: 0:56 in "time used" format)

I'd post your result in this case, but I don't know what version this is.  Lemmix or DOS?  Or something else?  If you remember, can you let me know?

[Clam]

Lemmix and DOS are equivalent now, since ccexplore fixed the LemmixPlayer to reproduce bugs like nuke glitch, pausing for time, steel digging etc. By the way, you'll need to download this version to view my latest replay

[Dullstar]

I thought the clock is faster in Lemmix, which would cause the results to be a larger number than they should be.  Did ccexplore fix that as well?

Well, the clock being faster is just what I heard, anyways.

[Clam]

I don't recall the clock being too fast in Lemmix at any point. That must have been a different version (Amiga perhaps?)

Anyway, since Lemmix emulates DOS properly now (or near enough), you can merge the two into a single list. When you watch a replay in the new LemmixPlayer, make sure the "pausing for time mode" is on (press "z" to toggle this).

[Dullstar]

Hmm, we could record how long 5 seconds is in Lemmix and then in DOS and compare the difference...
However, they're staying as separate lists until someone proves that the clock in Lemmix is the same as in DOS.

[Clam]

Okay, I know. Take a level that takes a long time to complete but has a really simple solution. Like Tricky 2. I just played this level in DOSBox, LemmixPlayer (with pause for time mode off), and the editor playtest, without changing the release rate. In each case, there was 15 seconds left at the end, i.e. it took 2:45 to complete. Over this length of time, if the number of frames per second was different by just 1, it would change the time taken by nearly 10 seconds. Go ahead and try it yourself (I recommend digging the last lemming only, otherwise it will get quite tiresome. The last one is the only one that matters anyway with regards to time taken.)

[Dullstar]

I'll get to merging those a little later.  Although, if you do it, it'll get done sooner.  Either way, it'll get done.

[ccexplore]

However, they're staying as separate lists until someone proves that the clock in Lemmix is the same as in DOS.

One of the things I did when helping Erik with the game mechanics is measuring the time limit.  I created a test level consisting solely of 1 lemming falling out of the trapdoor and walking to the exit, with a 1 minute time limit, no pause trick.  I keep moving the exit away by 8 pixels until the lemming couldn't reach it anymore (ie. game says I saved 0% after time runs out), then move it back to the last position it is reachable.  I then move the trapdoor up by 3 pixels each time (3 because that's how many pixels a lemming falls in one frame) until again the exit becomes unreachable in time.

It's testing like this that led me to discover the slight difference between original DOS Lemmings and ONML in the position where lemmings started to enter (entrance.x + 24 in Lemmings, entrance.x + 25 in ONML).  So I'm fairly confident that Lemmix should match DOS exactly in terms of time, but I'll double-check again later tonight to be sure.

I've also been going through more of the game's code in the past day or so, and I think I finally have the correct number of frames gained by the pause trick as well.  I'll test out and make the changes in Lemmix later today or tomorrow.

================

One thing I should note is that the amount of real time per frame in Lemmix may or may not match DOS exactly.  However, in terms of game time (ie. how many frames before the game clock goes down 1 second, etc.), Lemmix should already be matching DOS Lemmings, and my understanding is that we're measuring time in terms of game time here.

[Clam]

I've also been going through more of the game's code in the past day or so, and I think I finally have the correct number of frames gained by the pause trick as well.  I'll test out and make the changes in Lemmix later today or tomorrow.

Is it more or less than what it gives you now? Come on, don't leave me hanging...


Here are my provisional results for the first 10 levels of Fun:

1. 0:15
2. 0:17
3. 0:18
4. 0:34
5. 0:33
6. 0:21
7. 0:43*
8. 0:05
9. 0:29
10. 0:08

* Still a work in progress. If cc gives us one more frame on the pause trick, I'll have this down to 42 for sure. Otherwise, maybe.

I'll post replays after I check the timing on the new version.

[ccexplore]

I've also been going through more of the game's code in the past day or so, and I think I finally have the correct number of frames gained by the pause trick as well.  I'll test out and make the changes in Lemmix later today or tomorrow.

Is it more or less than what it gives you now? Come on, don't leave me hanging...

Unfortunately it's less I think.  Basically, due to a quirk in the game's programming, you cannot pause during the first 2 frames.

As part of the latest round of looking at the game's code, I'm also going to review the order in which Lemmix handles certain things compare with DOS Lemmings, to make sure there aren't any differences that can affect anything.  I don't think there'll be any, but as you said, it's worth it to get it right.

