Wolfenstein 3D and Spear of Destiny

2012-11-22 06:31:52 pm

The Crate one

I've been working with these games for months, and I have discovered so many tricks and bad programing to abuse to make the game faster to run. The results I have are pleasing to say the least, but I'm sure there's more. The engine I'm using is Wolf4SDL and SOD4SDL to run these games since it is impossible to be able to do anything with this game in Dosbox, I went through the time and trouble to be able to use this engine for my runs.
I am constantly working on these games to find better and faster ways to get through them, and trying harder not to get stuck on some walls, or make mistakes during my runs where I completely forget my way.

Here's an example of some of the things I'm doing during a run. Please note that I am using the Keyboard only to play this game, no mouse is being used.
My in game controls are the standard arrow keys for movement, Left Shift for run, A for fire, S for Strafing, and D for use.

Thread title:

I've been working with these games for months, and I have discovered so many tricks and bad programing to abuse to make the game faster to run. The results I have are pleasing to say the least, but I'm sure there's more. The engine I'm using is Wolf4SDL and SOD4SDL to run these games since it is impossible to be able to do anything with this game in Dosbox, I went through the time and trouble to be able to use this engine for my runs.
I am constantly working on these games to find better and faster ways to get through them, and trying harder not to get stuck on some walls, or make mistakes during my runs where I completely forget my way.

Here's an example of some of the things I'm doing during a run. Please note that I am using the Keyboard only to play this game, no mouse is being used.
My in game controls are the standard arrow keys for movement, Left Shift for run, A for fire, S for Strafing, and D for use.

[url]http://www.youtube.com/watch?v=97PEQwfVcGE[/url] [url]http://www.youtube.com/watch?v=Xqcl8W3KLes[/url] [url]http://www.youtube.com/watch?v=8h6Sc3d4pGg[/url] [url]http://www.youtube.com/watch?v=ml9ubK6apOU[/url]

LotBlind

2012-11-23 01:01:19 pm

Really interesting stuff. Does it take any trickery to move around so fast?

Reaku

2012-11-23 01:15:01 pm

The Crate one

That's called Wall Grinding, it's explained in one of the videos I have put up. Simply, in the game, there are 3 speeds. Walking, Running, then Wall Grinding which is strafe running at a 40-50 degree angle against a wall, it makes you go much faster than you normally would be able too. Another thing is that strafing if a tad slower than doing this, but it's still faster, so I strafe around quite a lot in my runs.

AAA

2012-11-24 01:27:01 am

FYI there is already a Wolfenstein 3D thread: http://forum.speeddemosarchive.com/post/wolf3d_done_quick_team.html
It has runs on the hardest difficulty through episode 5, but has been stalling on episode 6 for nearly a year. Maybe you could help finish it?

Reaku

2012-11-24 09:51:27 am

The Crate one

Quote from AAA:

FYI there is already a Wolfenstein 3D thread: http://forum.speeddemosarchive.com/post/wolf3d_done_quick_team.html
It has runs on the hardest difficulty through episode 5, but has been stalling on episode 6 for nearly a year. Maybe you could help finish it?

I am working on all 6 episodes, though on Difficulty 1, I'm going to figure it out. If they need any help, I can assist in any way.
Other things I may consider doing is learning the lost missions for Spear of Destiny.

CapnClever

2014-09-03 03:38:55 pm

Should've made this post a long time ago, maybe I didn't because I wasn't sure if I'd pick up this game seriously. Now, however, I'm knee-deep in these runs, so it's a good opportunity to revive this thread and maybe some interest too.

