I just uploaded the next phase, which is a 4 Layer MAM with Crystals... th-cam.com/video/2pNIZ8Jg33o/w-d-xo.html Blueprint for Phase 3 is in the Shapez 2 Discord.
Man... doood..... i just made a MAM myself for one layer no crystal/pins.. I did coloring after cutting, so i needed a train per corner lane.. this color before cutting is so simple yet a game changer. Love to watch videos about the same thing i made without a guide then finding out there is a better way, but the first time making it without a guide is essential
For your insane MAM problem controlling 5 levels with 8 global colors,.. assign 1 or 2( or 3) colors to be a 2nd digit. So all global signals would be 2 colors, the "tens" being either of the 1 or 2( or 3) 2nd digit colors or null color as 0/off and the 1st digit being either of the 7 or 6 (or 5) other colors. That would get you 14 or 18 (or 20 ), respectively, control channels to work with.
After watching your videos, I'm thinking of trying to make the logic separate from the everything, so I only need to make some basic routing block blueprints and the main logic block, because of that I think it might be overall bigger but easier to work with and design. That also means the MAM can work with 3 lanes. I thought of a few things that could speed things up: 1. making the MAM closer to the vortex, 2. using those jumping belts when the belts are straight, 3. not upgrading the capacity of the train so it starts working earlier, 4. every routing block would compare to see if the shape is enough and would send to a space belt to jump over the next unnecesary steps like colouring.
From your last video, you mentioned the turnaround time for new shapes can take upwards of 10 minutes just for shapes to make their way down the belt. From how you’ve laid out your MAM, it looks like the majority of time spent waiting is just when your shapes are traveling down the belts. Have you considered using belt jumpers to shorten that turnaround time? You have plenty of straight sections of belts where jumpers can be placed, and they move items SIGNIFICANTLY faster than normal belt lines. They can also be placed at all 3 height levels, so there are opportunities to place jumpers in a bunch of places.
I love how efficient and satisfying this look, but I feel dumb at the same time😅. I just finished my design to mix the colours (forgot about white, will add that) and it's nothing like this, but I will have fun and just watch these videos to add to my knowledge👌.
47:13 Quite simple, as I understand. You can flatten the entire painting section by using belt jumpers to the point that nothing sticks into the second layer. They send an item over an obstacle while item remain on the same layer. After that, you take the flattened design and copy on the remaining 2 layers. This way you can relatively easy flatten your entire design except the stacking phases (these actually require more work) and triple the throughput. Well, dumping items from second and third floor require a bit of extra work and space around the hole, but that's easy to arrange.
I looked into doing that this morning. Unfortunately the icon on top of the block doesn't change, which makes it very confusing from a visual standpoint. :(
wow that is interesting how you did the pins, I thought of it a different way like how you explained for detecting pins, we know when you analyze a layer it will output a signal for the north east quadrant and the color, if it analyzes one of the 4 shapes, or a crystal, it will output a positive signal for both, and if there is nothing in the space it will output a negative for both, but if it analyzes a pin, it comes out as a positive signal for something being in that quadrant, but as nothing for the color. so what I did was compare the color and shape in a xor gate, and send that signal to the next layer for each quadrant, and send it through a belt filter, if its negative it will just go to the side and continue on to stacking, but if its positive it will go forward, pin the shape, and then join back on the same track for stacking. I have not gotten to crystals either so I'm not sure how I will implement that. Its really interesting to see just how other people go about building a mam because there are so many different ways of going about it, thanks for sharing yours!
A slightly easier way to make the 4-pin shape: take the original input shape, split it, and crystalize both halves, then use the half-swapper to merge them back together. This will give a line with a solid crystal in all four quadrants and a line with the original input shape (which can be trashed). Take the 4-crystal shape and run it through the pin-pusher and then through both sides of the half-swapper. This will break the crystal, leaving just the 4-pin shape as output. I haven't figured out if there's a better way to make the 4-crystal shape yet, so the first part may still have some room for improvement.