[Clam]

Wait, does that mean the most you can gain is less than 2 seconds? I could have sworn it was more than that

[ccexplore]

Wait, does that mean the most you can gain is less than 2 seconds? I could have sworn it was more than that

Well, from the game's code, it looks like the entrance starts to open at the 36th frame (counting from 1), so I think the gain is still exactly 34 frames =  2 game seconds compared with no pause trick.  However, this does mean my current method in Lemmix of resetting the clock is giving you 2 more frames than you could actually have in the real game, and that it's making the game clock go down 2 frames later than they should, I think.  To be 100% sure I also need to go thru Lemmix's code also to see whether it's opening the entrances at the exact same time it would in the real game w/o the pause trick.

[ccexplore]

Okay, I've learned what I needed from looking at the game's code and Lemmix's code, and unfortunately it's not going to make my job any easier.  In fact, it leads to some interesting findings that may affect how you want to measure time for this challenge thread.  For example, do you know that in high-performance PC mode (where the game fades in and fades out slowly), if you press ESC to abort, during that slow fade-out, the game continues to update the lemmings, but does not update the timer?

Anyway, I'll mention all the gory details in the "LemmixPlayer ccexplore updates" thread.  No actual updates though--some care is needed to make all the necessary changes w/o invalidating all existing replays, and certainly some thorough testing will be called for.

[Clam]

So if I were to read the time off the game clock with the pause trick active, how far off would I be? If I know this, I can post some actual results without the need for a "fixed" version.

I don't know whether I should switch over to "High-performance PC" mode or just stick with the default "PC compatible" style. Glitches are one thing, but switching game modes to get faster times seems a bit... wrong. I don't know how to explain it.


In other news, I've improved my time for Fun 1 to 0-14, and Fun 6 to 0-17. For some bizarre reason, I was blocking my lemmings at the start before instead of letting them walk. Again, I'll post replays once the times are confirmed. (And I'm keeping notes too.)

[ccexplore]

So if I were to read the time off the game clock with the pause trick active, how far off would I be? If I know this, I can post some actual results without the need for a "fixed" version.

I still need to find out whether the clock display in the real game reflects the internal clock variables on the frame or on the previous frame.  If we take the current Lemmix model, where the time you see in a frame is exactly the values in the internal clock variables, and ignore the thing about ESC and/or high-perf, then right now, I think Lemmix is 3 frames too early (2 from not being able to pause on first 2 frames in real game, and 1 from the entrance opening one frame early in Lemmix).  I think.

For something as simple as Fun 1 and Fun 6, if you really want some results now that are guaranteed replicable in the real game, you can use my "vgalslow.exe" slowed-down version of DOS Lemmings.  It'll be a bit of a pain to use due to the slowness, but should be tolerable for short Fun levels.

Another thing to keep in mind is that if your solution is optimal in terms of number of frames, then it is an optimal solution regardless of how the game clock is treated, just that your reported result in terms of the game clock might be inaccurate.

[Ramon]

I'd do some for the SNES version of the original Lemmings. (With ZSNES.)

But which timer counts? When the lemming counter is 0, or when the screen has faded out and just about to display the results?

Note: Pause trick (pause before trapdoor opens) for more time is NOT possible on the SNES port.

I'll edit this post when I get more.

Fun 01: 0'35" 100%
Fun 02: 0'25" 100%
Fun 03: 0'42" 97%
Fun 04: 0'44" 100%
Fun 05: 0'51" 100%
Fun 06: 0'37" 96%
Fun 07: 1'06" 100%
Fun 08: 0'34" 99%
Fun 09: 0'40" 100%
Fun 10: 0'19" 95%
Fun 11: 1'04" 100%
Fun 12: 0'46" 100%
Fun 13: 0'54" 100%
Fun 14: 2'00" 100%
Fun 15: 2'50" 100%
Fun 16: 1'54" 100%
Fun 17: 0'56" 100%
Fun 18: 1'00" 92%
Fun 19: 2'00" 100%
Fun 20: 1'03" 100%
Fun 21:
Fun 22:
Fun 23:
Fun 24:
Fun 25:
Fun 26:
Fun 27:
Fun 28:
Fun 29:
Fun 30:

[Clam]

But which timer counts? When the lemming counter is 0, or when the screen has faded out and just about to display the results?

Does SNES have an option to quit to the post-level screen (like pressing Esc in DOS)? If so, then you can just stop the timer once you reach the required percentage.

[Ramon]

Nope, sadly it doesn't. I have to nuke in time.  Oh and I'm trying to save as many lemmings as possible in shortest time btw.

[Pooty]

Actually, it does. If you hold Select and press Start, the game will fade to the post-level screen. However, when you do that, the clock will read 0:00.

