HR2 TAS Movie

[Grazza]

How true. The excitement of watching demos can't have anything to do with any doubts about whether the player reaches the exit.

Although...

When I first watched Selim's ph22x528, I didn't know if it was going to play back OK (I watched it using Doom2.exe, and it hadn't been verified to work with that). It made it really exciting to watch, as I genuinely didn't know whether he was going to reach the exit. I kept thinking it must surely have desynched, but each time he somehow survived.

[Volte]

True, but that can happen with TAS demos too. :) In fact, TAS demos are probably even less likely to play back alright in any given port.

[Grazza]

ph22x528 is a TAS demo (first pacifist on HR map22 - this remains undone non-TAS). I couldn't get DosDoom to work at the time, so I had nothing better than to try to watch it with Doom2.exe.

[myk]

VolteFace said:
Generally when you watch a demo of someone completing a level, you know they won't die no matter what anyway.

Well...

[Donce]

Yeah, some of them desync with doom2.exe when converted, for example my demo of Nulspace only plays back with TasDoom only.

[cyber-menace]

Sure, professional demoes don't have the player dying before he reaches the end, but what about the demoes from the little people. Sometimes they can be entertaining, and sometimes they even die and try again for some odd reason...

[Schneelocke]

And then there's those demos of levels with a boss brain where the player dies but later on finishes the level anyways when the boss brain is killed by infighting monsters. :) I recall watching one recently of Grazza performing this stunt in a rather horrible level, but I don't remember the thread he mentioned this in right now.

[Grazza]

Well, VolteFace's use of the word "die" was a little careless, but I chose not to nitpick, since his basic point (that when you watch a demo, you already know in advance that the player will definitely reach the exit, barring a desync) was good.

Demos where the player dies in the process of exiting are of course quite standard. For instance, in a TAS speed movie on HR.wad, the player would almost certainly be at zero health at the end of map09 and map23, and maybe map30 too.

If you've seen a demo where the player dies and then respawns/restarts the level, then either:
1) It's a DM demo :)
2) It isn't a valid demo.
3) It is a recording of an entire session of playing.
4) It has desynched (maybe you're watching it with the wrong exe).
or
5) It might be an old coop demo; I recall that in the H2H contests, respawning after dying was considered OK in coop.

[Angus Thermopyle]

Grazza, you're right.

Look at Scythe Map10 and Map20 for example. The player cannot finish these maps without dying.

[Searcher]

On This page from Grazza:
http://www.doomworld.com/tas/lmps.html
There is a link for tasmbf, I would love to have it.
ftp://ftp.doomworld.com/tas/tasmbf.zip

The link is dead. 3D hates my system so can't go there.
Anyone have a link for tasmbf.zip I can grab?

[Schneelocke]

http://www.3ddownloads.com/doomworld/sda/tools

[Searcher]

Schneelocke said:

http://www.3ddownloads.com/doomworld/sda/tools

thanks for trying but:

3d downloads hates my system. I wait in line for 30 to 45 minutes and then get a failed to download error message.

anyone else?

Edit:
OK, got it From Donce. Thanks!

[Donce]

Searcher said:

Anyone have a link for tasmbf.zip I can grab?

Check your PM.

[OnyxStar]

Wow, a lot of replies in this thread, but I was hoping for some more feedback from viewers, either positive or negative. There are some things I would like to address though, namely:

AdamW said:

cyber-menace: um, you do know what TAS means, right? there's no such thing as luck in a TAS demo...and not much skill, certainly just running around and not dying isn't hard in TAS at all.

Before I break this down, I of course realize that I could never, ever perform on my own as good as this demo illustrates. But, to say that there was no luck or skill is something I completely disagree with.

Luck: I can't count the number of times I relied on luck to complete this. I don't control the monsters in any way, and rely heavily on luck when running between several enemies that could take me down with one hit.

Skill: I suppose the player just stumbled onto the correct routes through the level. I don't assume to be the leading example for the fastest way through each level, but I came up with all of this on my own, and am rather proud of several ideas I had.

Of course I am not going to die anywhere in the demo, that would miss the point completely. Although, I suppose if you were to get technical, you might watch the end of level 30 again... I must say though, I died many, many times while creating this. Just as anyone making a real speed demo does, and they restart and go at it again.

All that being said, I completely agree that all good, clean, non-cheated demos take much more skill than this. I get to start recording wherever I want. Of course, I must perform correctly to get to where I want first. What I love about TAS demos is that you never have time to breath. I just watched the new run through episode 4 of the ultimate doom, and I noticed at some points that my heart started beating faster, trying to keep pace with the mind bending speed. Is it real? No. Is it fun to watch? Yes.