For starters, this is what I've already accomplished. It's the first four episodes, single-segment, all on hardest difficulty (I Am Death Incarnate!). Currently my play setup is DOSBox version 0.74, max 100% cycles (the game caps at 70FPS so there's nothing engine-breaky going on): the videos are recorded using DOSBox technology. Make note that these times are based on the in-game timer at the end of each episode, which DOESN'T include the boss level (and that is sometimes very significant), so these can't be immediately compared to, say, Xvirus's listed times in his segmented runs. (In parenthesis is a rough estimate of what the total time probably is.)
Wolf3D E1 4:06 (~4:30)
Wolf3D E2 5:08 (~6:00)
Wolf3D E3 6:04 (~7:00)
Wolf3D E4 5:06 (~5:40)

For those of you who watch all of these, I'd like to mention that these runs were not learned in episodic order, so the skill displayed's gonna be inconsistent if you watch from top to bottom. Each of these is more or less the efforts during a single period of learning and haven't been picked up since. (If you want to watch in learned order, go 1->3->4->2.)

I've looked into Wolf4SDL and, while it's a nice source port, it's not going to completely sync with vanilla. One of the most significant cases I've found to this effect is pwall (pushwall, i.e., anytime you open up a secret) movement: in vanilla you can't move into a pwall's sector until it's completely moved, while in Wolf4SDL you follow the pwall pretty closely. Even desync aside, you're able to move slightly faster by hugging the walls in Wolf4SDL, giving roughly half or a full second per case. I know it's not HUGE, but it gives me pause to play/learn on Wolf4SDL. Believe it or not, I'm doing my best to slowly learn the source code and make a competition-friendly source port (one reason I haven't quite returned to episodes yet). In many ways it'll probably end up like Wolf4SDL, but priority will be given to demo synchronization followed by additional features to be speedrun-friendly, such as adding game times for boss/secret levels, increased time precision, and modern control usage.

My routing is "watch Xvirus, figure out what can be done reasonably, adjust from there". I'll explain more in detail some other time, but if you know what the segmented runs look like and wonder why I don't do the same, it's sometimes because the odds of it going well are far too inconsistent (or at least inconsistent at my skill level). Or, more likely, I messed something up, as these runs are far from optimal.

Should've made this post a long time ago, maybe I didn't because I wasn't sure if I'd pick up this game seriously. Now, however, I'm knee-deep in these runs, so it's a good opportunity to revive this thread and maybe some interest too.

For starters, this is what I've already accomplished. It's the first four episodes, single-segment, all on hardest difficulty (I Am Death Incarnate!). Currently my play setup is DOSBox version 0.74, max 100% cycles (the game caps at 70FPS so there's nothing engine-breaky going on): the videos are recorded using DOSBox technology. Make note that these times are based on the in-game timer at the end of each episode, which DOESN'T include the boss level (and that is sometimes very significant), so these can't be immediately compared to, say, Xvirus's listed times in his segmented runs. (In parenthesis is a rough estimate of what the total time probably is.)
[url=http://youtu.be/HDP0vBDaWBc]Wolf3D E1 4:06[/url] (~4:30)
[url=http://youtu.be/kI-HverX9hg]Wolf3D E2 5:08[/url] (~6:00)
[url=http://youtu.be/D4naBP7crEE]Wolf3D E3 6:04[/url] (~7:00)
[url=http://youtu.be/Oc-SQPulMDw]Wolf3D E4 5:06[/url] (~5:40)

For those of you who watch all of these, I'd like to mention that these runs were not learned in episodic order, so the skill displayed's gonna be inconsistent if you watch from top to bottom. Each of these is more or less the efforts during a single period of learning and haven't been picked up since. (If you want to watch in learned order, go 1->3->4->2.)

I've looked into Wolf4SDL and, while it's a nice source port, it's not going to completely sync with vanilla. One of the most significant cases I've found to this effect is pwall (pushwall, i.e., anytime you open up a secret) movement: in vanilla you can't move into a pwall's sector until it's completely moved, while in Wolf4SDL you follow the pwall pretty closely. Even desync aside, you're able to move slightly faster by hugging the walls in Wolf4SDL, giving roughly half or a full second per case. I know it's not HUGE, but it gives me pause to play/learn on Wolf4SDL. Believe it or not, I'm doing my best to slowly learn the source code and make a competition-friendly source port (one reason I haven't quite returned to episodes yet). In many ways it'll probably end up like Wolf4SDL, but priority will be given to demo synchronization followed by additional features to be speedrun-friendly, such as adding game times for boss/secret levels, increased time precision, and modern control usage.

