UV-Max shortcuts

[40oz]

I love watching UV-Max demos posted here and on youtube. I like to watch how certain people handle certain situations the quickest and most efficient way. A lot of times as I watch these demos, I'm sitting there thinking of even more efficient ways to handle the situations the speedrunner is enduring in the map.

Sometimes I'll see the speedrunner activate a trap with two cacodemons and a hell knight in it. The speedrunner doesn't have any weapons greater than a chaingun so instead of killing all three, he gets the hell knight and one of the cacodemons to fight to the death, and watches patiently for one to die, then when the cacodemon dies, the runner then proceeds to kill the wounded hell knight, then the other cacodemon. Wouldn't it have shaved off a few seconds if the player was killing the other cacodemon while the other two were fighting?

Another common scenario I see is when the runner runs through a huge room crowded with monsters to get to the section of the map with the BFG then come back to take them all out. The runner is already carrying some heavy firepower such as the SSG, plasma rifle, and a rocket launcher. Wouldn't it have been more efficient to shoot rockets at the horde as the player was running to the BFG? Or does it not matter since the BFG kills many monsters simultaneously?

Or maybe it's just unfair to expect players to posess the dexterity to be able to kill monsters and reach the desired destination as fast as possible at the same time. I've been thinking of mapping a bunch of short doom levels that each feature the common scenarios that occur in many Doom 2 PWADs to help investigate the time differences of getting monsters to kill each other versus killing them yourself, or both. Or things like leading monsters into crushers or barrels may be more time-efficient than most weapons would do at shooting the monsters, or it could just be beneficial to conserve ammo for weapons that might be most effective at some other portion of the map.

I know the Doom Wiki features mathematical statistics showing the damage statistics of all weapons and monster attacks, which means routes for playing the map most effectively can be mathematically planned to save the most time. But of course many of Doom's values are pseudorandom. I guess it would be best for speedrunners to always expect that weapons and monster attacks will always do the minimum amount of damage.

What are the common situations that appear in Doom levels that are still up for debate about which is the fastest most efficient method to complete?

[Phml]

Presented as is, that cacodemon/hellknight example sounds plain weird and I would chalk it up as user error.

There are situations in which it's more efficient to grab the BFG and others in which it's better to fight with the weapons at hand, it's not a simple yes/no proposition as distance to BFG, infighting possibilities, actual ammo count, health, monster opposition and just a whole bunch of other variables can play a role.

Additionally, unless you're looking at TAS demos exclusively, you can't just assume every player is going to be a ruthless and perfect machine of destruction. Even the best ones are going to make mistakes over long enough runs, and sometimes just staying alive is hard enough.

Yet other things can seem nonsensical in a vacuum but make sense given specificities that might not be blatant at a quick glance. I would suggest that when you see possible improvements, you try to record demos - either you will be right and you do end up with something better that you can share for everyone's enjoyment, or you will realise there was more to it than meets the eye and will be able to appreciate that specific demo differently.

[dew]

fair questions.

first of all, the cold brutal truth is that some players just don't try hard enough at optimization. they will just max the level at their natural skill level and won't press much further. this is, of course, much more comfortable, doesn't drain your runner stamina and keeps you interested in recording, because you can hop across maps/wads at a steady pace. from the current demo scene, i'd say blob and jario are perfect examples. when they bite lip and press forward, they produce top notch demos, however most of the time they're in cruise mode, because they feel like there's so much stuff to record on. at the opposite extreme you have looper and sav, those guys are ridiculous perfectionists and you will have hard time spotting ANY slack in their demos, but on the downside you get just a handful of demos per month/year from them. what i'm trying to say is that many players KNOW they aren't solving situations optimally, but they don't bother with improving, because the necessary grind is past their pain threshold.