I've seen a good amount of TAS demos, and I decided to contribute, instead of just sitting back and hoping others do it for me. My demo is nowhere near the perfection of other TAS demos, but I will pat myself on the back for the fact that it is fun for even me, the recorder, to watch.

If anyone else has some comments on the demo, I would love to read them! Thanks.

[Searcher]

OK, Back on topic for a moment.

I for one really enjoyed watching the utter chaos in those runs. So, I say, damn fine job!

I have watched them at least 15-20 times each and am quite impressed with the tight spots and near death experiences in the demos.

God knows I don't quite have the patience to die 50 times or more making one of these so have to applaud the guys that will take the time to do it. I give up to quickly for the most part, after only dying only 20 or 25 times.

So, thanks.

[VinceDSS]

In the case of Compet-N demos (which are not cheated), some demos needed months and a thousand tries...

[Grazza]

Simply put, TAS demos and non-TAS demos are two very different art forms. It doesn't make sense to try to compare them directly.

A top-class TAS demo involves a great deal of work (see, e.g., 30uv1617.txt), and so does a first-rate compet-n demo (see, e.g., lv24-036.txt). The type of work involved is rather different though.

I see a parallel with the difference between over-the-board chess and correspondence chess. Both require excellent skills, but one is all about what you can do on your own, in real-time, and the other depends on how perfectly you can play with plenty of time to eliminate errors and with all manner of automated assistance.

[AdamW]

my comment on the movie is above - i watched the first two maps, which was enough to show me it's nowhere *near* as optimised as a TAS demo ought to be, and the horrible turning makes me want to vomit. there, glad you asked? :)

[OnyxStar]

AdamW said:

my comment on the movie is above - i watched the first two maps, which was enough to show me it's nowhere *near* as optimised as a TAS demo ought to be, and the horrible turning makes me want to vomit. there, glad you asked? :)

I hope this won't turn into a little battle, but I must point a few things out. Nowhere *near* as optimized huh? I will grant you that map 2 isn't that great. I was still learning the ins and outs of TAS demo recording at that point. This isn't a 'professional' TAS demo where I shave off every last nanosecond. It's just supposed to be fun to watch, and still have the levels completed quickly, which they clearly are. As for map 1, I think it's pretty well optimized, considering I don't use Strafe 50 at any time in the demo, and I also now realize that I was using forward 49 tics since prboom was programmed with them, instead of forward 50.

Onto the turning aspect. I agree with you to a point. It is a little annoying at first, but I find you get used to it. Every other TAS demo I have seen has this too, and it's just what's going to happen when you record at half speed. Based on what you've said, I think it's safe to assume you've seen other TAS demos too, and I don't recall you whining about the turning in those. If you have, that's my mistake.

Finally, the fact that you only watched 2 maps, and have these complaints, while others have enjoyed the whole thing, is very interesting to me. Oh, and yes, I am glad I aksed. :)

[Schneelocke]

Maybe AdamW could do a HR2 TAS movie to show us how he would like it done?

(Note: I'm not proposing this because I think he couldn't do it; I'm just genuinely interested in seeing what he would come up with).

[xit-vono]

Well I watched the first 15 maps of your demo and I found it rather boring. There were many parts where you shot at monsters where it would be possible to simply pass them. I consider the TAS movies by Andrey and Sedlo to be much more impressive. Also 30hr by Yonatan is fun to watch and demonstrates what a TAS demo should look like on more difficult maps. Still hopefully you will improve and make some demos that are more worthwhile to watch.

[Opulent]

I think we're all in agreement, just we are looking at this from a different pov.
OnyxStar recorded this demo because it wasn't done yet... setting out to record a TAS-quality speedrun that is entertaining to watch.
This has been done successfully. Any demo I don't ever take my eyes off the screen is clearly well made; and I really enjoyed this one personally.
On the other hand, it is not Sedlo-fluid or PlanetAdam-Compet-n impressive... but how many demos are?

[myk]

Schneelocke said:
(Note: I'm not proposing this because I think he couldn't do it; I'm just genuinely interested in seeing what he would come up with).

I'm more interested in his real pacifists and speedruns.

[Grazza]

I think it was very impressive for the first demo that OnyxStar publically released (I think it's his first - right?). Certainly a very difficult and exceptionally ambitious project to start with, and he had very few pre-existing speedruns to get ideas from.

TAS is not a magic wand. Just as you can't expect someone's first ordinary demo to have perfect flowing movement, so you can't expect this when it is a TAS demo.

Some of the original TAS demos were not at all time-optimized, particularly where they were taking on undone tasks.

As for jerkiness of movement, maybe Sedlo's TAS demos have spoilt everyone...