My routing is "watch Xvirus, figure out what can be done reasonably, adjust from there". I'll explain more in detail some other time, but if you know what the segmented runs look like and wonder why I don't do the same, it's sometimes because the odds of it going well are far too inconsistent (or at least inconsistent at my skill level). Or, more likely, I messed something up, as these runs are far from optimal.

CapnClever

2014-09-15 04:51:09 pm

For anyone wondering about Spear of Destiny, I'll pick it up after I get "okay times" on the six episodes of the original game. In the meantime, however, funkdoc worked on the game for a week or so and managed to get a good time. I'll be using a lot of his run as the basis for what I eventually learn myself. He plays on Wolf4SDL, by the way, which is why some things may seem a bit different.
Spear of Destiny 29:17 (RTA, twitch)

As for myself, I popped an exit in E5. It's not a great run but exiting in E5 is hard enough with all the precise door tricks and disadvantageous combat situations.
Wolf3D E5 4:22 (~4:50)

E6 is really, really challenging. I'll be lucky to get a exit, let alone a decent goal like sub-10.

CapnClever

2014-10-15 12:35:26 am

Dropped my E3 time by well over a minute!
Wolf3D E3 4:50 (~5:30) or, if you want an embedded video which appears to be common around here:

The old single-segment tried its best to mimic Kimo's route, but it falls short in a few areas (level 4 and 7 door tricks are really bad to get with consistency, seeing as you can't hear the doors involved). Not to mention the extremely cautious chaingun pickup on level 8, mostly to not have the end of that level go horribly wrong.

With these things in mind, I planned a new route that grabs the chaingun on level 2 instead. This lets me ignore the shenanigans I did on level 8 previously, while smashing through various other levels that take quite a bit longer without segmented luck. Originally, I just played level 2 "as intended", with the chaingun secret opened early to save a bit of time otherwise waiting on the pwall. Tonight I realized I could probably get the door trick to work anyway, and the result is more or less what you see in the video. There's still a little luck involved, but as long as you only start shooting after nabbing the chaingun and keep your movement tight until that point, it tends to work out rather well.

I could certainly aim for an even lower time (level 3 sucked big time, for instance), although I think 5:10-5:20 (considering all nine levels) is about as good as this route can manage. I reckon it's FAR more consistent that what Kimo pulled off with the machine gun, though. Haven't really tried the level 4 door trick much, maybe it's possible to set up a timing? That'd certainly bring this closer to five minutes flat.

Dropped my E3 time by well over a minute!
[url=http://youtu.be/sAQvKoaElLk]Wolf3D E3 4:50[/url] (~5:30) or, if you want an embedded video which appears to be common around here:
[url]https://www.youtube.com/watch?v=sAQvKoaElLk[/url]

The old single-segment tried its best to mimic Kimo's route, but it falls short in a few areas (level 4 and 7 door tricks are really bad to get with consistency, seeing as you can't hear the doors involved). Not to mention the extremely cautious chaingun pickup on level 8, mostly to not have the end of that level go horribly wrong.

With these things in mind, I planned a new route that grabs the chaingun on level 2 instead. This lets me ignore the shenanigans I did on level 8 previously, while smashing through various other levels that take quite a bit longer without segmented luck. Originally, I just played level 2 "as intended", with the chaingun secret opened early to save a bit of time otherwise waiting on the pwall. Tonight I realized I could probably get the door trick to work anyway, and the result is more or less what you see in the video. There's still a little luck involved, but as long as you only start shooting after nabbing the chaingun and keep your movement tight until that point, it tends to work out rather well.

I could certainly aim for an even lower time (level 3 sucked big time, for instance), although I think 5:10-5:20 (considering all nine levels) is about as good as this route can manage. I reckon it's FAR more consistent that what Kimo pulled off with the machine gun, though. Haven't really tried the level 4 door trick much, maybe it's possible to set up a timing? That'd certainly bring this closer to five minutes flat.

