HL2 TSA

[CptCool]

Me and typical have decided to do a TSA run of hl2. This is just a side project we're both doing. Any thoughts?

[Inexistence]

No. No thoughts.

[CptCool]

We have finished this run. Thank you all for your contribution.

/troll

Anyways we're gonna start this really soon, we won't post new topics for each map itself, just each chapter and have the maps inside that topic.

[typical]

Heheh. Will be running on hard and will BEAT every time that hl2dq got.

[Foxtreme]

I expect a sub 1 min run. GL !!

[CptCool]

Typical already estimates 30 minutes deduced.

[typical]

I do?

[Inexistence]

You do.

[typical]

I'm actually amazed at how good the HL2DQ team were at finding shortcuts in HL2. They were incredibly through in comparison to Ep1. Then again Suga was technically on his own.

[lowerlogic]

I would love to help on this TSA run in whatever way possible. I'm not sure if I'll be able to compete on the same level when it comes to actually running it, though I'd be willing to give each map my best shot. I could explore and test all sorts of creative routes and make sure the TSA run is using the fastest possible routes. I also have some basic audio and video editing experience from putting together my crowbar-only speed run of hl2, so if needed I could probably take all the demo files of the TSA run and assemble it into a single video when its done.

This post has been edited by lowerlogic on Oct 31 2009, 05:50 AM

[CptCool]

Sure, but you'll be the last person we'd allow in to help. The smaller the group, the better. Add me to steam, CptCool4.

[typical]

Any input about route or otherwise would be greatly appreciated. The crowbar only run was pretty good by the way.

[lowerlogic]

Thanks typical .

Here's a thought/observation:

Playing at a time scale of 0.2 makes dialog scenes, "waiting" scenes, and each map overall much longer to run, so I'm open to the concept of segmenting this TSA speed run more than hl2dq or other speed runs of hl2.

For example of what I'm talking about: In hl2dq the first map takes a time of 2m:40s to complete at normal timescale. So at an 0.2 time scale, it takes about 13m:20s to complete an attempt for the map (assuming we manage to teleport the NPCs perfectly), which means it takes about an hour to get just 4 full attempts in. I've tried changing the time scale part way through the run (so the initial train part and the later dialog part with Barney are played through at normal speed), but that also alters the demo playback speed (even when I also set demo_timescale correctly).

So I think we could split the first map, and many other "long" maps, into a few extra parts. Perhaps part 1 of the first map could end right when the train doors begin opening (so we don't need to wait 3 minutes riding the train each time), part 2 could end when the combine soldier opens the door to the hallway, and part 3 could be the rest of d1_trainstation_01.

Segmenting at well-defined points like doorways makes it easier and more fun to run, but also will probably make the video of the final run less smooth at the transition points from one segment to the next, so there is a trade-off. I'm interested in hearing your and CptCool's thoughts about what we should do.

[typical]

Since the timescale switching while recording seems to screw up the playback I figure that we should make a segment for each time we change the timescale. For example: in big open area's I usually use 0.7 while in tiny finicky area's I use 0.1!

As for cut-scene's. It's much easier to teleport npc's everywhere when viewing them slowly as you can see exactly when they start moving and therefore teleport them very successfully.

[Inexistence]

alias time "host_timescale 1; wait; time"; time

Put that command into the console. It'll keep timescale at 1 constantly no matter what. I'm pretty sure it works for the demos. If not, replace host_timescale with demo_timescale.

If you want to change timescale again, use the command:

alias time ""

This post has been edited by Inexistence on Nov 2 2009, 05:04 PM

[typical]

Thank you very much for the fix. Now we won't have to segment as much/ at all for most maps.