We talking like... where the screen itself is positioned horizontally? Or where objects happen to be comparative to be in position? Or is it something to do with the number of pixels on the screen at a given moment?
That's an excellent question which I have no qualification to answer xD If you think of a totally level surface (like level 7), multiple lemmings on that surface cause headaches to draw in unless you scroll away. But a surface that's even slightly uneven seems to drastically change that outlook. I don't think it would be due to pixels on screen because overflow sprites are drawn into the background as I understand it. In any case the lag is pretty easy to manage so far. If you look at my latest video for Fun 3, you'll notice I amp the RR up to 90 and it is the fastest way to get lemmings home - that's partly because after Nuke I scroll the screen anyways, but there are a couple of other levels where you can allow many lemmings on screen at once and have the optimal solution.
Last night I busted Origins and Lemmings down to a 1:39, saving 30 seconds. And I found a one-digger solution for level 13 although using it in a run is a long way off. So maybe I can get the time closer to 35 than 40 minutes!
Practice run of the first 15 levels totalled 12:39.33, saving 2:28 from the old run. No resets and no flubs.
Ah, I think I understand. Like, you were saying that when lemmings are moving horizontally, the slow down is worse than when they lemmings are moving at an angle?
Basically yeah; it seems to be like that. A lot of levels in Fun have relatively loose save requirements though, so as you'll see in the above playthrough, I can often kill a few at the start which maximally reduces any potential lag even from on-screen bunches. Calculating travel time across a level also makes it possible in some cases to release late, such as Fun 19, which even reduces otherwise unavoidable lag frames.
Hope everyone out there had a lovely Xmas and New Year! I took a break from the game to pick up something else. Now I'm going to finish optimising the last 9-10 levels of Fun and take a proper crack at a sub 40 minute run. Here's the latest:
I had a run last night just as practice that turned into a serious attempt on the WR. Going into Level 29 I was on course for a 38:59, total time, then I flubbed the final bridge to the exit :( I'm just finishing up my last optimisations including copious practice on that level to get a reliable method of bridging, then I should be able to work out the final projected time.
Each between-level transition (from the frame where OUT reaches 0, to the frame input is accepted on the level description screen) takes as an absolute minimum, 240 frames, or 4 seconds precisely. It works best if you hyper tap the 2 button to achieve this, since if you press early and hold the button, the input won't be accepted. So this time added to all optimal levels will be 1:56 at the least, or more like 1:58-2:00.
With current optimal strategies (bearing in mind I am yet to tackle a few of the worst levels), the fastest possible completion comes out at 35 minutes 6 seconds. So it DOES look like you can just about go sub 35. A more realistic target would be sub 36 minutes.
This has to be one of the most insane optimisations I've found yet. You save time not only by removing lemmings from screen quicker, but also from not bombing them. That is, if I let 20 lemmings cross the screen and have 10 die at the web, that's still quicker than exploding 10 at the start... but what if I kill them another way?
This saves TWELVE AND A HALF SECONDS :D Total completion time is now in the high 34 minute mark, and with L28/29 to go it could yet go into the 33s.
Given the lag has to be dealt with, and in as such, excess lemmings killed asap, then yea, if there is a way to just direct a given number off the course as opposed to exploding them separately, that seems like it would be way to go. I guess you just got to know exactly which lemming to start digging with though, in this case.
Luckily it seems that it works with the first lemming, it's more a case of timing it to trigger him ASAP then making sure to scroll over and use the diagonal area to set floaters.
I now optimised all 30 levels, I found a perfect reference point to use the long builder in L28 thanks to the brick design in hell, and I found a perfect bridge building point in the cage at L29 too. I'll upload the last few bits and pieces but the estimated perfect completion time so far comes to:
Thirty three minutes, twentyone point three seconds
33:21.3 Based on this, anything below 35:00 is going to take some serious beating. If other systems' respective runners look into their versions it will be highly interesting to see whether anyone can top that; I remember Usumgallu saying somewhere he had a run optimised to within the 30s somewhere.
Unsure how many interested readers I have but I've started optimising Tricky levels again, and I'm having a comfy time of it thanks to good RR management and lag management. Check out this insane optimisation of T5, which knocks 11 seconds off my previous effort and is almost frame perfect:
I did a commentary over the Fun WR and so if anyone is interested, here it is :)
I have found a way to utterly destroy Tricky 7. With the normal strat it took 3 minutes 47.
I don't want to wait around though, so I decided to cut 2 whole minutes off.
My work optimising Tricky levels is now virtually complete! You can watch the playlist here:
https://www.youtube.com/playlist?list=PLOTsjhPmEfJeElqnBUB6wPCj2k98ds9H8
The current estimated fastest completion time for Tricky is 40:11 which is surprisingly quick! This may be subject to further downward revision. Realistically speaking, it would take a lot to remain within the envelope of both no restarts and no significant time loss across each level, so my target for my next WR attempt will be somewhere around 45 minutes, giving me safety margins in deadly levels and accounting for human error.
