In the crystalliser build, if you pull from the bottom two belts and put into the top two belts, with the non-processed belts swapped in between the in and out, that'd simplify scaling the build.
Did a MAM myself. A not completely, but a lot different approach than you did. E.g. the shapes are fed from the side, colors from above in a matrix 4x16 like structure where each layer consists of 16 shape inputs (as each side can contain each of the 4 shapes). Pins are added last as they never get painted - but I, too, took them as a "fifth" shape if you will. The other ruleset (check level below for pins) basically is destroyed once you have pins on the top layer or pins above pins (you would need to create 3 loops for the top layer, 2 for the 2nd, one for the 3rd and none for the bottom layer), but you cannot remove the top so I stick to have crystalizers first and crush them to get pins, too. Right now, I am trying to figure out a way to integrate the crystalizing part logically with missing parts of a layer or stacked shapes onto crystalized layers. A great example of this is task Nr. 35 with missing layers, remaining pins and shapes put onto whole crystalized layers. It's quite tedious to actually integrate EACH possible shape into a MAM. Especially, if you think you figured it out (just with pins in this example) and then you face something, your totally overcomplex MAM cannot even do: Nr. 30 with pins on the top layer. I seriously was fuming as my logic to compare each and every layer of existing pins to forward this to the pin pushers was extremely complex and took literal hours to build and in the end was deleted... But in the end, it's funny to see other approaches that somehow have very similar steps and logics here and there. For example, the "null" to "0" converter is almost identical in my build. And the resigning look on my face when looking at your approach handling the crystalizer test-MAM-logic as empty parts of a layer will immediately stop shape combiners from working. It took me quite some time to figure stuff like this out and it requires a whole lot of re-checking each in- and output to verify fitting data. It's great to see there still is a video coming. Love the content.
You shouldn't need an 'alternate' crystalizer. Instead, just have the let and right lanes swap in the crystalizer build itself. This means that at the start you take the 2 left lanes, then swap the left and right lanes. This way both builds are identical because in the 2nd build, the right lanes are now left, and are taken off as is normal.
At the random operator shape #1 (crystal / bottom right) I got this shape at level 72: crCucrSu:cmcmSuWu:cmRuWwcm:P-P-P-P- As you can see it has pins at the top level, so the rules that you told are not quite right. Fortunately I managed to build a circuit that handles pins. Very helpful series btw. :)
You could also simulate stack one layer with any full shape and destack it, so you will get layer without crystal, build layer like in normal MAM and in the end just compare if layer is the same like in the target layer, if not - add crystal, if so, pass thru.
I figured a base for crystal design(on paper, not shapez 2), which includes all layers. So I won't watch this video until I turn the initial base design into a working design to not spoil the fun.
Yep, that's what I implemented into my MAM. Build the bottom layer and crystallise. Build the second layer, stack it on top, then crystallise. Then the same for the third and the top. It works pretty well, but won't allow for gaps or pins in lower layers - not sure if those come up in the random generator.
Idk why this works (where else would the 7200 limit be relevant?), but it does make space management so much easier. I have a compact pipe gate array too, consider blueprinting one per space belt.
Where did you get the "rules" for the random shape milestones? It would be nice if these rules were listed in the game :) Maybe that's coming in a later update?
Just a heads up. Your logic that replaces crystals with pins in the shape will fail if there are empty spaces in the layer, since the stack operations will jam. I learned that the hard way.
so my guess for todays episode; you'll have to linearize the layers before joining them with splitters. basically each qaurter needs to be stacked individually then go through the crystal process and go back with spliters without damaging anything. idk how to join pins and crystals with them on top of each other; but if it's possible then you can do that individually in each quarter. otherwise you should be able to at least have one color on each layer.
when i taught cs100, one of the examples of "people who almost invented a computer but ran out of funding" was a guy who was too lazy to calculate trig tables so decided to invent a computer instead. Lazy FTW :)
My fully automatic mam is working, 12 x180 belts, 4 layers, pins and crystals. The only problem wyth cristal is that take 15min more to swap beetwin 2 random operator shape because we have to stack layer by layer to presevre crystals. TO make this more efficiant i've créated some complex purge system and i work with quarter parts instead of full shape. Let me know if you what to see that.
@@jelster2 Those 'rules' that he establishes are his design rules, not game rules. My current shape has cw--cw-- as the bottom layer, and the 3rd layer is multi-colored crystal cgcrcrcg. The only rules that the game has is that no crystals appear in random shape 1. This is noted on the game's official wiki.
Can somebody explain to me why we are needing the whole change crystals to pins logic, shouldn‘t this work just fine without the pins? Just cut the crystals out in the beginning and the rest is doing its thing. Also very nice Video, as always!!
@@Bionicmann01 I think the main reason you need the pins is in the case where the entire layer is crystal. You would end up with no shape and you need pins there to have something to crystallize. In my MAM I’m doing what you are saying: just get rid of the crystals in the shape, and do a last check for that one case. If so, just create a layer with a single pin. Everything is working so far.
The logic around 27:47 is a bit silly, you can just NOT the color instead of having the color signal gate some other signal and then NOT that one. The value of the signal coming from the yellow receiver isn't used for anything!
