Factorio Forums

Hurkyl wrote: Thu May 29, 2025 2:21 pm IMO the plastic requirement is the limiting factor here.

If the goal is just to have "blue circuits" of high quality you don't necessarily need plastic of high quality ! If you use something like this :

viewtopic.php?t=125125

Hurkyl wrote: Thu May 29, 2025 2:21 pm I think the most competitive alternatives would have to be

Competitive in what ? what thing are you gonna measure to say which is the "most competitive" ? Because it seem to me you are now discussing "how to make legendary plastic" , but that's not necessarily useful for "blue circuits" or LDS , the "objectives" ( which were themselves already not precisely defined i have to say it's not only on you , i'm blaming Tertius too there x) ! ).

I tend to think when it's not precised that it's a competition about being the funniest, so i'm going to continue but i'm sure everyone can have their own interpretation .

Now if you want to talk about setup to make legendary plastic, you can also do in Gleba ! It's very competitive in terms of ressource efficiency, similar to asteroids, because it's coming from plants and they are infinite and free, so it's infinite legendary free plastic with no risk of running low on input !

Hurkyl wrote: Thu May 29, 2025 2:21 pm running plastic through a recycler directly. That's super lossy, but I can imagine circumstances conspiring to make that preferable.

That's lossy like asteroid upcycling ? What do you mean by lossy ? if you lose something that's infinite and bound to spoil anyway ?

I'm a bit provocative maybe, but when i see "what is the better?" or "the most competitive" i'm still having the impression that i read ( both you and Tertius to a certain extend ) players talking about "the best potato" but you can have "the heaviest" , "the largest" , and "the tastiest", which are "more objective ways" of defining things ( one being a personnal preference). That's why i'm asking "what do you mean?" all the time ! or "lossy in what ? plants ?".

The cheapest to build for a given amount of production per minutes at the end ? the smallest footprint for the amount of production per minutes ? The easiest to setup early in game ? the one that pay for itself the fastest so you can snowball ? the ultimate setup you build and then you have achieved your objective you can't run the save on your computer anymore, really this time you have achieved the most SPM possible ?

==> example :

Hurkyl wrote: Thu May 29, 2025 2:21 pm My current thinking is that if you do have an alternate quality plastic source that it would be better put towards superconductors than blue circuits, but it would depend on your needs or if you prefer a different method of getting quality superconductors.

So the better could be just the one you prefer ? or one based on your need ? either one ? That makes the term "better" pretty useless to qualify any setup then imo, let's just forget the "one you prefer ?" because that's endless, people have different opinion, what's left is ===> IT DEPEND ON YOUR NEEDS ! Indeed ! That's the number 1 thing you need to explain and have a precise definition, but most of the time players don't really defined their needs, that's my point, they want advice on a setup or "what is the best method", but really it depend on what is your goal ! ( what is the 'best' potato makes no sense, it depend what you gonna use it for and what you prefer too).

I believe when player uses shortcuts like those ( the best / the more competitive without precision ) in their explanations it is bound to create misunderstanding because players would value different thing as being the "best", so they would argue about their personnal preference, and not about something that can be measured and on which comparaison would be rationnal and helpful. ( imo).

Feel free to explain why my setup is "not the most competitive" x) if that can help you define more precisely what you mean by "most competitive" in a measurable way. It's a setup that makes legendary plastic from plants forever, that was my goal, infinite inputs, but i don't think it's good for making legendary blue circuits ! x)

It's very expansive, require a lot of machine and ressources invested to build it and UPS to let it run but then you have infinite ressources. All and all very similar to asteroid mining no ?

mmmPI wrote: Thu May 29, 2025 8:03 pm

Hurkyl wrote: Thu May 29, 2025 2:21 pm IMO the plastic requirement is the limiting factor here.

If the goal is just to have "blue circuits" of high quality you don't necessarily need plastic of high quality ! If you use something like this :

viewtopic.php?t=125125

Congratulations, you have repeated both what I said and what the person I was replying to said.

The point I was explaining is that you have to grind quality on something with plastic in its production chain -- a red circuit derivative or an LDS derivative -- or recycle the plastic itself.

Or, I suppose, cannon shells. Seems an unlikely choice but maybe there is some world where that makes sense. (Or, maybe the tesla gun/ammo? I don't remember if I ruled those out because you can't get plastic back, or just because it eats holmium)

Anyways, the very sentence following the quote you snipped out of context is precisely opining that upcycling blue circuits is likely the best option of the red circuit derivatives.

Hurkyl wrote: Thu May 29, 2025 8:48 pm Congratulations, you have repeated both what I said and what the person I was replying to said.

I believe it's not the case, it seem that you are misunderstanding, and/or omiting a lot of the ideas that were mentionned, or have in general a very limited assement of the situation. Have you missed that I also posted a picture of the "best setup" according to "ressource input" for demonstration purposes ?

Hurkyl wrote: Thu May 29, 2025 8:48 pm The point I was explaining is that you have to grind quality on something with plastic in its production chain -- a red circuit derivative or an LDS derivative -- or recycle the plastic itself.

You said other wrongs things, and in this reformulation , you are adding one more imo, because it depends on your goal, as you said earlier :

Hurkyl wrote: Thu May 29, 2025 2:21 pm [*] You have plastic coming from your asteroid upcycling operation[/list]

If you do have such thing, which i doubt you ever done now because then you'd know you could upcycle carbonic asteroid to legendary, so you have legendary carbon and sulfur, and you make legendary coal from that to make legendary plastic, you do not ever need then to grind quality on something with plastic in its production chain, as you said . Carbonic asteroid do not contain plastic. No, you grind something that is a raw material for plastic.

That's just a detail that doesn't adress the big problem imo that the whole reasonning is baseless if the quest is attempting to find the "best most competitive setup" before defining what you think should be considered valuable in the definition of "competitive". Hence why i posted my setup which is not an imaginary situation.

Why isn't it the best setup ? how do you compare that to asteroid upcycling ?
No wonder you didn't answer in retrospect ......

Once you have a proper understanding of the different solutions, then you can measure different builds, then you can choose the more appropriate for your objectives, as always , don't forget the very basic comon sense x).

mmmPI wrote: Thu May 29, 2025 8:03 pm I'm a bit provocative maybe, but when i see "what is the better?" or "the most competitive" i'm still having the impression that i read ( both you and Tertius to a certain extend ) players talking about "the best potato" but you can have "the heaviest" , "the largest" , and "the tastiest", which are "more objective ways" of defining things ( one being a personnal preference). That's why i'm asking "what do you mean?" all the time ! or "lossy in what ? plants ?".

