Factorio Speedrun in Less Then 4 Hours!

[mikehendi]

Hi guys,

===============================================================

Update again:

Broke the 4-hour mark!

https://forums.factorio.com/forum/vie ... =20#p66038






===============================================================

UPDATE:

I did a new run in 4:44:42 -->

https://forums.factorio.com/forum/vie ... 730#p64730

Video's are online now:

The run itself:
http://youtu.be/3BBr_xtI5do

Commentary video:
http://youtu.be/2fgTJskwDlM

===============================================================

I am working on a speedrun and I just finished a tryout/template kind of run in a very promising 5:23:43. I'll upload a video of the playthrough to youtube soon, if the replay works (I'm still on version 0.11.8, it should work as long as I don't update right?)

I still made a lot of small mistakes and some big ones, especially the end is slow (didn't manage the research efficiently enough). under 5h is definitely possible, maybe even 4:30. We'll see, hopefully I'll get a couple more runs in before the new endgame content is added


Edit: lpw suggested I'd include a statement about settings and such:

Disclaimer: this is a run on peaceful difficulty, with high resource richness. The map is generated with the normal factorio map generator, no custom edits were made. I believe in the speedrun community these rules would fall under a "any%, no glitches" category (in which you can pick the most favourable settings the game itself provides, but not make custom edits or cheat, or use glitches like for example the famous Minecraft duplication glitch.
However, the way this (and likewise, all factorio runs I've seen so far) differs from a traditional gaming speedrun style, is use the in-game timer as a benchmark instead of real-time (The game timer, for instance, doesn't include the time spent in the tech selection screen as the game timer is paused at those time). Also, I had to reload from an autosave at the very end due to an unexpected hilarious event (I'll show this in my commentary video). Likewise, I reloaded from saves two or 3 times when I screwed up big-time (I didn't do excessive reloading, as you'll see there's still quite a few errors in my run, including a big one which costed me 15 minutes or more! I'm thinking sub-4 hours should be possible with excessive reloading) I imagine most (or all) of the runs so far have a few reloads, anyway there is no way to check if this happened or not while watching a replay of a game, so unless you're running live in an official kind of way, one has to assume this happened. I guess that these kind of runs fall somewhere between a live run (where you play live on a real timer in one sitting) and a TAS run (Tool Assisted Speedrun, where you can use emulators, savestates, and basically program all controller input frame-by-frame to achieve the optimal results). I'm quite new to speedrunning, so if I gave a wrong description about any of the above, feel free to point it out.

[-root]

Very nice!

That's an incredible time. I can scarcely believe it and I'd love to see the reply just to confirm you're not using hax.

Goes to show what efficiency improvements can do to a factory!

[Xterminator]

Wow that is quote impressive! I know I could never even attempt to do it in that amount of time.
-root is right, very amazing what an efficient factory can do for ya.

[sillyfly]

Not to sound too suspicious, but - how did you manage to get enough alien artifacts for all the research+modules with so little spawners destroyed?
Or maybe it just looks a little because I like to go on massive killing sprees?

[mikehendi]

Killing aliens was actually one of the weak parts of my game. I mistakenly set alien bases to small, which meant I had to spend 20 minutes laser-turreting small and far apart bases to collect the stuff (you only need ~380, 100 for the modules and 260 + losses for science). I don't know what the average drop amount is for destroying spawners, but I had to take down 55 to get the required amount, which is about 7 per spawner.

I initially thought I would get under 5h, but I screwed up by not automating/supplying enough smart inserters, which meant I had to wait on finishing the RD tech.

I should mention that this is on peaceful and very good richness resource patches (so mines won't run out during the run), but there are no custom edits to the map, it's a normal generated map.

I tried to record the replay at 4x speed, but halfway the repay realy lagged out so I had to interrupt it. I'll try to record at 1x speed overnight, and see if it works out better.

[SHiRKiT]

5 hours! That's quite impressive, I think my best was around 20-25 ish. Congrats!

[mikehendi]

I just finished a new (and for now my last) run in 4:44:42

About 2h in the run I knew this time would probably be quite an improvement, so I didn't bother trying to film-capture the old 5:23 run. Instead I'll make a video on this one, including all the numbers and strategies I figured out. It might take me a few days if I want to include audio commentary and make a nicely edited video. If people are really interested, I'll upload the raw soundless video at 4x speed or so. Also, I'm not the quickest guy with keyboard controls and accurate placement of stuff, so you guys shouldn't have too much trouble breaking this time 4:30 should be possible.

[-root]

That is positively crazy.

Can't wait for the video

[Lurch1985]

You, sir, are a factory building MACHINE!

Well played!

[kovarex]

When we get to stable, it would be best to test if the replays work properly on this.