[Clam]

Wait, I should have known that - I watched a SNES Lemmings speedrun the other day. I posted the link in this thread, if you're interested.

Before you get discouraged by this:
1) It's a tool-assisted run, and not a genuine speedrun,
2) It aims for quick completion in terms of real time, and not game time, and
3) It starts at Mayhem. Fun to Taxing are completely untouched.

[Ramon]

Actually, it does. If you hold Select and press Start, the game will fade to the post-level screen. However, when you do that, the clock will read 0:00.

I can't believe I never found that out yet. :p

Wait, I should have known that - I watched a SNES Lemmings speedrun the other day. I posted the link in this thread, if you're interested.

Before you get discouraged by this:
1) It's a tool-assisted run, and not a genuine speedrun,
2) It aims for quick completion in terms of real time, and not game time, and
3) It starts at Mayhem. Fun to Taxing are completely untouched.

I myself also use savestates and rewind states to complete the levels in the fastest time :p but I need to add I'm just an average Lemmings player, not an indepth one...

[Clam]

Oh... well, there's nothing wrong with that. I'm doing the same for the DOS version. Heck, I'm taking it even further by using a specialised tool rather than an emulator

[Clam]

I've been quiet over the past couple of days, but that's really just because the forum's going through a bit of a lull. I'm still busy working on this challenge, and I'm really starting to get a feel for it now. It's hugely more involved than any other challenge so far, as you would expect. So progress is gonna be slow of course, since I insist on doing this "properly", and going for the absolute minimum possible (in seconds, not frames - that would just be nuts ).

As I said earlier, I have replays and notes for all the levels I've done so far (about 30 - though this includes trivial ones like Fun 2), and once Lemmix is fixed I'll check all my times and post this stuff.

[ccexplore]

and going for the absolute minimum possible (in seconds, not frames - that would just be nuts ).

I'd think in most cases, the difference in number of retries for seconds vs. frames would merely be (nuts - 17) instead of plain nuts.

[Dullstar]

and going for the absolute minimum possible (in seconds, not frames - that would just be nuts ).

I'd think in most cases, the difference in number of retries for seconds vs. frames would merely be (nuts - 17) instead of plain nuts.

Is this some kind of inside joke?  Anyways, I'm gonna copy Ramon's results into our list.

[Clam]

I'm pretty sure that was intended as a joke.


I guess I should stop holding my breath waiting for ccexplore to fix the LemmixPlayer, and post the results I've got (48 levels done now). But first I'd just like to confirm the timing error in the current version. Let's see if I've got this right...

In LemmixPlayer as it is, compared to PC Compatible mode:
- The pause trick gives you 2 frames too many (+2)
- The entrance opens one frame too early (+1)
- You don't get the opportunity to gain a frame by pressing Esc just as the last lemming is about to exit (-1)

And so to account for this, you need to add 2 frames to times obtained from using the LemmixPlayer. Is this correct?

[ccexplore]

Is this some kind of inside joke? 

Hmm, guess the joke is too obtuse.  The point is that since in DOS Lemmings it's 17 frames to 1 game second, if your result is good down to game seconds, then you're within no more than 17 frames from the least amount of time in terms of frames, so it shouldn't take more than 17 additional tries to nail it down to the least amount of frames.  That would seem to be relatively miniscule amount of work compared with however long it'd take just to get down to the least amount of game seconds. 

Anyways, I'm gonna copy Ramon's results into our list.

Not sure what list you're talking about, but keep in mind that the SNES game clock runs faster than DOS Lemmings, so it's rather pointless to mix results together like that.  Keep a separate list for SNES if you must.

[ccexplore]

I guess I should stop holding my breath waiting for ccexplore to fix the LemmixPlayer, and post the results I've got (48 levels done now). But first I'd just like to confirm the timing error in the current version. Let's see if I've got this right...

In LemmixPlayer as it is, compared to PC Compatible mode:
- The pause trick gives you 2 frames too many (+2)
- The entrance opens one frame too early (+1)
- You don't get the opportunity to gain a frame by pressing Esc just as the last lemming is about to exit (-1)

And so to account for this, you need to add 2 frames to times obtained from using the LemmixPlayer. Is this correct?

My latest finding also shows that the game clock is updated (ie. displayed) a frame later than the internal variables that keep tracks of the time.  Also, I believe it's actually giving you 3 frames too many from the pause trick--the frame count is currently being reset to 0 after the code that increments it, so in effect it "ate" that increment as well causing an additional 1-off in the counting.

