It's midnight and my eyes are bleeding (aka Doom 2 UV Speed run in 14:02)

myk said:
I haven't watched these but I don't recall Sedlo's 16+ minute UV and 18+ minute NM demos being ugly.

Er.. Miguel, this is a BUILT run.

Kristian Ronge said:
Er.. Miguel, this is a BUILT run.

Artistry needn't be lost in the effort, even if it costs a handful of seconds.

I wonder if a bot could be used to automatically find and build the quickest paths between segments (might not look any better, but would be quite a coding experiment). Kind of like a "speed run SLIGE".

myk said:

Artistry needn't be lost in the effort, even if it costs a handful of seconds.

I wonder if a bot could be used to automatically find and build the quickest paths between segments (might not look any better, but would be quite a coding experiment). Kind of like a "speed run SLIGE".

Sounds NP-hard to me. :P

As for artistry... this is about quickness, not beauty, right? :) Besides, to me at least, a perfect TAS run that uses every available trick to get the fastest possible time *is* beautiful, and it's beautiful precisely *because* it doesn't sacrifice time in order to make the run look more "natural". :)

Schneelocke said:
and it's beautiful precisely *because* it doesn't sacrifice time in order to make the run look more "natural". :)

I don't care how big those silicones made her boobs look; they still feel and look like a pair of plastic baloons.

I don't care if they call it "modern art"; it still looks like an ugly mess of lines on a canvas.

Schneelocke: Sounds NP-complete to me. :-)

myk said:

I don't care how big those silicones made her boobs look; they still feel and look like a pair of plastic baloons.

Then stick with natural breasts (and with unassisted speedruns). :P

Kristian Ronge said:

Schneelocke: Sounds NP-complete to me. :-)

NP-complete implies NP-hard. :)

Yeah, well, it implies NP-hard *and* in NP. I'm one of those guys who find the researchers claiming P=NP the most interesting. ;-)

Kristian Ronge said:
I don't care if they call it "modern art"; it still looks like an ugly mess of lines on a canvas.

I like (some) modern art. To me it's more like "photoshopped art" that uses plugins to apply means to obtain effects.

As far as the issue about artistry goes. This run was all about time, not about smooth movement. In fact I've become so use to watching built/tas demos that they seem completely natural to me at full speed. I could've done smooth turns at some points but making it look 'nicer' at the cost of speed is totally pointless when I'm building in the first place, if you want reasonable looking fast runs that's what TAS is for.

On the issue of computer generation that is something I have thought long and hard about. In fact if you look in the text file you'll see that I actually did brute force out a small section to perform a bar glide. However even with extremely limited movement types and no turning there was still 100000 possibilites to try. As far as mathematically calculating routes as doom essentially is a chaos engine this makes it np-complete. So brute forcing under conditions with pass checks is essentially the only way I can think of but I can't see any real way of making it better or faster than doing it by hand except for a few specific situations such as annoying barglides etc.

A very entertaining demo. I would also look forward to seeing more built demos either on doom2 or on another map. If someone redoes this I would be interested in seeing the map15 normal exit. It may be slower than map15+map31, but I don't recall ever seeing it in a movie, so it would be more interesting.

Producing great Built or TAS demos takes a lot of work, practice and patience. But built is probably the pinnacle of hard work, writing every single tic one by one... retrying countless times until you passed this monster or this obstacle w/o losing time... it's OMG... kudos Rambobones.

PS : Hi Jim, been a while I didnt hear from you :)

RamboBones said:
As far as the issue about artistry goes. This run was all about time, not about smooth movement.

In the end it's a display of an optimized run. The most relevant factor is the route; whether all the known time improving tricks and best routes are in the run. Whether any seconds (out of minutes) were shaved off doing extrmely sharp turns or turns with sr50, is secondary.

It's kind of sad some tricks start appearing in TAS or built demos and haven't even been used in real recordings (unlike the recent Map07 pacifist). Although recording those takes really good players, and these are not always active or available, so it can't be said it's anyone's fault.

In fact I've become so use to watching built/tas demos that they seem completely natural to me at full speed.

Heh, from all the built demos I've watched (and many of the TAS demos), they certainly look quite different than when one plays the game.