When you study software design or just good programming, you learn it's important to use the proper algorithms. If not, you're wasting resources. For example, if you often need to find some value in a list with 1024 entries, a very bad algorithm is to just search item by item until you find it. If you need to search often, your program will spend all the time with that (operations=1024/2=512). Instead, you could store the list sorted, then use binary search. This will use a fraction of CPU time (operations=log2(1000)=10). Or you could compute a checksum (hash) of the values, then use the hash as index into an array and just access the array element without any search (operations=1). Yes, the last algorithm needs 1/500 of the effort of the first.

It's the same with upcycling Factorio items: with the wrong upscaling "algorithm", you spend all your resources and don't get much output. But with a better approach, you build the same amount of machines but use much less ingredients for much more output in much less time.

This is what I'm trying to do: find a good, if possible the best, algorithm. I don't want to waste resources. I don't want to build a 1000 tile factory and see a 100 tile factory in the next screenshot with double the output. And since there are so many approaches with so many starting points, I want to profit from the experience of other people. I don't want to copy a blueprint, I want general information about the general approach.

You already sent me on a fool's errand by mentioning "use legendary bacteria". That's junk. I wasted an afternoon trying to work out how to do that - in the end it's dependent on legendary bioflux, which cannot be crafted efficiently. You don't need a huge factory for legendary bacteria, but you need the huge factory for legendary bioflux. Getting rid of all the unwanted non-legendary byproducts is the real challenge and the real waste. I didn't try but simply washing lava-to-molten and molten-to-plate with 10 foundries will probably have the same yield of legendary plates. Or the other methods I mentioned and already built.

mmmPI wrote: Thu May 29, 2025 9:47 pm*snip*

Congratulations. After observing I did not put enough caveats in my statements to exclude people who are obsessed with modules in lava to molten metal recipes or roleplaying as bees and wouldn't have the awareness I was not talking to their goals, you caught me failing to make a sufficiently encompassing statement about ways to obtain legendary plastic (and derivatives) and got to pretend I'm ignorant about it despite explicitly mentioning the omitted method. Are you proud of yourself?

Tertius wrote: Thu May 29, 2025 10:35 pmYou already sent me on a fool's errand by mentioning "use legendary bacteria". That's junk. I wasted an afternoon trying to work out how to do that - in the end it's dependent on legendary bioflux, which cannot be crafted efficiently. You don't need a huge factory for legendary bacteria, but you need the huge factory for legendary bioflux.

As an aside, I've thought about this myself in the past, and I think rare bacteria would be more practical -- I was estimating that you can probably get enough rare bioflux to sustain it from the quality rolls on mash/jelly recipe and on the bioflux recipe.

A sufficiently large operation probably could keep legendary bioflux commong randomly just with the natural rolls, but rare seemed more likely to be achievable with more modest operations.


Or a more complex solution would be to have the infrastructure to allow bacteria of all qualities spinning. The higher qualities would stall out when you run out of bioflux, but since the common loop runs constantly, it can provide the uncommon bacteria to restart the loop when uncommon bioflux comes in, and so on up the line. Maybe wait for a few bioflux rather than just one.

Bioflux keeps for long enough that I expect it to last until you can produce a new high quality bacteria to pair it with.

(disclaimer: I have not done much to compare its effectiveness versus other approaches)

Of course, this approach assumes you're prepared to work with ore of multiple qualities as input to further quality refining operations.

Tertius wrote: Thu May 29, 2025 10:35 pm

mmmPI wrote: Thu May 29, 2025 8:03 pm I'm a bit provocative maybe, but when i see "what is the better?" or "the most competitive" i'm still having the impression that i read ( both you and Tertius to a certain extend ) players talking about "the best potato" but you can have "the heaviest" , "the largest" , and "the tastiest", which are "more objective ways" of defining things ( one being a personnal preference). That's why i'm asking "what do you mean?" all the time ! or "lossy in what ? plants ?".

When you study software design or just good programming, you learn it's important to use the proper algorithms. If not, you're wasting resources.

[...]
This is what I'm trying to do: find a good, if possible the best, algorithm. I don't want to waste resources. I don't want to build a 1000 tile factory and see a 100 tile factory in the next screenshot with double the output. And since there are so many approaches with so many starting points, I want to profit from the experience of other people. I don't want to copy a blueprint, I want general information about the general approach.

I think i sort of understand what you mean, but feel like maybe i wasn't clear in my explanations :

You could be optimizing for the CPU speed, or you could be optimizing for the amount of memory the algo need to handle the task. And even in those, you can optimize the CPU speed to be the most stable over the course of the task ,or the least amount of time before completion, or you can try to find a version that is neither the "best" in CPU speed nor "the best" in memory cost, but is a trade off that is somewhat in the middle because you personnaly value speed vs memory a certain way, or you want constant production, or you want the quickest way "on average" or "no average".

That's the same for "quality" to me, if you search for "the best sorting algo" that's like "the best way to upcycle", it depends. When people say "this setup is the best or is better" , without precision it makes no sense to me, so i ask, and when i hear as answer "least amount of input ressource" and "least amount of machine" it's not making sense to me , because if you want to save on input ressources you use things like productivity modules, but that can have as a consequence to require "more machines".

It's pretty much always possible to find a situation where you have setup as follow :

Setup A takes 10 input for 100 output per second, cost 1000 to make, composed of 5 machines.
Setup B takes 20 input for 120 output per second, cost 1000 to make, composed of 5 machines.

Which one is the best ?

if you value the input ressources, one is almost twice as efficient. because you do 100 ouput/10 input vs 120/ 20.
If you don't, the other is roughly 20% better. because you do 100 output/5 machine or 120/5.

You can also vary the "cost to make" !

If you need legendary material to make your setup that produce legendary material, on paper it may be a very good setup, super efficient, super dense, as you described, as a "late game solution to mass production", but it can also maybe happen that you're left with great difficulty putting up enough legendary material together in the first place to build it. And if that leads to a situation where you spend 150 hours getting the setup running, and you let it run 50 hours and you have all the legendary stuff you ever need, maybe there's something to adress that isn't about "which setup is the best when late game is reached" ,but also "how do you plan to use the setup in a game ?" , "when ?" , "what are the tools available to you already ?" To reach a situation where you only need 50 hour to have a setup running, and you let it run 150 hours. Sounds like the setup is worse because it takes 3 times as much waiting for the legendary material, but really in a game sometimes you need the cheap one, so it runs while you build the rest of the factory instead of waiting forever to have the material to build "the final" blueprint.

