Let the transmutation begin! - Rock, Paper, Zachtronics

[AlexClockwork]

Hi there, I just joined the group and am adding people to Steam. I played SpaceChem back in the day (and abandoned it at some point, although, to be fair, there’s lots of games I don’t finish), and just got TIS-100 and Infinifactory in the Humble Bundle. I’m loving them so far.

I’m not all that into optimizing, so that may be why I prefer TIS-100 so far, as just completing a level is more of a challenge, but I guess I may start optimizing if I start to see people’s scores around.

[alms]

I’m not all that into optimizing, so that may be why I prefer TIS-100 so far, as just completing a level is more of a challenge, but I guess I may start optimizing if I start to see people’s scores around.

And welcome! Not optimizing is the key to success, lest you end up like me timing different array deduplication routines, instead of doing actual work, or hunting medals in DG and not finishing the main campaign.

[LTK]

Hi there, I just joined the group and am adding people to Steam. I played SpaceChem back in the day (and abandoned it at some point, although, to be fair, there’s lots of games I don’t finish), and just got TIS-100 and Infinifactory in the Humble Bundle. I’m loving them so far.

I’m not all that into optimizing, so that may be why I prefer TIS-100 so far, as just completing a level is more of a challenge, but I guess I may start optimizing if I start to see people’s scores around.

Premature optimisation is the root of all evil. That said, I remember Helio’s advice about replacing conveyor belts with platforms wherever possible, which allows one conveyor belt to push all of your other products around. It gives you a strong reduction in blocks used while taking a lot more cycles to complete. I’m only a few levels into Infinifactory and it has allowed me to get some pretty decent scores.

My solution on Wave Detection Array is soaring high on the leaderboard in terms of blocks used, but the very next puzzle, Guided Javelin Type 1, completely stumped me. I can arrange the product’s parts just fine but I have no idea how to move the completed product to the output. I could do it if I had a not-gate or a vertically oriented conveyor belt but I have neither of those things. I can do the next puzzle as well, which at least I can make a bit more progress on, but I can’t hack my way into a solution yet.

[alms]

That said, I remember Helio’s advice about replacing conveyor belts with platforms wherever possible, which allows one conveyor belt to push all of your other products around. It gives you a strong reduction in blocks used while taking a lot more cycles to complete.

Grey blocks simply aren’t counted - that is actually in a loading screen tip, and one the loading screen lottery didn’t present me with until many hours in the game, which is why one Rule of Good Design is to never confine important information about your game to a loading screen tip.

One more thing, obvious but easily missed when starting out - there is a reason for having 3 solution slots: it’s very rare (probably impossible) that a solution will excel in all three metrics, while a solution can be brilliant and be excellent in, say, cycle performance, while also being efficient with blocks usage, to go for the absolute best result you need to focus on only one.

This brings me to something I don’t like about the game: we should be given more slots for solutions, so that we can try new stuff without throwing away our previous PR solution in any category.

[LTK]

Well, I completed Guided Javelin Type 1. My problem was not being able to delay a piston so that it pushed the complete product to the output only when the nosecone was welded onto the javelin, because sensing the top block would cause the other components to trigger the piston before the javelin was assembled, and sensing the nosecone in another place would cause it to trigger the piston before it was welded onto the top.

I got the idea of connecting a sensor to a piston that pushed an unconnected platform onto a conveyor belt, which carried it past a sensor and then back onto the block in front of the piston. When the nose cone passes the first sensor, the loose platform is pushed, and is carried to the next sensor five ticks later, which is enough time for the nose cone to drop onto the javelin, at which point a piston pushes the complete product to the output.

Am I going to have to use this kind of trick more often?

[alms]

Generally speaking solving sync problems is something I have found myself doing pretty regularly whenever I didn’t just magically come up with a solution that worked beautifully. I’m pretty sure that, whenever that happened, it was the game subtly pointing me in the right direction than me being super lucky or above average.

As for the details, I think the most skilled player among us who also likes to discuss their solutions is likely to be Gus.

[Heliocentric]

Am I going to have to use this kind of trick more often?

If you are racing the clock yes, else no. you can block the inputs to freeze the flow on any level, and allow yourself a sterile 1 in 1 out workflow which will ever choke, even if only for small parts of the flow. Just avoid things vertically stacking, unless you are lifting things up with a lifter until they are needed I guess, that’s a pretty good overflow stack.

Ultimately, Infini becomes a programming game as much as a manufacturing one.

[Gus_Smedstad]

While I didn’t have to employ any delayed response tricks with Guided Javelin Type 1, in general it can be very handy to create sub-assemblies whose only purpose is logic. I rarely employ delays, but I do frequently create AND gates and state machines.

[LTK]

If you are racing the clock yes, else no. you can block the inputs to freeze the flow on any level, and allow yourself a sterile 1 in 1 out workflow which will ever choke, even if only for small parts of the flow. Just avoid things vertically stacking, unless you are lifting things up with a lifter until they are needed I guess, that’s a pretty good overflow stack.