the other issue is route optimization. first max submissions on even slightly non-linear maps will do something 'wrong' and the player doesn't even realize it. then the next guy will tweak something and everyone goes 'oh, of course.' this is most notable on doom and doom2 compet-n max demos. waldon, anders or radeg were world-class runners and some of their stuff is nigh unbeatable, yet you can find other stuff from them improvable by minutes. it takes collaborative effort to find optimal max routes, which also leads to a challenge for the record. that drives people to a) find something new, b) risk more or grind for a luckier outcome. the latter means some people will simply lose interest, because it takes far too long to get everything in the whole run right.

now to your examples:
optimally, i'd probably saw (or shoot) the second caco to death, then proceed to kill both of the other guys if they're still standing. waiting for the HK to finish his job and provoking a second infight is definitely slower and "shouldn't" be done unless the map suffers from severe ammo drought.
in the second scenario, i'd shoot my current weapon as much as i could, optimally to weaken/distract archviles, bait some monsters into provoking infights, etc. you only shouldn't fire if you need all the monsters to infight/not follow you, or it's important to stay silent (eg. easily beat a teleporter ambush at the bfg, because monsters are asleep).

anyways, you're right when you assume it's one thing to want to do things one way and it's something else to actually achieve it. :)

[Grazza]

There was a good quality discussion on Max methods and how to tackle maps for maximum efficiency in one of the early threads in this forum. Some of the posts there shed light on situations where something surprising-looking can actually be a good idea. dew's post above is excellent too, of course.

If the demo where you saw these apparently odd decisions was a compet-n Max on a popular map, then chances are there were good reasons. Maybe to give infighting a chance to kick in, or to avoid the player's path being blocked a little later, or the player was preserving the ammo for a later battle, feeling that it would be better spent that way, and the current room could be cleared just as quickly without firing at that point. Some of the more optimized Max demos are worked out with a high degree of precision: in the text-files, you might read that a player decided to save his last rocket for a later part of the map, etc.

If it was just a random Max on a map that hadn't received much demo attention, then it's possible that the route and strategies weren't very carefully worked out in advance, and there was some trial and error being used. Sometimes previous attempts suggest that a particular approach should work best in one area. But during one particular attempt the player happens to reach that room better equipped than normal, and doesn't adjust his strategy to make use of the unexpected windfall (a hard thing to do when working in real-time). Sometimes these things are obvious when watching a demo (and might be mentioned/explained in the text too), but re-recording in order to reproduce the situation and exploit it better may not be so easy, given the randomness of many maps.

[vdgg]

40oz said:
I know the Doom Wiki features mathematical statistics showing the damage statistics of all weapons and monster attacks, which means routes for playing the map most effectively can be mathematically planned to save the most time. But of course many of Doom's values are pseudorandom. I guess it would be best for speedrunners to always expect that weapons and monster attacks will always do the minimum amount of damage.

No, that's at least not how I understand it. I'll deal with weapon damage, monster damage is just simply what I call "taking risks".
Example #1, in Doom2 MAP07, taking mancos with plasma and arachnotrons with rockets is TAS optimum (two rockets per arachnotron), but in real life the probability of two rocket arachnotron kills every time is way too low, so human optimum demands using rockets/SSG on mancos (three rockets or two rockets + 1 SSG shot on a single mancubus is what we aim for, it is not too unlikely to kill them all this way). Saving rockets for arachnotrons is a "waste" in a sense that you'll end up with killing most of them with three rockets each, 500/3 = 166 2/3, that's a very low damage per rocket.

Example #2, from Tatsurd's Max on Requiem MAP02

In this run, I was able to kill 7 demons with 15 shots of a shotgun,
14 shots is perfect though, needless to say. I killed them with 14
shots only one time, but I made a mistake in the second half, and
the run ended up with 1:04. It was not easy for me to finish the
latter part of this map without shot errors or movement errors.
I think the time is pretty good.

You see, if the run is short, a perfect player should expect the maximum damage and if he does not achieve it, he will just start recording again. Now it makes sense expecting low probability scenario, because it is still reasonably high - you don't spend your entire life waiting for 7 demos killed in 14/15 shots.