Also, regarding crystals - I think the way to go here is to create a mold (three arbitrary shapes with a hole in the quadrant where you want the crystal to be). Then, when you go to assemble, stack the individual quadrants first, then use the half-swapper to merge things into the final shape. The idea would look something like this: Consider a shape with the following code, "cb------;Ru------;cm------" (so, three layers with only the top right quadrant filled, in the order of blue crystal, gray rectangle, magenta crystal). Step one, create a mold that looks like "--CuCuCu" (blank in the top right, gray circles elsewhere). Step two, wrap each layer with the mold (leaving the mold empty for the layer with the crystal). So, layer 1 looks like "--CuCuCu", layer 2 is "RuCuCuCu", and layer 3 is "--CuCuCu" Step three, run layer 1 through the crystallizer, giving "cbCuCuCu". Step four, stack layer 2 on layer 1. There's no crystal on this layer, so bypass the crystallizer. The shape should now be "cbCuCuCu;RuCuCuCu" Step five, stack layer 3 on top, then run the crystallizer again. This should give "cbCuCuCu;RuCuCuCu;cmCuCuCu". Step six, use the half-swapper to separate out the top right quadrant (and merge with other quadrants as necessary), giving the desired shape: "cb------;Ru------;cm------". Since you never have to drop a crystal, there's no risk of breaking, so this should be able to generate any desired single-quadrant stack. The only question is what to do about voids. I think the answer here is to use a full crystal layer instead of a mold. So, if you had a quadrant that went "blue crystal, void, magenta crystal", the for the second layer you could stack four pins ("P-P-P-P-") and crystalize (I think any color works, although you might have to avoid using the same color as the layer above or below). That way, when you use the half-swapper to assemble the final shape it will break the crystals and leave a void in those layers.
Thanks HM, I restyled to your latest and it fixed my initial stacking issues. Very streamlined, I like it. Now I have a 2 layer shape working perfectly, 3 layers....not so much. Can't figure out the logic if there isn't a fourth layer so it will not stack the 3 layers together. I'll watch a few more times and try and pause when you have the final platform in view to see if I can figure out the method to your madness :D
This comment is from my Phase 1 video, A is layer 3, B is layer 4. Basically, say you are merging A and B. If B == null, send A directly to the output of the platform. If B != null, send A to the stacker. To check if something is null, use the Equals Logic Gate, and pass in the shape on one side, and a Signal Producer with "null" as it's configuration in the other side. It's pretty much the same logic used in every stacker, from the first 2 quadrants, to the final combination of 1&2 and 3&4 layers.
@@HairymuppetGaming I think I have it sorted. Thank you for the help and the great design. I had a couple of the wires crossed and a couple of conveyor gates pointing the wrong direction. I had a blueprint with the errors and it got pasted by mistake.
@@HairymuppetGaming I just got a shape my machine is having a tough time with SuWcCb--:Cu----Ru:------Su The empty space on the bottom layer is allowing layer 2 and 3 in the top left to fall into the gap.
@@joeymack5810 You have to make each layer completely, and only then stack the full layers on top of each other. If you stack each quadrant vertically first, then overhangs will never work.
@@HairyStarcraft I know but that just happened to be my level 19 Channel 0 shape. lol. I did it manually kind of slapped together but it is good to know that it can come up as a random. You did talked about something similar explaining pins in Phase 2 video. The thought experiment on a pin dropping to a lower layer.
Because there are 4 basic shapes, pins, crystals, 4 rotations for each, and like 7 colours for each, so you've got 4 locations, each can contain one of 5 different things, that can be 1 of 7 different colours, plus an extra thing, 5 things with 7 colours is 35 possibilities, 4 locations for them increases that to 140 possibilities, and then you have pins for an extra 4 possibilities making 144 possibilities, so you'd need to have a machine capable of producing 144 belts of output, each platform section has 4 lanes and 3 floors so 12 outputs, 144/12 is 12 platforms of output which coincidentally is 12 space belts full of output This mam can produce anything it needs and it doesn't need 12 belts of output per layer to do it as it only makes what it needs, it's not a make everything machine it's a make anything machine
The odd thing is that in my normal difficulty save I started on my first of the two random shapes has red crystal on it. I'm not really sure that pins or crystals are actually gated behind levels of any kind. Could simply be a lower chance of those features. I've yet to unlock the access to those shapes but it's there all the same.
I'm up to level 57 of the first random shape. Pins only started showing (randomly) up at lvl 50+, and crystals have never shown up. I'm pretty sure only the 2nd of the 2 random shapes will ever have crystals, at least according to shapez2.wiki.gg/wiki/Operator_Level. If you are seeing red crystal on your first random shape, it could be a bug or something...
Instead of doing it manually do you have space to just put the checker up there so that it doesnt have to send the signal? I cant remember what the space is like over there.