[mikehendi]

Thanks, I'll do my best to get it up quickly!

@kovarex/devs:
I'm still on version 0.11.8 (hint:no blueprints were used , I didn't update on purpose because that disables the replay ability (even though worms won't shoot you anymore in the newere versions )
Yesterday night I managed to play the entire replay at 1x speed (while capturing it on film), so replaying a game in the same version worked well for me, for what it's worth.

[lpw]

I think when you upload record times like this, you should always include all your settings, especially peacefull and increased resources...

[TediBare123]

This is amazing!
I'm about 25hrs into my 3rd attempt to complete the game and I'm still nowhere near rocket defence, well done!

[mikehendi]

I think when you upload record times like this, you should always include all your settings, especially peacefull and increased resources...

Good point, I've edited the opening post with a statement. Also, I'll talk about this somewhat more comprehensive in the upcoming video.

[mikehendi]

Hi guys,

I've just posted the video's on Youtube:

The run itself:
http://youtu.be/3BBr_xtI5do

Commentary video:
http://youtu.be/2fgTJskwDlM

Hopefully we can get a discussion going about further optimization!

[Koub]

This is totally incredible. Watched the replay ... Wow.
However, if I may criticize, peaceful mode does help a lot. When you care neither the defenses, nor the pollution, you can go really faster.
But I don't deny it, even in peaceful, I'd be unable to such an achievement.

[Trepidati0n]

This is totally incredible. Watched the replay ... Wow.
However, if I may criticize, peaceful mode does help a lot. When you care neither the defenses, nor the pollution, you can go really faster.
But I don't deny it, even in peaceful, I'd be unable to such an achievement.

A speed run is a speed run is a speed run. It is defined as the fastest time to rocket defense based upon baseline configuration parameters given to you by the base application. Hint..SPEED!

But maybe we should define a speed run as very sparse resources, the smallest area, with the least richness, smallest starting area, high segmentation, with largest enemy bases with least richness as well.

[t33]

Simply amazing!

[ssilk]

This was really cool to see.

Some critics (I try to be constructive):

What really surprised me is the number of times you filled up furnaces by hand ... Is there really no way to automate that? It's so inelegant to always run around and filling chests/stacks. I'm with you with the trick for the module production, but I feel, that running around more than three times to fill some chest is in general ineffective.

You used no burner inserters... They are much cheaper and you don't need any wires. https://forums.factorio.com/wiki/inde ... r_inserter

You lost over three minutes by continuously rebuilding the storage tanks. You can use the steam engines to fire that unused stuff: https://forums.factorio.com/wiki/inde ... eam_engine (Using other liquids)

Why didn't you build a simple inserter factory?

[mikehendi]

Thanks for the feedback

Good point on using the heavy and light oil for power, I didn't think of that. Just hooking up a couple boilers and steam engines (does someone know how much fluid/s a boiler uses?) might save time, as I'd have to set up power anyway. It'd have to be a reliable constant source of power though). I'm not sure about the amount of time that is lost though. I had my run planned with fixed number of red and blue assemblers, once those are up I can play the rest of the run somewhat more relaxed, as I just have to do the rest of the stuff within the time it takes to complete 3k blues and 14k reds.

I also didn't look into burner inserters. I'm not sure theyre worth it, even in the furnace areas, which I assume you are aiming at? (in the other areas you would have to put up wires anyway). Most of my furnace areas required a full belt (both sides) of ore, which means I need either long handed inserters (so you'd have to take time to both set up power and feed the burner inserters coal), or make multiple rows of furnaces which requires a lot of tedious/precise belt placing, splitting and joining (which requires more time, and circuits you saved on the burner inserters). I tend to look mostly for easy setups which are easy to remember and fast to place (not sure if my red/green science setup qualifies for that though...)

Inserters are easy to make by hand when you need them, especially if you have gear wheels and circuits on you (I probably should have setup a better circuit producer early on though). If I'd run out it would be slower to run across the map to go collect them from the chests anyway. Also, if you have gear wheels and/or circuits on you, you can craft stuff by hand faster then you can place it anyway. Last, I didn't have huge stacks of materials sitting around, so overproducing one thing can lead to slower production of other stuff. Inserters can't be changed into gear wheels and circuits again. On the other hand, gear wheels and circuits can be made into anything you need on the spot.

I don't have a good solution for the by-hand filling of coal. I kept needing the resources they provided though. I feel it's fine for steel to do it by hand, but I agree continuesly filling up the starting area burners with coal might have been a mistake. Perhaps it's better to add a couple more electric mines and furnaces to the initial setup for red and green science to generate more overflow of iron and copper plates, and use those instead of keeping the burners fed.