CapnClever

Edit history:

CapnClever: 2014-12-08 12:47:56 pm

2014-12-08 12:47:35 pm

Finally, a good E6!

Wolf3D E6 7:37 (~9:00)

This episode kills you in so very many ways, and the lack of health during some segments (generally the last half of M4 until M7) demands absolute execution. Fortunately for me, I get lucky a few times instead. The last level is unfortunately rather poor, having missed the skip on the first group of enemies, but the rest of the run goes incredibly well (with a couple best segments to boot). The eighth level is usually a huge slog combined with ammo starvation, but most of it went so well that I got an incredibly fast time relative to my average (for reference, two minutes flat is considered good).

And with that all, the episodic runs are all recorded with at least decent times: I've made an improvement to E1 but I just noticed I haven't uploaded that so expect a link in the next update. Of the other episodes, 2, 4, and 5 are the ones left to improve without going completely insane, 4 being the most likely I'll go back to.

Next up is Spear of Destiny! I'm learning the route with RTA runs, and I've already managed a time that improves funkdoc's record by a couple of minutes.
Spear of Destiny RTA 26:41 (twitch)
Sum of best is already 22:23, which I don't think I would have ever expected. Single-segment I hope to see a time of at least sub-25 RTA (unfortunately IGT is far less meaningful in Spear because there are several boss levels, all of which ignore game time: for reference, the linked run is something like 15:40 IGT).

Finally, a good E6!

[url=http://youtu.be/dNuLlxFPj7I]Wolf3D E6 7:37 (~9:00)[/url]
[url]https://www.youtube.com/watch?v=dNuLlxFPj7I[/url]

This episode kills you in so very many ways, and the lack of health during some segments (generally the last half of M4 until M7) demands absolute execution. Fortunately for me, I get lucky a few times instead. The last level is unfortunately rather poor, having missed the skip on the first group of enemies, but the rest of the run goes incredibly well (with a couple best segments to boot). The eighth level is usually a huge slog combined with ammo starvation, but most of it went so well that I got an incredibly fast time relative to my average (for reference, two minutes flat is considered good).

And with that all, the episodic runs are all recorded with at least decent times: I've made an improvement to E1 but I just noticed I haven't uploaded that so expect a link in the next update. Of the other episodes, 2, 4, and 5 are the ones left to improve without going completely insane, 4 being the most likely I'll go back to.

Next up is Spear of Destiny! I'm learning the route with RTA runs, and I've already managed a time that improves funkdoc's record by a couple of minutes.
[url=http://www.twitch.tv/capnclever/c/5640919]Spear of Destiny RTA 26:41[/url] (twitch)
Sum of best is already 22:23, which I don't think I would have ever expected. Single-segment I hope to see a time of at least sub-25 RTA (unfortunately IGT is far less meaningful in Spear because there are several boss levels, all of which ignore game time: for reference, the linked run is something like 15:40 IGT).

Hepe

Edit history:

Hepe: 2014-12-11 07:33:59 am

2014-12-11 07:33:38 am

Rye bread, beer and snus

So I just watched your newest Spear of Destiny PB and it kinda got me wondering, are you aiming just for a decent SS run, or would you by any chance have any interest in doing a segmented run? And could you shave off some time from Kimo's W3D ep6 run by going through the secret exit Doom 2 style in level 3?

CapnClever

2014-12-16 06:11:03 pm

Haven't really thought about segmented before... the main problem I'd have compiling a segmented run is that the tool Kimo used to make his runs, WolfDem, doesn't seem to be compatible with the latest wolf3d version, for whatever reason. (Or I suppose it could be DOSBox, something to look into anyway.) I'm not even sure it'd work with Spear of Destiny! So it wouldn't be the same standard of Kimo's run (nevermind that I play on DOSBox). Many of the strats I've developed are about as risky as you can get anyway, the segmented run would primarily be about optimizing health and ammo.

One day I'd really like to properly provide a source port for this game so that demo recording is more like Doom; for the time being I'll keep to single-segment runs. It's definitely a possibly for the future though (and maybe individual levels as well).