Tertius wrote: Thu May 29, 2025 10:35 pm You already sent me on a fool's errand by mentioning "use legendary bacteria". That's junk. I wasted an afternoon trying to work out how to do that

I'm really sorry you could have thought i meant it seriously as a "go for this 1rst degree", are you not a bee ? i had tried to explain that it was the best in the "fun department" and because it rhymes, which are both "not serious" things to consider ! It was my way to point out that you didn't explained what you defined as "best", so for me "best" is "the most fun" and i showed a setup that is "the best" !

If you look carefully at the picture you could see that most of the biochambers are not working with "legendary bacteria", they use all sorts of quality because the lower level are much more comon, it was quite the challenge to try and make the legendary spoilable, because you need to maintain constantly a level of quality bacteria, and if you stop for a moment, it takes a while to restart from with the low quality bacteria so overall they are working mostly with uncommon and rare

Tertius wrote: Thu May 29, 2025 10:35 pm in the end it's dependent on legendary bioflux, which cannot be crafted efficiently. You don't need a huge factory for legendary bacteria, but you need the huge factory for legendary bioflux. Getting rid of all the unwanted non-legendary byproducts is the real challenge and the real waste. I didn't try but simply washing lava-to-molten and molten-to-plate with 10 foundries will probably have the same yield of legendary plates. Or the other methods I mentioned and already built.

Alow me to share a few picture from my experience then :

Here is the part that makes bioflux from quality spoilable : Here is a wider view : ( Disclaimer : i have actually built the setup and thus have experienced first hand the things i'm talking about x) i can send blueprint or save game if needed to turn speculations into facts ^^ )

I'm not saying it's "the best setup" ( apart in the fun department ), but if you are going to say "another setup is better" i'd expect you explain to me why, and not just because you prefer it, and when explaining why, there must be something objective i suppose a certain metric that would be "cost of the setup/ production per minute" ? or "number of machines / output" ? or "amount of input consumed ?" x). Judging by the fact that you answer 10 foundries would have the same yield "maybe-to-be-verified". But if you ask for "the best setup" and i don't know in which category, i can't answer, that's why i try to have you explain what you mean by "the best "!

You said earlier :

Tertius wrote: Wed May 28, 2025 5:56 pm Blue circuits are one of the crucial items. My approach is to do this on Vulcanus. Just uses calcite (mined) and normal plastic bars (from a coal liquefaction-based process, or perhaps imported from Gleba). Create normal blue circuits with productivity bonus and speed beacons in electromagnetic plants.

Now if you mention "imported from Gleba" i believe it's because you want infinite ressources.

But using infinite ressources is going to require a larger setup with more machines, than just going for the easier setup that uses ressources that deplete. So which one is best ? how can we tell ? it's up to you to be more precise, the choice of the "proper" setup i believe will just feel like a consequence otherwise when it's possible to use or the other and it's personnal preference that matters, you can't really ask for advices on what is the "best", more "what are the things that exists" and their pros and cons, that's the situation that will turn one or the other into the most appropriate.

If you do the setup on Gleba like i've shown, it may means you have quality biochamber, and if you try on Vulcanus it may means you have quality foundries, so in both case it's probably as important if not more to have a way to have the quality foundry or biochamber ! Because depending on how much time you need to create 1 legendary foundry or biochamber or productivty module or quality module, or one platform to do asteroid mining, one or the other method can be magnitude of order "faster" to setup and rip the benefit of. To me that's the most important in some real game when "i'm going fast" but in others or later, why not change the design to have something more UPS efficient, even though that means adding some "depletable" ressources to the mix, because i know i have reached a point where it's not going to effectively be the case.

To take your example :

Tertius wrote: Thu May 29, 2025 10:35 pm I don't want to build a 1000 tile factory and see a 100 tile factory in the next screenshot with double the output.

Then that's kind of an answer, because someone else may think :

But if you built your 1000 tile factory in 10 minutes after landing on Vulcanus and then you go on another planet while the other screenshot require 10 hours of waiting and upgrading with a trickle of material and produce nothing before it is finished ...

Tertius wrote: Thu May 29, 2025 10:35 pm

mmmPI wrote: Thu May 29, 2025 8:03 pm I'm a bit provocative maybe, but when i see "what is the better?" or "the most competitive" i'm still having the impression that i read ( both you and Tertius to a certain extend ) players talking about "the best potato" but you can have "the heaviest" , "the largest" , and "the tastiest", which are "more objective ways" of defining things ( one being a personnal preference). That's why i'm asking "what do you mean?" all the time ! or "lossy in what ? plants ?".

When you study software design or just good programming, you learn it's important to use the proper algorithms. If not, you're wasting resources. For example, if you often need to find some value in a list with 1024 entries, a very bad algorithm is to just search item by item until you find it. If you need to search often, your program will spend all the time with that (operations=1024/2=512). Instead, you could store the list sorted, then use binary search. This will use a fraction of CPU time (operations=log2(1000)=10). Or you could compute a checksum (hash) of the values, then use the hash as index into an array and just access the array element without any search (operations=1). Yes, the last algorithm needs 1/500 of the effort of the first.

It's the same with upcycling Factorio items: with the wrong upscaling "algorithm", you spend all your resources and don't get much output. But with a better approach, you build the same amount of machines but use much less ingredients for much more output in much less time.

This is what I'm trying to do: find a good, if possible the best, algorithm. I don't want to waste resources. I don't want to build a 1000 tile factory and see a 100 tile factory in the next screenshot with double the output. And since there are so many approaches with so many starting points, I want to profit from the experience of other people. I don't want to copy a blueprint, I want general information about the general approach.

You already sent me on a fool's errand by mentioning "use legendary bacteria". That's junk. I wasted an afternoon trying to work out how to do that - in the end it's dependent on legendary bioflux, which cannot be crafted efficiently. You don't need a huge factory for legendary bacteria, but you need the huge factory for legendary bioflux. Getting rid of all the unwanted non-legendary byproducts is the real challenge and the real waste. I didn't try but simply washing lava-to-molten and molten-to-plate with 10 foundries will probably have the same yield of legendary plates. Or the other methods I mentioned and already built.