I’ve managed to fit my crystallizer on a 3x3: it takes 2 of each primary as input, and outputs a full belt of any 7 colors. I’d love to know how compact that functionality can get.
Unfortunately, the 'rules' for this design make this MAM not very useful. I've encountered multi-colored crystal on same layer (eg: cgcrcrcg), and crystal-only layers with holes (eg: cw--cw--). This is a good MAM design for ROS#1, but not for ROS#2 since the 'rules' are broken by the game.
13:46 - 1 Space Pipe can "consume" and support 72 Liquid Thrower. Since 4 Liquid Thrower support 1 regular Pipe 1 Space Pipe can support 18 Pipes ... or with a Pipe on each layer of a 1x1 Platform ... 1 Space Pipe = 6 Platforms (with 12 Liquid Thrower each). Did you even calculate? Ôo 19:12 - I'm just shaking my head. Why? Why using pins if not necessary? That's a pretty convoluted approach. Just replace the crystal segment with a Null signal and fill the gaps after merging the not crystal shapez with ... crystal. Using pins unnecessarily for this purpose is a pure waste of resources.
![Multi-Level CRYSTALS To Complete The MAKE ANYTHING MACHINE [MAM] | 06 | Shapez 2 | Tutorial/Guide](http://i.ytimg.com/vi/u393BXQMBZk/mqdefault.jpg)
![Multi-Level CRYSTALS To Complete The MAKE ANYTHING MACHINE [MAM] | 06 | Shapez 2 | Tutorial/Guide](/img/tr.png)
![MAKE ANYTHING MACHINE [MAM]: Shape Selector | 01 | Shapez 2 | Guide/Tutorial](http://i.ytimg.com/vi/E6qntnB09YU/mqdefault.jpg)
In the crystalliser build, if you pull from the bottom two belts and put into the top two belts, with the non-processed belts swapped in between the in and out, that'd simplify scaling the build.
So sad there is just one episode left, great series, thank you!
Until hexs
ikr, i save them up till they were all available for a nice binge and now i can only think... noooo, it's almost over. love the series.
Did a MAM myself. A not completely, but a lot different approach than you did. E.g. the shapes are fed from the side, colors from above in a matrix 4x16 like structure where each layer consists of 16 shape inputs (as each side can contain each of the 4 shapes). Pins are added last as they never get painted - but I, too, took them as a "fifth" shape if you will. The other ruleset (check level below for pins) basically is destroyed once you have pins on the top layer or pins above pins (you would need to create 3 loops for the top layer, 2 for the 2nd, one for the 3rd and none for the bottom layer), but you cannot remove the top so I stick to have crystalizers first and crush them to get pins, too.
Right now, I am trying to figure out a way to integrate the crystalizing part logically with missing parts of a layer or stacked shapes onto crystalized layers. A great example of this is task Nr. 35 with missing layers, remaining pins and shapes put onto whole crystalized layers. It's quite tedious to actually integrate EACH possible shape into a MAM. Especially, if you think you figured it out (just with pins in this example) and then you face something, your totally overcomplex MAM cannot even do: Nr. 30 with pins on the top layer. I seriously was fuming as my logic to compare each and every layer of existing pins to forward this to the pin pushers was extremely complex and took literal hours to build and in the end was deleted...
But in the end, it's funny to see other approaches that somehow have very similar steps and logics here and there. For example, the "null" to "0" converter is almost identical in my build.
And the resigning look on my face when looking at your approach handling the crystalizer test-MAM-logic as empty parts of a layer will immediately stop shape combiners from working. It took me quite some time to figure stuff like this out and it requires a whole lot of re-checking each in- and output to verify fitting data.
It's great to see there still is a video coming. Love the content.
Thanks to you I started playing Shapez 2. I allready feel overwhelmed, and I havent even unlocked the space platforms yet...
@@COKO040 once U reach space platforms IT gets way easier ITS Just Design one blueprint that works and then Go from there
Making a MAM is well beyond the average player, just enjoy it!
You shouldn't need an 'alternate' crystalizer. Instead, just have the let and right lanes swap in the crystalizer build itself. This means that at the start you take the 2 left lanes, then swap the left and right lanes. This way both builds are identical because in the 2nd build, the right lanes are now left, and are taken off as is normal.
At the random operator shape #1 (crystal / bottom right) I got this shape at level 72: crCucrSu:cmcmSuWu:cmRuWwcm:P-P-P-P-
As you can see it has pins at the top level, so the rules that you told are not quite right. Fortunately I managed to build a circuit that handles pins.
Very helpful series btw. :)
that is sooo cool but it wrecks my brain so hard at the same time ....
You could also simulate stack one layer with any full shape and destack it, so you will get layer without crystal, build layer like in normal MAM and in the end just compare if layer is the same like in the target layer, if not - add crystal, if so, pass thru.
This doesn't account for the crystal color
I figured a base for crystal design(on paper, not shapez 2), which includes all layers. So I won't watch this video until I turn the initial base design into a working design to not spoil the fun.