One tiny nitpick is there's really no reason to dump full pins At 31:15 is it at all possible that the cutters could back up further down the line from overflow variance?
It's possible that one set of cutters will be delayed more than others, at it's outputs take longer to get to the stackers in the next block, but once the shapes from the other side of the stacker show up the whole thing syncs up nicely. It also helps that all 4 cutter blocks are fed by a single belt, so there's very little that the cutters can do to cause problems further up the line.
I just uploaded the next phase, which is a 4 Layer MAM with Crystals... th-cam.com/video/2pNIZ8Jg33o/w-d-xo.html
Blueprint for Phase 3 is in the Shapez 2 Discord.
Man... doood..... i just made a MAM myself for one layer no crystal/pins.. I did coloring after cutting, so i needed a train per corner lane.. this color before cutting is so simple yet a game changer.
Love to watch videos about the same thing i made without a guide then finding out there is a better way, but the first time making it without a guide is essential
For your insane MAM problem controlling 5 levels with 8 global colors,.. assign 1 or 2( or 3) colors to be a 2nd digit. So all global signals would be 2 colors, the "tens" being either of the 1 or 2( or 3) 2nd digit colors or null color as 0/off and the 1st digit being either of the 7 or 6 (or 5) other colors. That would get you 14 or 18 (or 20 ), respectively, control channels to work with.
IMO you don't need global channels at all. You can just use the same simulated buildings to decompose the random shape every time you need it.
Got recommended this right after i watched the first phase! Excited to watch
After watching your videos, I'm thinking of trying to make the logic separate from the everything, so I only need to make some basic routing block blueprints and the main logic block, because of that I think it might be overall bigger but easier to work with and design. That also means the MAM can work with 3 lanes.
I thought of a few things that could speed things up: 1. making the MAM closer to the vortex, 2. using those jumping belts when the belts are straight, 3. not upgrading the capacity of the train so it starts working earlier, 4. every routing block would compare to see if the shape is enough and would send to a space belt to jump over the next unnecesary steps like colouring.
From your last video, you mentioned the turnaround time for new shapes can take upwards of 10 minutes just for shapes to make their way down the belt. From how you’ve laid out your MAM, it looks like the majority of time spent waiting is just when your shapes are traveling down the belts. Have you considered using belt jumpers to shorten that turnaround time? You have plenty of straight sections of belts where jumpers can be placed, and they move items SIGNIFICANTLY faster than normal belt lines. They can also be placed at all 3 height levels, so there are opportunities to place jumpers in a bunch of places.
I love how efficient and satisfying this look, but I feel dumb at the same time😅. I just finished my design to mix the colours (forgot about white, will add that) and it's nothing like this, but I will have fun and just watch these videos to add to my knowledge👌.
this game is blowing out of proportions
You can just select the space belts and press "i" to clear the contents. No need to erase and replace.
47:13 Quite simple, as I understand. You can flatten the entire painting section by using belt jumpers to the point that nothing sticks into the second layer. They send an item over an obstacle while item remain on the same layer. After that, you take the flattened design and copy on the remaining 2 layers. This way you can relatively easy flatten your entire design except the stacking phases (these actually require more work) and triple the throughput. Well, dumping items from second and third floor require a bit of extra work and space around the hole, but that's easy to arrange.
Instead of NOTing equal logic blocks, you can configure them to be NOT EQUAL themselves. (As well as less than, greater than, etc.)
I looked into doing that this morning. Unfortunately the icon on top of the block doesn't change, which makes it very confusing from a visual standpoint. :(
@@HairymuppetGaming ye, they do need to change that ;)
Thanks for the update to show us how to do pins and some fixes with it, now I can let the game run in the background XD
wow that is interesting how you did the pins, I thought of it a different way
like how you explained for detecting pins, we know when you analyze a layer it will output a signal for the north east quadrant and the color, if it analyzes one of the 4 shapes, or a crystal, it will output a positive signal for both, and if there is nothing in the space it will output a negative for both, but if it analyzes a pin, it comes out as a positive signal for something being in that quadrant, but as nothing for the color.
so what I did was compare the color and shape in a xor gate, and send that signal to the next layer for each quadrant, and send it through a belt filter, if its negative it will just go to the side and continue on to stacking, but if its positive it will go forward, pin the shape, and then join back on the same track for stacking. I have not gotten to crystals either so I'm not sure how I will implement that.