I decided to give this a good read through before I posted. Hmm...

I wouldn't disagree with your logic and where I might expand on it, I think I need to be rather careful. You conditioned the input situation rather carefully and that is what is the key to algorithm selection.

With quality you can have an input situation where exactly the algorithm that works is completely wrong for another use case. And there's a gambling excitement that makes people discussing it a bit too excited to be smart. You can find an input condition that mocks an algorithm in comp-sci as well. But that's a creative hobby and not a serious discussion of its merits in the context of an issue.

If I'm going for buildings I'm happy to do it at an item a minute, by using a production building, a recycler to breakdown, and a chain of re-manufacture buildings. I believe Nine-Nine pointed that out early on and that's a good way to get a production line that can in a way that nothing else will. Edit: I'll start this strategy on T3 Quality modules because that speeds the process up the most. Next I might do modules at T2 because they're fast enough and cheap enough. I'll do forges and beacons, then whatever strikes my fancy.

If I'm going for components to make buildings, I probably have a crushing amount of material motivating me. So that's my input condition. At that point, I'm looking for a quality bonus of 18% and a building that can do production/re-manufacture at a rate that'll convince me I'll use my input IPS up. I might adjust quality up if I convince myself I'm getting a rate worth mentioning. Recycler is going to take out 1/4 for each tier of item you put in, so even if I get a perfect upcycle I get 1/4 of a red chip out and (1/16)*20 green ones worth of items out of a blue one. The least wasteful processes take time and require a petrol chem product.

So you can play games like compare 30% batteries (in a cryolab) to 24% iron boxes (in asmblr 3), and while you lose more stuff boxes spits out more items in a smaller footprint. It's pretty funny in the map editor at 64x speed. Not great when you're looking at it in game at real life rates and at scale.

As a final note: I'll do it on Vulc. Because I have Long Tall Sally (the railgun) and I can use the geography for voids and a lot of the products I'll need at scale. Vulcs problems can be solved by just having a mining bonus. Other places need a shipping line.

coffee-factorio wrote: Sun Jun 08, 2025 3:30 am With quality you can have an input situation where exactly the algorithm that works is completely wrong for another use case. And there's a gambling excitement that makes people discussing it a bit too excited to be smart. You can find an input condition that mocks an algorithm in comp-sci as well. But that's a creative hobby and not a serious discussion of its merits in the context of an issue.

I agree that the use case is what will determine what algo to use considering you meant which factory layout, because they all have trade-offs and as such the first question to answer in "how to upcycle?" is "what is the expected result ?" because what works for small quantity or long term or mass productior isn't the same, even more when you consider the subjective evaluation of the cost of "input" can apply to ressources consumed per cycle, or initial cost of the instrastructure, or any ratio in between.

And not every player value the certainty of having something the same way, if you want 1 armor you want it 100% chance, but if you want 100 beacon , it's ok to have 90-110 maybe, so the method may differ.

But to me it appears both a serious discussion and a creative hobby and is more than enough to get me excited. I think the same can be said about the part where you use your "smart" to remove the "gambling" aspect of upcycling, and you make it a statistic, that's exciting !

coffee-factorio wrote: Sun Jun 08, 2025 3:30 am As a final note: I'll do it on Vulc. Because I have Long Tall Sally (the railgun) and I can use the geography for voids and a lot of the products I'll need at scale. Vulcs problems can be solved by just having a mining bonus. Other places need a shipping line.

Isn't that the case for every planet except Aquilo that mining bonus solve everything ?

Vulcanus is good but if you're mentionning mining bonus to me it means you're mining the coal, and thus to me the same "problems can be solved with mining bonus" applies to Nauvis and Fulgora.

mmmPI wrote: Sun Jun 08, 2025 2:50 pm Isn't that the case for every planet except Aquilo that mining bonus solve everything ?

Fulgora is another exception as mining bonus can actually be detrimental to designs that can't destroy byproducts fast enough.

I've been doing legendary science lately and also got excited with quality designs! It's quite fun as the optimal algorithm (e.g. best yield with reasonable UPS at required throughput) keeps changing up to productivity 30s and can require a combination of different steps. Like upcycling uranium ore until rare, then process into U-238, then upcycle U-ammo to epic then convert into U-235 and upcycle nuclear fuel (the end product) to legendary.

h.q.droid wrote: Tue Jun 10, 2025 3:58 am

mmmPI wrote: Sun Jun 08, 2025 2:50 pm Isn't that the case for every planet except Aquilo that mining bonus solve everything ?

Fulgora is another exception as mining bonus can actually be detrimental to designs that can't destroy byproducts fast enough.

I've been doing legendary science lately and also got excited with quality designs! It's quite fun as the optimal algorithm (e.g. best yield with reasonable UPS at required throughput) keeps changing up to productivity 30s and can require a combination of different steps. Like upcycling uranium ore until rare, then process into U-238, then upcycle U-ammo to epic then convert into U-235 and upcycle nuclear fuel (the end product) to legendary.

I share the excitment, but i don't see Fulgora as an exception, i was reusing the person's to whom i was adressing words when mentionnig "mining bonus solves everything" in Vulcanus.

I understand the lava can be used to void stuff unlike in Fulgora, but it is a different thing for me than the mining bonus and how it "solve everything" in Vulcanus in a different way than it would in Nauvis, or (or Fulgora).

If you rely on "mining bonus" on Vulcanus, it means to me you have choosen a planet with infinite ressources, except 2 calcite and coal, and you haven't setup things to make those "infinite", as such the similar situation could occur on Nauvis of Fulgora, using non-infinite ressources, but made quasi-infinite due to high level of "mining bonus".

The exception are Aquilo and Gleba to me because you can't mine iron and copper there so mining bonus is of limited usefulness.

I have prefered investing in asteroid productivity than mining productivity at some point in games that last long enough for legendaries.

mmmPI wrote: Sun Jun 08, 2025 2:50 pm

coffee-factorio wrote: Sun Jun 08, 2025 3:30 am With quality you can have an input situation where exactly the algorithm that works is completely wrong for another use case. And there's a gambling excitement that makes people discussing it a bit too excited to be smart. You can find an input condition that mocks an algorithm in comp-sci as well. But that's a creative hobby and not a serious discussion of its merits in the context of an issue.