I could've done smooth turns at some points but making it look 'nicer' at the cost of speed is totally pointless when I'm building in the first place, if you want reasonable looking fast runs that's what TAS is for.

Yes, TAS plus built sections for some super hard tricks could probably make a great "optimal display" demo of this sort. I think those would (as far as manufactured demos go) be the kind of thing speed running sites could host to represent DOOM as a game (in addition to real runs like Radeg's 26:09, of course). At least the DOOM marine wouldn't seem to be epileptic in those.

VinceDSS said:

I wonder how fast a built map15 normal exit would be ... faster than 15 secret + 31 ?

15 + 31 took Rambo 55 seconds. I can reach the normal exit in precisely 55 seconds if I run through the map without any monsters, and that's despite hitting a lot of corners and of course using only strafe40. It's true that there are lots of monsters in the way, especially near the end, but maybe they can be handled with some cleverness (BFG and all)?

Interesting - so it sounds like the normal exit ought in principle to be a tad quicker overall. But note that Rambo also lost some time in 31 by taking the secret exit - the normal exit is 2 seconds quicker (source).

I also wonder (idly) if there might be some tricks to quicken up the normal exit route in map15. After all, it is not one that has been scrutinized at all closely from a TAS or Built perspective.

Maybe it's possible to flip the switch that opens the door to the blue key from below? Didn't get it to work, but someone else might.

Oh, and 0:55 is an outstanding nomo time, Fredrik.

Maybe playing in ZDoom had something to do with it (not that I used jumping or anything like that, though).

Ah, now I see why: the ending section can be played much faster in ZDoom (try it with a doom2.exe-compatible port and compare it!). :-)
0:55 is still a good time, though...

Guys, can someone, anyone turn it into a video file?

Do you have an aversion to using a port that can play it back properly?

HobbsTiger1 said:

Do you have an aversion to using a port that can play it back properly?

I guess this stands for "no"...

I lost my Doom 95 installation when my computer crashed a while ago, and I can't find the CD-ROM. So converting to .avi would be a great service.

That Level 32 could be skipped entirely for a faster time has been brought up ever since the 16'17" run. Please, do so in the next run. No point in wasting unnecessary time.

I can't wait to see this. Level 32 issue aside, this sounds awesome.

So, possible improvements:

-0'01" Level 1
-0'02" Level 31 (normal exit)
-0'11" Level 32 (skip)
------------------------------
-0'14" Total

Brings the possible time down to 13'48". Next run should aim for 13'30" or 13'37" (leet).

skipping Map32 is Compet-N rule allowable... I think Rambo was recording the full game.. so skipping a level doesn't really count.

The issue of skipping map 32 has come up before, in fact I recall being involved in a discussion about Vile's 30nm2956 run where he did not visit map 32. In the end it is at the discression of the replay maker, as this replay was purely a counter to andrey's replay and in his replay he did visit map 32, I decided that I should do the same. One of the goals of this built demo was to attempt to show the levels done to the highest levels of perfection (time wise) possible, so to do that I needed to at least visit every level. I don't know if I will ever attempt to do a better run, if I can find a way to get the zombieman in map 3 to move into place and allow me to exit in 13 seconds then it may be worth remaking for the sake of that trick alone, in which case I would *******ely skip map 32, but I wouldn't hold your breath.

RamboBones said:

I don't know if I will ever attempt to do a better run, if I can find a way to get the zombieman in map 3 to move into place and allow me to exit in 13 seconds

you need a rocket launcher for map05

Unless the same thingrunning trick is possible on map05...

Fredrik said:

Unless the same thingrunning trick is possible on map05...

I dropped a :23 UV run into C-N's /incoming/ a while ago that uses it. I felt bad replacing a perfect Sedlo run with a run that was still rough, but even so...

See -- nothing's impossible in this game.

How do I run this demo?

It keeps giving me errors, and in prboom it screws up.

Exactly which version of prboom? (Full version number please.)

If you are using a current or recent version of prboom, then chances are the problem is your iwad - i.e. it is an old version or it has been modified in some way. What is its exact size in bytes? (Not kilobytes or megabytes, but bytes.)

Vile said:
I dropped a :23 UV run into C-N's /incoming/ a while ago that uses it.

Yeah, that demo is intriguing; funny how the Player uses the Shotgun guy(s) to hoist himself.