A surplus of components wasn’t the issue though. If I want to move a stack of five welded components onto a conveyor belt, a piston has to push it there, and pistons only accept input as OR. What sort of sensor setup will activate the piston if, and only if, the stack of components is complete? Any sensor in the stack is going to trigger the piston when a block is falling past it, pushing too early, and any sensor outside of the stack triggers it too early, before the final component has a chance to be welded. I don’t see any alternative solution to mine.

Ultimately, Infini becomes a programming game as much as a manufacturing one.

It’s a Zachtronics game, I expect nothing less.

[Gus_Smedstad]

If I want to move a stack of five welded components onto a conveyor belt, a piston has to push it there, and pistons only accept input as OR.

As I said, you can construct an external AND gate. Not that it’s necessary to solve this, but it’s an option.

My preferred solution to vertical stacks is to send the parts is to assemble them without dropping parts into place. In a 2-item stack, if the top part arrives from the side, the assembly is ready when the top part is in place. You can build any stack from a series of single welds this way.

My 100 cycle solution to Guided Javelin is a bit messy, but you can see this happening in several places.

Center left (largely obscured) fuselage pieces 2 and 3 weld by dropping together in front of the welder - not my preferred solution, since it’s highly timing sensitive.

Center right, segment 1 joins by sliding on top of segments 2 and 3. While I’m again relying on timing, you could put a sensor to detect segment 1, and push the assembly off as segment 1 arrives.

Note that the assembly of the fuselage works regardless of input speed. The buffer area (top left) detects that 3 fuselage pieces are available before pushing all 3 simultaneously into the assembly area.

Left (again somewhat obscured), the nose cone slides on top of the fuselage. A sensor detects when the fuselage is moving in to place, and moves the nose cone as the fuselage moves past.

Bottom right, sensor 1 detects the fuselage falling in to place and pushes the engine below it. 1 cycle later both are in place and welded. Sensor 2 detects that the engine has arrived and pushes the completed assembly off.

The messy bit is that, since I’ve got the input rate set to max, the nose cones and engines tend to overflow. This is unavoidable, because the max input rate gives you 1 engine pieces for every 2 fuselage pieces, not every 3. The design deals with this by throwing excess parts into the void.

[Heliocentric]

A surplus of components wasn’t the issue though. If I want to move a stack of five welded components onto a conveyor belt,

Sensors on pushers, you have the 1st item sensor trigger a pusher which bares a sensor which powers wires but only if the 1item is there too, essentially the pusher and sensor/wire serve as an ‘and’ logic check.

[LTK]

I don’t get it, can you show it in action?

Gus, nice solution, I half-heartedly tried to address the problem with bottom-up construction, sort of like you did, but I don’t think I would have arrived at your method unless I spent hours more iterating.

[Heliocentric]

Anything attached to the pushy side of a pusher, not connected to the ground or anything else will move when it’s pusher is activated, done mindfully, you can push out multiple squares at once.

I’ll go make a demo later.

[Gus_Smedstad]

I don’t think I would have arrived at your method unless I spent hours more iterating.

Well, I did. My original solution was much simpler and slower. Later, I felt challenged by Alms to improve my solution times, so I went back through most of the puzzles optimizing them (though I made a point of only using pieces normally unlocked at that point, no premature lifters). It was entertaining in its own, different way. My take on the early puzzles was quite different once I was more experienced with the game.

[Jurple]

Stuck (for now) in IF’s Proving Grounds. This does not bode well.

In Spacechem, at least I managed to get past the tutorial.

[AlexClockwork]

Wow, that solution is beautiful. Mine was simple but slow, making all pieces wait until there were three of the middle one and then joining the whole javelin in a single place, using the lowest block to push the previous one into a conveyor belt.

I feel like I really should go and try to improve my stuff, but I guess I’ll try to complete further puzzles first.

I’m still more interested on TIS-100, although I admit some of the later levels do look like a bit of a chore once you have an idea about how to solve them.

[LTK]

Alex, if I understand correctly, that’s an ingenious solution, I’ll have to try that.

Edit: Yes! It’s accurate to a single cycle! Thanks for the inspiration, that gives me a solution of only 42 blocks. It’s almost supernaturally fortunate how little tweaking it took to arrive to that solution.

Stuck (for now) in IF’s Proving Grounds. This does not bode well.

In Spacechem, at least I managed to get past the tutorial.

Spacechem’s tutorial was just a basic walkthrough, you had to do exactly what the game told you. Infinifactory’s Proving Grounds are true to their name, the tutorial came before them. Don’t worry, some of them had me scratching my head as well and my solutions aren’t the best, but with a bit of persistence you can do it.

[alms]

Big smile on face seeing more people enjoying InfiniFactory.

[Heliocentric]

Here is an example of a bulk “And” test, 5 blocks that are all sequentially tested and simultaneously required.

https://www.youtube.com/watch?v=Y_4-cjpNkDU

[LTK]

So conduits can be connected by moving them! That’s good to know.

My first solution to Small Excavator involved a blocker and an eviscerator and I didn’t realise I wasn’t supposed to use them yet until later. I was struck by inspiration and decided to redo my solution and it’s definitely my most elegant one yet. The eviscerator is optional, the excess blocks can just overflow to that side, but it was neater just to destroy them. It’s a shame I haven’t found a better way to deal with the slow pace of the block sculptor.

https://www.youtube.com/watch?v=ckXD9-9asE0