I agree that the use case is what will determine what algo to use considering you meant which factory layout, because they all have trade-offs and as such the first question to answer in "how to upcycle?" is "what is the expected result ?" because what works for small quantity or long term or mass productior isn't the same, even more when you consider the subjective evaluation of the cost of "input" can apply to ressources consumed per cycle, or initial cost of the instrastructure, or any ratio in between.

And not every player value the certainty of having something the same way, if you want 1 armor you want it 100% chance, but if you want 100 beacon , it's ok to have 90-110 maybe, so the method may differ.

But to me it appears both a serious discussion and a creative hobby and is more than enough to get me excited. I think the same can be said about the part where you use your "smart" to remove the "gambling" aspect of upcycling, and you make it a statistic, that's exciting !

Quite the contrary. I cautioned that the excitement makes people too excited to make good moves. If you approach the issue creatively quality has a habit of rewarding you with a halted assembly line or worse; it flips your logic on its head and burns you out.

When I do what you say and use my smarts, I build what is effectively a Monte Carlo sim in a map editor. And I run that. And when I get a number out the other end that suggests that for 18 ips iron ore input I get 1/4 ips output legendary quality gears with a combination of 4 quality modules in every system (forges, ect). I'm doing this at 64x speed for a minute (60*64 >=3600, so that's running an assembly line for an hour). And when I do this I'm thinking of making a legendary science pack.

Here's how I be the man Tertius is scared of: I swap in green modules and a little productivity I have that same system running 18+ ips in the same footprint. My forge line shrinks because it goes faster under that model and I swap in green modules so my power plant shrinks too. If I do 2 legendary green modules and 2 legendary t3 productivity module on the drills then I see I get 3.49 instead of 3.0 out of each drill. In theory I do the same thing to my gears line, pessimistically I limit my ambitions to a small power plant because I already know how this story is going to go.

There will be a six bonus on the legendary packs which completely fails to account for material waste in the system. Your bonus at the lab is f(x, I) = L*P*I*x. L=2, that's longevity from buildings. P = 4 that's from tier 3 legendary modules in final assembly and research. x is final output hitting the labs and I is the longevity bonus on the items. But x will be ~20 when I is set to 1. And 1/4, 6 with a bad cycle. If I do a bad cycle then that's 48/4 = 12 ips real research out. If I do what I always did that's 20*8 ips research out.

If I consider a process with a 1.5 productivity bonus Tertius can get a line that can perform in terms of material savings... if all he wants to do is red science. The bonus I'm seeing suggests 6x longevity bonus, 5.8 ips as my inputs then. We're talking 30*8=240 ips out. I'm seeing 1.5, 1.7 out in which point "Huh, Kovarex's number is usually 1.6. Mine is 2." The process I'm using is batteries, which is completely impractical but gives me a view of what happens when I have to shred a line of copper wire down from an EM Plant to get plates for red science.

But if I put legendary speed and productivity on the forging steps, or put calcite into a foundry, I'll arrive at compounding mechanical advantages for everything but the difficulty of shipping items that just outperforms what quality can do in terms of startup difficulty. Because that 18 ips that becomes 20 ips is going to really get multiplied by 2.5 with some calcite and productivity, then 2.5 again when I make raw plates out of a foundry, and when I work gears I'll probably have enough left over to do automation science. And then that becomes 125 ips raw * 8 from lab longevity and productivity. I can see a way which might deliver a similar advantage to quality but I've yet to be given a 2x general advantage which is going to convince me to use it in any seriousness. I respect the hustle but it's just depressing to see that the ideas aren't delivering. Edit: Moved for clarity: I get intrinsic bonuses and gears and steel that are persuasive and a mining productivity bonus becomes easy enough to purchase that the material savings of quality is useless, because the system doesn't scale horizontally or vertically.

Someone might be persuaded by having 4x productivity on a few items. How do you upcycle sulfur for chemical science? Remember, my criteria is a 2x general advantage. There's several special cases that have had me thinking for a long time I'll never do quality science outside of a simulation. Sulfur is one of them.

mmmPI wrote: Sun Jun 08, 2025 2:50 pm

coffee-factorio wrote: Sun Jun 08, 2025 3:30 am As a final note: I'll do it on Vulc. Because I have Long Tall Sally (the railgun) and I can use the geography for voids and a lot of the products I'll need at scale. Vulcs problems can be solved by just having a mining bonus. Other places need a shipping line.

Isn't that the case for every planet except Aquilo that mining bonus solve everything ?

Vulcanus is good but if you're mentionning mining bonus to me it means you're mining the coal, and thus to me the same "problems can be solved with mining bonus" applies to Nauvis and Fulgora.

No. The other planets have hidden costs in the form of maintaining a defense network or i/o bounds imposed by landmass scale. These problems are best solved with a shipping line and persist even if you have a mining bonus: you can have a 240 ips drill on Fulgora but what's your plan to bring that drill to a processing center at that rate? Several kilotons of foundation will enable a solution. No research will make a multi-car train bigger or allow it to load and unload consistently.

Are you planning on solving pentapods without a gunline? Even if you kill them all, you'll be left with a map where you need to keep the spore clouds from touching unexplored chunks. The best solution to that will be shipping in biter eggs in the long term; and I was not unaware that the amount of geography you take is your resource bound on Gleba. Not your mining bonus. But to use it well, you need shipping. Nauvis is the same because a legendary rocket launcher line will hold forever as long as you have a space platform that can deliver. But that's energy and running robots to run that.

The thing about Vulcanus really is that I need to import a gun, break the back of some big worms, and have enough health to hug the small ones (they have the same adorable derp face as a pitbull). However, if you are just looking for buildings and do science on Nauvis, then all you have to really do is ship Vulcanic science there and have enough petro-chem for expensive runs of a thousand objects over a few hours. And using artillery you can get enough space that I think you can pull it off even without Earandel's rifle by making small and medium demolishers go extinct.

As for asteroid productivity, you'll always have a bad petrolchemical logistic on a platform relative to a surface but otherwise it's competitive with Vulcanus. And the time investment is large; but I love the things so I'll fight anyone who says that the investment isn't worthwhile. The space bound and petrol bound at asteroid productivity 10-20 prevented me from getting out a high tiered module line and there isn't stone up there. But I did do a blue chips line that more or less rebuilt Nauvis in a long game so... \o/

From what I gained from the thread, a valid approach is to first create normal ingredients with largest productivity bonus+modules possible (and speed increased by speed beacons), then start upcycling the product you want in machines with inherent productivity bonus.