In any event, you're probably better off just posting results now.  Keep in mind that (ignoring high-perf mode) your results in LemmixPlayer will always be better than or equal to the actual game, and I can guarantee you that all replays will remain forward-compatible, so you'd never need to redo a solution from scratch to find the correct best time once I've updated LemmixPlayer.  Instead you just need to replay the solution to see what the correct time should be.

[Dullstar]

Anyways, I'm gonna copy Ramon's results into our list.

Not sure what list you're talking about, but keep in mind that the SNES game clock runs faster than DOS Lemmings, so it's rather pointless to mix results together like that.  Keep a separate list for SNES if you must.

The OP currently contains results for Gameboy, DOS, Lemmix, and SNES.  A seperate list is kept for each one, although sometime I intend to merge the DOS and Lemmix lists.

[Clam]

In any event, you're probably better off just posting results now.  Keep in mind that (ignoring high-perf mode) your results in LemmixPlayer will always be better than or equal to the actual game, and I can guarantee you that all replays will remain forward-compatible, so you'd never need to redo a solution from scratch to find the correct best time once I've updated LemmixPlayer.  Instead you just need to replay the solution to see what the correct time should be.

Yeah, but in some cases I'm going to have to find even more "nuts" solutions, when they're already extremely nuts as they are. I'm trying to anticipate that here by adjusting my timing.

But anyway, my results up to now are attached to this post. 49 levels done now, in no particular order. The text file contains notes on my solutions, to get you up to speed with strategy if you want to try some speedruns yourself, and also my handy formula for working out how many lemmings you need when the nuke glitch is used. These scores assume a 2 frame difference in timing between versions.

EDIT: Reuploaded the zip since the notes were messed up a little. (Plus I improved one of the solutions.)

[ccexplore]

About once every month the results will be released in a spreadsheet and you can view them.  Two are provided - one in the MS Excel format, and another that's in an open document format (for those of us who don't like proprietary formats).  EDIT:  May be able to provide PDF instead.

Don't kid yourself, PDF is just as much a proprietary format as Excel, just because it isn't Microsoft doesn't mean anything.  For a truly simple and open document "format", just save it as ASCII text, you know, like something that opens even in Notepad.  Even Excel lets you save as text in its Save As options.

[Gronkling]

I would like to work on this challenge using Lemmix for more accurate physics to DOS however the methods for keeping track used here are really outdated (I mean it is from 2009, I was told to bump), the NL time thread is much more precise.

My way of getting a frame-precise reading of time from Lemmix is to nuke on the first frame possible where you've saved the correct percentage of lemmings (you would normally press esc to exit the level here, but the replay won't show this). You can then read the replay text file and read the frame where you nuked (it'll show up with N in the action section), giving you your final time in frames, which is how times ideally should be recorded (the current method here using the in-game clock is super imprecise for this challenge's purpose). You know when you saved enough as it's the first frame that the correct percentage is shown. I have checked and the pause-for-time glitch doesn't affect this number, so simple to just read off, no extra work needed.

Lemmix/DOS run at a rate of 17 frames per second IIRC, so it shouldn't be too hard to convert from frames to seconds (eg. 123 frames = 7 seconds, +4 frames).

To get things going here's an adaption of namida's NL solution to fun 1, giving a time of 302 frames (17 seconds + 13). I also attached the text file for this one so you can see where I got the number from if you are confused. Here I nuked on the first frame that 10% saved is shown.

[Minim]

Interesting and proficient way to record results this way! However, if you dig too deep into this idea you'll bump into Hail 6 of Holiday Lemmings: "Surprise Package?", one of the only levels I know that requires Nuke to beat it.

I've been working hard improving more levels in the NeoLemmix version of this challenge within the past week. This will be a big update as I've managed to break records for so many levels, but once I release the updated version feel free to have a look at these.

[Gronkling]

It shouldn't be a problem still, afaik you can just nuke again (but i doubt i'll ever get to xmas lems )
Anyway did a quick round over the first 10 fun levels, mostly seem to match or slightly beat the times in the first post
I'll add a marker (G) to solutions that use glitches

FUN 01 - 299 frames - 17s +10
FUN 02 - 339 frames - 19s +16
FUN 03 - 351 frames - 20s +11
FUN 04 - 626 frames - 36s +14
FUN 05 - 617 frames - 36s +5
FUN 06 - 379 frames - 22s +5 (G)
FUN 07 - 822 frames - 48s +6 (Definitely improvements to be made here, will be hard to implement NL solution due to lack of directional select)
FUN 08 - 409 frames - 24s +1
FUN 09 - 496 frames - 29s +3
FUN 10 - 205 frames - 12s +1