I certainly look forward to more demos (TAS or non-TAS) from OnyxStar. You're obviously aware that people expect more strafe-50 and wallrunning to be used in TAS demos, and would like to see smoother movement. You might also consider using a method to record it in sections, TASMBF-style - AFAIK all previous TAS demos of anything like this length used such a method. Without additional coding work, that might mean sacrificing Doom2.exe compatibility though. Swings and roundabouts, I suppose.

[entryway]

If you did not know, but first of all such runs are recorded for themselves.
I do not know how many times each of you has looked my runs, but I them have looked everyone on 300 times. They like me and I love them. And OnyxStar loves the run which he has made. And if someone does not like this run - simply do not write here anything. Because it is not pleasant for the author.

[Fredrik]

He asked for people's opinions. While I agree that bashing doesn't lead anywhere, constructive criticism is always good.

[AdamW]

schneelocke: I don't like playing TAS, it doesn't feel like fun, more like work - there's no excitement in it. I'd rather go do some stuff for that Vile wad, which I'm gonna try just now, or have yet another go at getting e1m7 back :). I made a TAS demo to prove the fast route for n2m5s was possible once, think that's the only real TAS I've done. I didn't mean to sound excessively harsh on this run, but IMHO if you're going to make a TAS demo you should make it something really impressive. I'm watching it through again and onyx just doesn't play with the best style for TAS - in TAS you should play as if you're God. You shouldn't kill anything that isn't immediately blocking your path unless you can do it on the run, you shouldn't accept taking any damage that isn't damn near inevitable, and you should above all always look as if you know *exactly* where you're going and what's going to happen next, because in TAS, you effectively do. Onyx's demo looks like fairly tentative uncheated play - he plays as if he doesn't know what's around the next corner, and plays to minimise damage and get through situations "safely", at least in a lot of places. I only really see the point of watching a TAS demo when it's something you'd never try in an uncheated run...

[Andy Olivera]

Alright, I finally got around to watching this. It's not perfect, but it certainly isn't bad. When you consider this is the author's first TAS demo, it's actually pretty good. It's got all the same mistakes as my first demos had, but it's quite a bit more ambitious, too.

It showed me quite a few things I missed while I was speed-proofing the levels. For example, MAP21 can be done without waking up a good portion of the teleporting monsters. In MAP26, you didn't have to visit any of the side rooms in the Red Key area. MAP14 didn't require a visit to either of the two side teleporters. The switch in the big Cyberdemon/Staircase room can be avoided in MAP15.

Here's a couple suggestions for future demos. First, never miss! Not only does it speed up the final product, it looks more impressive.

Second, along those same lines, kill enemies in the minimum number of shots. Sometimes the RNG can be in the wrong place, making this impossible, so I'll make the wasted shot up on another enemy in the same battle. It doesn't look quite as good, but there's no loss in speed or wasted ammo. In the case of the two rocket Arachnotron kill, I generally ignore this because it's so rare. One BFG Spiderdemon is far less rare, but sometimes it just refuses to work. One Berserk Demon/Spectre kills are about the same. Everything else is pretty much expected.

Third, always be prepared for what's coming. If something surprises you, try again. The example that caught my eye was on MAP19 when you're on the lift and you suddenly turn around to shoot the Chaingunner that was just revealed behind you.

And finally, here's my overall TAS philsophy: make it look as much like a real demo as possible, just without mistakes.

Oh, before I go, I have a complaint about a comment you made in your text file:

I built a way to skip ahead to the level I want, but it still has to calculate what happens on the way, even if it doesnt draw it. Even with my 3.0 Gigahertz computer, a half hour of demo takes about 15 seconds to run. When you are constantly changing a demo to get it right, it takes a lot of time.

Fifteen seconds to go through half an hour of gameplay? I don't want to hear ANY complaining! Do you know how long it would take TASDoom to fast-forward through that? Three minutes! I have to wait thirty seconds just to fast-forward to the next part of a five minute demo, so you have it freakin' EASY, man! :-)

[xit-vono]

Well I think Andy olivera's comments are very good here. One more comment I have is this, don't wait to kill monsters unless you have to. In a wad like hr2 you might have to make an exception to this because of the difficulty of the map, but still you killed too much stuff. Also make sure you do things like kill mancos in 3 ssg or rockets, and cacos in 2 ssg or rockets. I have a higher standard than Andy Olivera about weapon damage, so for me I do cacos in 3 punches, hell knights and arachnotrons in 3 punches, and mancubi in 4 punches. Theoretically a caco can be killed with 2 punches but I don't bother because the damage must be perfect both times. But I wouldn't worry about that kind of thing if I were you. JUst make it look clean and don't waste time.

[Opulent]

As long as no time is wasted, shouldn't matter how much shooting is done.
I'm sure the goal here was to make a satisfactory TAS demo; and that has clearly been accomplished.