The start is quality quality modules. So I started a factory to upcycle quality 3 modules. On Fulgora, because it requires superconductors, and because having quality electromagnetic plants with their increased speed lessens the amount of modules required, and electromagnetic plants can only built on Fulgora.

Started with 48 normal electromagnetic plants for the normal quality module 3 recipe, 10 for the q2 recipe, 4 for the q3 recipe, 2 for the q4 recipe and 1 for the legendary recipe, and 23 recyclers. All with normal quality modules 3. Every few minutes I collect the q2+ modules and replace the normal modules in the machines with the better quality ones. If the all have q2, I replace them with q3, then q4. Went rather fast, just an hour or 2. Now already a steady flow legendary quality module 3, and I'm replacing all previous modules.
That's really promising. This amount of plants for bootstrap is reasonable, and the amount of ingredients is high but manageable. The whole thing will shrink with quality electromagnetic plants.

So I will see how to create upcycled electromagnetic plants. These require a gigantic amount of holmium, so I need to check how much I can extend my scrap recycling facilities, since holmium is the bottleneck here. Current scrap recycling productivity is 14 (+140%). I tried an upcycling facility with 16 normal plants in map editor with great results, however they need ridiculous amounts of holmium, so I probably have to scale that down to 8 or 4 or something like that.

And it shows the way to the next thing to upcycle, or in parallel: quality productivity modules to increase availability of diminishing resources. Not sure if this will still take place on Fulgora but instead on Vulcanus - depends on how many green/red/blue circuits are still available from scrap recycling.

coffee-factorio wrote: Tue Jun 10, 2025 8:53 pm No. The other planets have hidden costs in the form of maintaining a defense network or i/o bounds imposed by landmass scale.

That's wrong for Fulgora. No ennemies and no bounds imposed by landmass in end game .... They are the same thing to pave Fulgora and Vulcanus ... It's just wrong to say "mining bonus solve everything in Vulcanus" to me. It makes no sense because if you go to Vulcanus to use the lava, you can make it so you don't need the "mining bonus" whatsoever, and then "mining bonus" is wrong research.

If you mine something in Vulcanus, you can just do the same in Gleba, or Fulgora or Nauvis, after all there too the "mining bonus solve everything". Since you are not going to use the "true" infinite ressources but rely on high productivity bonus.

Otherwise you keep saying "this is the best" , "this is the best", but that makes no sense to me to state your personnal preference like this. It seem a step backward.

Tertius wrote: Tue Jun 10, 2025 10:47 pm The start is quality quality modules.

That i agree with no reserve

Tertius wrote: Tue Jun 10, 2025 10:47 pm From what I gained from the thread, a valid approach is to first create normal ingredients with largest productivity bonus+modules possible (and speed increased by speed beacons), then start upcycling the product you want in machines with inherent productivity bonus.

If you don't have "many" quality quality module, the positive point of this approach is that you only use them at the end, on "high value target" or "dense material", you can have for your holmium production, a very large complex filled with productivity modules ( and speed beacon), and only a couple machine that makes "EM plant" (or module) while having the "quality quality" module inside.

The downside is that it's going to generate a lot of material from a single ore, before being reduced at the end, and the "middle" can represent a need for a lot of infrastructure to process when another approach where you have "many" quality quality module, could consist into using them as early as possible in the production chain, in the mining drill if possible, this cost a lot of "initial investment", because you can't beacon them and they "roll" on "low value target" or "not-dense material". However, this makes sure the subsequent step in the production chain are processing "refined" input, which also represent a lower quantity of material.

From a a "theory" point of view, i think it can be expessed this way :

1 Raw material makes 4 intermediate-1 which makes 16 intermediate-2 which makes 64 intermediate-3, which makes 128 intermediate-4 which ends up being 1 "finished product of the desired quality".

1=> 4 => 16 => 64 => 128 = > 1

But you can imagine you can "swap" the order, and you start by trying to achieve the desired quality level first, so you would need 128 raw material to make 1 of the desired quality, but then from it you could make 4 intermediate-1 of the desired quality, and 16 intermediate-2 and 64 intermediate-3 and 256 intermediate-4 that would be of the desired quality ....

128=> 1 => 4 => 16 =>64=>256 !

I think it illustrate how in one case there is "little to no" drain on the raw material, and on the other on it's the complete opposite, it devours raw ressouces, but set you up for good if the production chain are super long ( thinking 'mods' too ). Most of the infrastructure is dedicated to "raw material" and upcycling them. (the 128=>1 step doesn't require the same number of machines in both situation x).

The approach you mention is like the "first case", it minimize consumption of raw ressources which synergize well with the idea of then using "infinite ressources" or "quasi infinite ressources due to mining bonus". It is "valid" imo, in some situation it is even mandadory to some degree, because you can't make quality holmium plate from quality holmium ore, so you can't just "upcycle the raw material" that would be a waste, the "earliest" possible , the "rawest material" would be the plate, but then it makes more sense to transform them into EM plant and recycle them if you are going for "legendaries", "recycling at the last step".

I'm saying all this because in a "real game", if you consider the case of Vulcanus, you have scarce coal there, but you don't even need to use it if you get it from space. Now if you set up for "asteroid mining", you will still need to expand for tungsten eventually, until very high mining bonus, and in doing so you will "unlock" this "easily available coal". This can feel tempting to just "void it into legendary steel copper and plastic" as you can upcycle for legendary coal, and use for plastic, or you can make plastic first and only then recycle it into legendary. That's a very short "chain" where the optimal i think depend on your plastic produtivity research. Both would be "valid" imo, and the "best" can depend on things like "how many legendary cryoplant vs how many legendary mining drill you have available". Not for "science", but for "module", because for "science" you'd want the "sustained infinite from space", but for infrastructure or "module" that are "1 time build" even if you build a lot, it's different. There's a lot of subjectivity still.

I think this is very important to consider ! when you use things ! as much as "what kind of design".

Tertius wrote: Tue Jun 10, 2025 10:47 pm And it shows the way to the next thing to upcycle, or in parallel: quality productivity modules to increase availability of diminishing resources. Not sure if this will still take place on Fulgora but instead on Vulcanus - depends on how many green/red/blue circuits are still available from scrap recycling.