What are the common situations that appear in Doom levels that are still up for debate about which is the fastest most efficient method to complete?

Weapon choice.
- Plasma vs BFG (easily accessible plasma and a BFG hidden somewhere. If cells are limited on a map, the general rule is not to use plasma anywhere. However, sometimes getting BFG first involves too much backtracking and plasma is the best choice on some of the early fights)
- BFG vs rocket launcher. In HR2 MAP32 the fastest demos showed us that the "primitive" approach (BFG only) is the most efficient one, this is not always the case.

I recommend watching a few maxes for Evilution MAP17, Phml's is especially interesting as one which differs so much from the rest. You can see that everything can be done with different weapons, in a different order and we (the runners) don't have yet a definite solution which way to handle this map is the fastest one.

[j4rio]

I'd say that question is very map-specific. Nothing that could be debatable in general comes to mind. Also, if you like watching those demoies, why not trying to record something after all? It doesn't have to be anything top notch as plenty of guys stopped (paused?) recording lately.

dew said:
first of all, the cold brutal truth is that some players just don't try hard enough at optimization. they will just max the level at their natural skill level and won't press much further. this is, of course, much more comfortable, doesn't drain your runner stamina and keeps you interested in recording, because you can hop across maps/wads at a steady pace. from the current demo scene, i'd say blob and jario are perfect examples. when they bite lip and press forward, they produce top notch demos, however most of the time they're in cruise mode, because they feel like there's so much stuff to record on.

Actually I'm trying to mix it, getting something done properly while coming up with regular first (2nd at best) exits. Another cold and embarassing truth is that I suck at optimising. :P I really don't want to tell how long I spent on some stuff that passed as another of my supposedly random first or second exits. (et12 took me 3 weeks to get... just saying)

[Looper]

j4rio said:

(et12 took me 3 weeks to get... just saying)

Well, I'm pretty sure you didn't play 3 weeks constantly without sleep etc. so your metaphor doesn't tell anything :) It could be 2 hours or 50 hours. I'm pretty sure if you would concentrate on that level for a single day (24 hours) with only tiny breaks you would finish it with ease :p

When I try to improve max record, first I'll just stick to the current record's route, because it's pretty/very good 'template' for route. After playing the route few times, you remember how many rockets/cells you have left at each situation. Then it's easy to swap the use of BFG to more efficient situations. After that it's just shooting rockets with good splash.

There's so much to say about max routes because there are pretty much infinite amount of different situations, but IMO in most maps best tactic is to exit the level with 0 cells and 0 rockets :)

Phml said:

Additionally, unless you're looking at TAS demos exclusively, you can't just assume every player is going to be a ruthless and perfect machine of destruction.

Hey! Not even TASes are perfect! Not even UV Speed ones built! It's optimal/perfect when intermission screen says 'TIME :00' unless negative time is possible to reach :))

[j4rio]

Yep I figured it afterwards, although I'll keep it at that as any hour count would be just a very wild guess... so I'll you wonder ;). Sometimes those hours are just over before you know it. Besides, how the hell could some of you guys go on for 24 hours with little pauses is beyond me :P. Don't you, like, get entirely sick just of the view you get from starting screen?

[Looper]

j4rio said:

Besides, how the hell could some of you guys go on for 24 hours with little pauses is beyond me :P. Don't you, like, get entirely sick just of the view you get from starting screen?

Probably nobody plays 24 hours in a row (excluding korean SC-players and I don't know about SAV :)) unless there's a event or a REALLY must-do thing.
(End of derailing thread.)

dew said:

what i'm trying to say is that many players KNOW they aren't solving situations optimally

Yeah. It's really hard to solve maps in optimal way because some of the things works way too seldom, so it's better to use tactics which have the best expected value to reach one's goals in reasonable time.