For the multilayer crystal MAM you will need to stack the next shape and then crystalize the next layer, that should solve it
Yep, that's what I implemented into my MAM. Build the bottom layer and crystallise. Build the second layer, stack it on top, then crystallise. Then the same for the third and the top. It works pretty well, but won't allow for gaps or pins in lower layers - not sure if those come up in the random generator.
18:16 you should also be able to combine all thrower Inputs to a single Pipe throughput wise
Idk why this works (where else would the 7200 limit be relevant?), but it does make space management so much easier. I have a compact pipe gate array too, consider blueprinting one per space belt.
"how do i check if its a crystal?"
you already checked it when you figured out the color. if its any color, its a crystal ;)
Exactly what i came to say, colour exists = at least one crystal segment
Where did you get the "rules" for the random shape milestones? It would be nice if these rules were listed in the game :) Maybe that's coming in a later update?
They are really caveats, not game rules. he made them up so his theory could be tested on a simple MAM design first.
Just a heads up. Your logic that replaces crystals with pins in the shape will fail if there are empty spaces in the layer, since the stack operations will jam. I learned that the hard way.
does the random milestone generate such shapes though?
@@Czeckie It definitely does. It won’t create layers with missing pieces below layers with crystals, but it can create them on top.
@@Czeckie Can confirm that. Had exactly the same bug in my logic platform. The empty corner can happen on the top layer.
I had to make a circuit to push the live signal forward when I encountered a missing quadrant to fix that issue when I ran into it
One of his rules for this design was that there couldn't be any blank spaces on the layer. 1:17
so my guess for todays episode; you'll have to linearize the layers before joining them with splitters.
basically each qaurter needs to be stacked individually then go through the crystal process and go back with spliters without damaging anything.
idk how to join pins and crystals with them on top of each other; but if it's possible then you can do that individually in each quarter.
otherwise you should be able to at least have one color on each layer.
Nice one! I think I'm going to use that for all the tasks before leveling up. Because I'm a good programmer (a.k.a. lazy) ;-)
when i taught cs100, one of the examples of "people who almost invented a computer but ran out of funding" was a guy who was too lazy to calculate trig tables so decided to invent a computer instead. Lazy FTW :)
wait til you get to those hexagonal shapes.
If you have a problem with the cost of blueprints, why not make more blueprint currency?
My fully automatic mam is working, 12 x180 belts, 4 layers, pins and crystals. The only problem wyth cristal is that take 15min more to swap beetwin 2 random operator shape because we have to stack layer by layer to presevre crystals. TO make this more efficiant i've créated some complex purge system and i work with quarter parts instead of full shape.
Let me know if you what to see that.
Can your MAM handle cw--cw--, or cgcrcrcg ? If so, I'd like to know how you did that.
What if a layer has multiple crystal colors ?
Keep up the good work !
According to the algorithm used to generate the random shapes, this wont occur.
@@jelster2 Those 'rules' that he establishes are his design rules, not game rules. My current shape has cw--cw-- as the bottom layer, and the 3rd layer is multi-colored crystal cgcrcrcg. The only rules that the game has is that no crystals appear in random shape 1. This is noted on the game's official wiki.
the other assumption for this build is there are no empty spaces
Can somebody explain to me why we are needing the whole change crystals to pins logic, shouldn‘t this work just fine without the pins? Just cut the crystals out in the beginning and the rest is doing its thing.
Also very nice Video, as always!!
@@Bionicmann01 I think the main reason you need the pins is in the case where the entire layer is crystal. You would end up with no shape and you need pins there to have something to crystallize.
In my MAM I’m doing what you are saying: just get rid of the crystals in the shape, and do a last check for that one case. If so, just create a layer with a single pin. Everything is working so far.
The logic around 27:47 is a bit silly, you can just NOT the color instead of having the color signal gate some other signal and then NOT that one. The value of the signal coming from the yellow receiver isn't used for anything!
I’ve managed to fit my crystallizer on a 3x3: it takes 2 of each primary as input, and outputs a full belt of any 7 colors. I’d love to know how compact that functionality can get.
I also was also wondering why didn't he do that
Make any milestone machine 😔
Unfortunately, the 'rules' for this design make this MAM not very useful. I've encountered multi-colored crystal on same layer (eg: cgcrcrcg), and crystal-only layers with holes (eg: cw--cw--). This is a good MAM design for ROS#1, but not for ROS#2 since the 'rules' are broken by the game.
13:46 - 1 Space Pipe can "consume" and support 72 Liquid Thrower. Since 4 Liquid Thrower support 1 regular Pipe 1 Space Pipe can support 18 Pipes ... or with a Pipe on each layer of a 1x1 Platform ... 1 Space Pipe = 6 Platforms (with 12 Liquid Thrower each).
Did you even calculate? Ôo
19:12 - I'm just shaking my head. Why? Why using pins if not necessary? That's a pretty convoluted approach. Just replace the crystal segment with a Null signal and fill the gaps after merging the not crystal shapez with ... crystal. Using pins unnecessarily for this purpose is a pure waste of resources.
Pins are required for a layer that is full crystal, as you cannot crystallize an empty shape.
thanks spam bot