Hahahaha ! parallel or next is a difficult question , when do you consider you're done if you decide to do one after the other too x). That's highly situational, that's to me the core of the question, beyond the choice of design or which recipe to use, when they are appropriate.i found very difficult to know in advance. You don't want gigantic buffer of quality quality module sitting in chest while lacking productivty or the opposite.

Tertius wrote: Tue Jun 10, 2025 10:47 pm Started with 48 normal electromagnetic plants for the normal quality module 3 recipe, 10 for the q2 recipe, 4 for the q3 recipe, 2 for the q4 recipe and 1 for the legendary recipe, and 23 recyclers. All with normal quality modules 3. Every few minutes I collect the q2+ modules and replace the normal modules in the machines with the better quality ones. If the all have q2, I replace them with q3, then q4. Went rather fast, just an hour or 2. Now already a steady flow legendary quality module 3, and I'm replacing all previous modules.
That's really promising. This amount of plants for bootstrap is reasonable, and the amount of ingredients is high but manageable. The whole thing will shrink with quality electromagnetic plants.

This is exactly what i feel upcycling is about, the process that actually takes time during a game. Replacing the "normal" electromagnetic plant by higer quality ones, starting by those that already have the higher quality of quality module, and that's a process that happens with foundries and biochambers and recyclers too.... x)

It feels to me more about optimizing the path, rather what you have at the end, because once you get the quality rolling, which takes time, and once you get it good so you have decent legendary material output per unit of time, and not just "enough for an armor" you can "easily" make any "full-legendary setup" designed to make more in a very dense way. That's not the hardest part, the hardest is getting there in the first place.

48 EM plant for "normal quality module3" is a lot for me, i'd personnally make less and use a speaker to tell me when i need to do something about it be it in a few hours and i had forgotten, different playstyle ^^

Have you tried to mine quality scrap ?

My 2 cents: legendary quality 2s are much cheaper and provide better values than less-than-legendary quality 3s. My path was to first roll asteroids with normal quality 2, then get a bunch of legendary quality 2s and roll asteroids with that. And only upcycle EM plants once I got enough legendary quality 2s to fill everything.

mmmPI wrote: Wed Jun 11, 2025 12:51 am Have you tried to mine quality scrap ?

I tried, but as far as I see this becomes feasible with quality mining drills and quality recyclers, and quality quality modules of course. Before that, needs a rally huge amount of machines with low quality modules and low yield.

Upcycling is a loop, and the more often the loop is processed, the better use was made of the modules within that loop. However, with lower quality modules the drills and the initial scrap recycling is just the 1st step of such a loop, while the modules in the machines+recycler loop are visited as often as required to produce the wanted quality. So the former can be better replaced with means to increase initial throughput, for example speed modules. A thing that isn't probably necessary later, when quality machines become available with inherent speed boost.

That's probably the next thing on my checklist of quality machines to create: 1st: electromagnetic plant. 2nd: recycler. 3rd: big mining drill (because of the lowered source drain). Then foundry.

h.q.droid wrote: Wed Jun 11, 2025 2:03 pm My 2 cents: legendary quality 2s are much cheaper and provide better values than less-than-legendary quality 3s.

I tend to agree on this, it's necessary to use less-than-legendary quality 3s at some point, at least before you have them. legendary quality 2 are just fine to use in many cases, but for my particular case i'm more fond of epic quality 3 or "rare" quality 3, the blue ones. This is because every "legendary tier 2" is a potential intermediate product for my legendary tier 3 production chain from legendary material, whereas a "blue quality tier 3" is like a failed legendary from the "other" production chain, the one that tries to be lucky and recycle the "failed". Or in some case, i'm just using them, when i don't have enough legendary "yet". That may also have to do with the fact that i "prefer" going for "rare" quality everywhere, before going for "epic" or "legendary" anywhere.

Tertius wrote: Wed Jun 11, 2025 2:34 pm I tried, but as far as I see this becomes feasible with quality mining drills and quality recyclers, and quality quality modules of course. Before that, needs a rally huge amount of machines with low quality modules and low yield.

I don't know "how much quality" is the baseline for this. I have used this as a "slow and steady mechanism". It's not necessarily aimed at "legendary" at first, i'd be just scrap mining with quality, and recycling the scrap with quality module in the recycler, it's roughly similar in speed as if doing a non-beaconned base, so if you do it "early" game it doesn't feel "slower" than a regular Fulgora settlement to me. I have started such things on a another "large and empty island", compared to the "small and filled with ore". It felt to me like "utilizing" the "cheap" space gained when connecting such island which would otherwise be "useless" compared to how "expensive" is the space on the first island in which you settle in Fulgora. I was using "rare" quality probably everywhere before using "epic" anywhere.

Tertius wrote: Wed Jun 11, 2025 2:34 pm Upcycling is a loop, and the more often the loop is processed, the better use was made of the modules within that loop.

Well in the case of scrap mining, to me, it's a little different, it's not necessarily a loop, you have "science and rockets" which act as sink for every "normal quality" things, and you have the "quality material" as "byproduct" that end up piling up somewhere in Fulgora. If you were'nt going to use speed module or beacon in your "mining" island, you loose a bit of throughput by placing quality module instead of nothing, and you need more accumulator than if you had used efficency module, and it makes the train bringin scrap logic more complicated. But overall, it allows to "extract" some quality material scaling with the size of your science production somehow.

I agree that's not the efficient use of the module x) But i wouldn't use my first "legendary Q3" there, i would use some from "normal quality tier2" to "rare Q3" once i have too many, or maybe because i don't have "epic" or "legendary" yet. "It's the last place where i'd put module" but it's also a place that's available, that doesn't necessarily need to be made for them. If you start quality on Fulgora as soon as you arrive there, you can have such setup working for 30 or 40 hours, starting at the 20 hours mark, and you upgrade it maybe 2 or 3 time over the course of the game when you unlock "epic", and "legendary".

Just make big buffer chest and "void" everything that is above XXX quantity later in the game, when it's overflowing with quality material that is no longer needed x). I know this "works" because i used this. But it depend on when you start doing it in a game wether it make sense as a viable setup or not i suppose.

Tertius wrote: Wed Jun 11, 2025 2:34 pm However, with lower quality modules the drills and the initial scrap recycling is just the 1st step of such a loop, while the modules in the machines+recycler loop are visited as often as required to produce the wanted quality. So the former can be better replaced with means to increase initial throughput, for example speed modules. A thing that isn't probably necessary later, when quality machines become available with inherent speed boost.

That's probably the next thing on my checklist of quality machines to create: 1st: electromagnetic plant. 2nd: recycler. 3rd: big mining drill (because of the lowered source drain). Then foundry.

I fully agree on where the priority lies for the modules and why they are better used in "high-value/dense" process. That's considering a situation where "module is scarce", where to place them "first" to maximize their utility. That's not necessarily the case in a game if you use the "scrap mining" early, where you can have "many rare tier 2 quality module", just because of "upcycling holmium plate", passively, because you produce so many "recycler" on Fulgora, that you need scrap for their blue circuit and concrete, which yield "holmium ore" , more during a period than you need for science which is turned to plate which are "upcycled when above XXX quantity", even if it's not the "optimal way" to upcycle holmium, it is " one way" to create a lot of recycler on Fulgora, which "may yield" some quality holmium plate.

I'm not saying what you say is wrong, i'm saying that depending on how some players approach the game they won't have the same reasonning and situation, the module are always more efficient in high value process, but if you have plenty of modules "already", one may not value the "efficency of the module" as much and just put them in "already available places". I'm sharing it because i think the more method one "knows" their pros and cons and how to use them, the easier it is to pick the "most appropriate" in a game.

I think EM plant recycler drill and foundry , in this order makes a lot of sense.

Tertius wrote: Tue Jun 10, 2025 10:47 pm From what I gained from the thread, a valid approach is to first create normal ingredients with largest productivity bonus+modules possible (and speed increased by speed beacons), then start upcycling the product you want in machines with inherent productivity bonus.

The start is quality quality modules. So I started a factory to upcycle quality 3 modules. On Fulgora, because it requires superconductors, and because having quality electromagnetic plants with their increased speed lessens the amount of modules required, and electromagnetic plants can only built on Fulgora.

Yep. Quality quality modules is probably why you won't see many people stop to do a lot of quality objects outside high value items where "any % helps" like quality lightning collectors, tanks and spidertrons in the first hundred hours. The rate seems to be about the same when applied to an assembly line so it doesn't help you if you're trying to get somewhere quick, and a legendary speed boost cuts a layout in half on a building so it's hard to justify not just delaying till you get the research on Aquilo.

That and you trigger an upgrade cycle where you suddenly need less than half as many layouts no matter what you do, it's a lot of effort.

The geometric nature of quality means that you ultimately want the best bonus of it, because small decreases translate to huge amounts of material inputs. There's situations where you can overcome that but you're probably using a combination of game mechanics to have 1k of raw materials per second to do it. You'd need a certain kind off emotional strength to say "well, rare will let me cut my layouts in half once all force multlipliers are applied" while you're trying to get other upgrades that enable you to do things that are qualitatively different.

As for your buy list. I mean, I have opinions but the order doesn't differ much. I'd do as many modules as I could get in parallel with beacons and at the best tier available to avoid upgrade logistics in the long term. Look into legendary bulk inserters combined with normal stack inserters though. That ends up being a huge force enabler, because you can frequently stack at N ips with some difficulty at output, but you can't feed that into 9 slots at the rates quality allows. And bulk inserters are easier to get.

The only thing of note is that tier 2 module's are universally available and tier 3's require dedicated work stations. You can ship holmium and do the Fulgora chain anywhere; only em plants and recyclers (excuse me while I cry manly tears ) are actually limited to Fulgora. Fulgora gets so many banal rocket parts that shipping is reflexive. It's the place's great strength. Whatever I can say about an individual surface for getting components cheap, if you can half-effort 15 assembler 3's an hour, 60 beacons an hour and and 60 modules 1 an hour on every surface you can still probably snap the game in half within a few hours and buy anything else you want. Inherent productivity is nice but it doesn't make inserters.

Keep the idea of just doing raw materials to quality entities in your back pocket to avoid race conditions that will ultimately determine your rate of input. This happens whenever you play with spoilage. To give a practical problem case: look at your options for upcycling jelly from jellynut; in a timely fashion for stack inserters. And then apply the lessons learned to other things like biochambers and biolabs.

mmmPI wrote: Wed Jun 11, 2025 12:51 am Have you tried to mine quality scrap ?

Every now and then I run around in circles with ideas for 2-10 hours at a time. It was my first experiment with quality It's a 60 sort, so if you use bots you probably need lightning collectors to defend them from lightning Foundation will functionally ease the difficulty of that sort if you do it with splitters and you probably need about an accumulator per building. Foundation and quality will help keep those from overrunning your layout due to their power output limit. Oh. And since it's 1/25 * 1/4 to immediately get a quality item like LDS you'll naturally be drawn to look need to look at i/o rates of 100. Which will drain an individual train car in 20 seconds. A train has to accelerate and decelerate, and if you take the time to check what that does to i/o it'll quickly become apparent that you just can't get a great rate out of the train that island geography will enforce on you. It doesn't mean that it's impossible, just that you need 20 trains to make 20 seconds 400 seconds. Or you can drop in red belts and foundation to get 240 ips constant without touching trains at all. And since you can get 100 to 200 productivity researches, you can practically get over that rate so you probably will end up looking in that direction

And that's true on every surface.

It's not that I disagree with the sentiment that it's a step back. I can laugh at some of the stuff you said because it's solvable but not practical or not fully true. I really don't like what that number 40 does to my conversations. The challenge is worth the price of admission even if I'm being prodded in a way which makes me cast it in a vain light. Until that point. And then all I can say is; this stuff is a lot more brittle than you'd think, so there's some stuff where just kind of hurts but that's what best honestly is after 800 hours and 2 runs. If you got a better number tell me again so I can justify fun with a big machine. Edit: That and I'm doing some Krastorio right now, so I can see what stack size 200 does to game balance. It seems like an oversight to not push it up when you need to move raw materials, since the game has a lot of moving parts. I'm running Deadlock's stacker's and played some Nullius, so I'm seeing where this idea was and wasn't fully baked based on that.

coffee-factorio wrote: Wed Jun 11, 2025 5:00 pm Inherent productivity is nice but it doesn't make inserters.

It's true I'm obsessed with efficiency, but I'm not THAT obsessed to not upcycle such a thing. The only items I will not upcycle are items without quality bonus except hit points. Of course I need quality inserters.