Its really interesting to see just how other people go about building a mam because there are so many different ways of going about it, thanks for sharing yours!
You approach was my backup approach if I couldn't get a flat set of pins as an explicit input into the system. :D
Looks great, man! Really polished it since the last update!!
A slightly easier way to make the 4-pin shape: take the original input shape, split it, and crystalize both halves, then use the half-swapper to merge them back together. This will give a line with a solid crystal in all four quadrants and a line with the original input shape (which can be trashed). Take the 4-crystal shape and run it through the pin-pusher and then through both sides of the half-swapper. This will break the crystal, leaving just the 4-pin shape as output.
I haven't figured out if there's a better way to make the 4-crystal shape yet, so the first part may still have some room for improvement.
Also, regarding crystals - I think the way to go here is to create a mold (three arbitrary shapes with a hole in the quadrant where you want the crystal to be). Then, when you go to assemble, stack the individual quadrants first, then use the half-swapper to merge things into the final shape. The idea would look something like this:
Consider a shape with the following code, "cb------;Ru------;cm------" (so, three layers with only the top right quadrant filled, in the order of blue crystal, gray rectangle, magenta crystal).
Step one, create a mold that looks like "--CuCuCu" (blank in the top right, gray circles elsewhere).
Step two, wrap each layer with the mold (leaving the mold empty for the layer with the crystal). So, layer 1 looks like "--CuCuCu", layer 2 is "RuCuCuCu", and layer 3 is "--CuCuCu"
Step three, run layer 1 through the crystallizer, giving "cbCuCuCu".
Step four, stack layer 2 on layer 1. There's no crystal on this layer, so bypass the crystallizer. The shape should now be "cbCuCuCu;RuCuCuCu"
Step five, stack layer 3 on top, then run the crystallizer again. This should give "cbCuCuCu;RuCuCuCu;cmCuCuCu".
Step six, use the half-swapper to separate out the top right quadrant (and merge with other quadrants as necessary), giving the desired shape: "cb------;Ru------;cm------".
Since you never have to drop a crystal, there's no risk of breaking, so this should be able to generate any desired single-quadrant stack. The only question is what to do about voids. I think the answer here is to use a full crystal layer instead of a mold. So, if you had a quadrant that went "blue crystal, void, magenta crystal", the for the second layer you could stack four pins ("P-P-P-P-") and crystalize (I think any color works, although you might have to avoid using the same color as the layer above or below). That way, when you use the half-swapper to assemble the final shape it will break the crystals and leave a void in those layers.
Thanks HM, I restyled to your latest and it fixed my initial stacking issues. Very streamlined, I like it. Now I have a 2 layer shape working perfectly, 3 layers....not so much. Can't figure out the logic if there isn't a fourth layer so it will not stack the 3 layers together. I'll watch a few more times and try and pause when you have the final platform in view to see if I can figure out the method to your madness :D
This comment is from my Phase 1 video, A is layer 3, B is layer 4.
Basically, say you are merging A and B. If B == null, send A directly to the output of the platform. If B != null, send A to the stacker. To check if something is null, use the Equals Logic Gate, and pass in the shape on one side, and a Signal Producer with "null" as it's configuration in the other side.
It's pretty much the same logic used in every stacker, from the first 2 quadrants, to the final combination of 1&2 and 3&4 layers.
@@HairymuppetGaming I think I have it sorted. Thank you for the help and the great design. I had a couple of the wires crossed and a couple of conveyor gates pointing the wrong direction. I had a blueprint with the errors and it got pasted by mistake.
@@HairymuppetGaming I just got a shape my machine is having a tough time with SuWcCb--:Cu----Ru:------Su
The empty space on the bottom layer is allowing layer 2 and 3 in the top left to fall into the gap.
@@joeymack5810 You have to make each layer completely, and only then stack the full layers on top of each other. If you stack each quadrant vertically first, then overhangs will never work.
@@HairyStarcraft I know but that just happened to be my level 19 Channel 0 shape. lol. I did it manually kind of slapped together but it is good to know that it can come up as a random. You did talked about something similar explaining pins in Phase 2 video. The thought experiment on a pin dropping to a lower layer.
Why not make every single shape component in every colour, store them all, and then just pull the ones you need and combine them?
Because there are 4 basic shapes, pins, crystals, 4 rotations for each, and like 7 colours for each, so you've got 4 locations, each can contain one of 5 different things, that can be 1 of 7 different colours, plus an extra thing, 5 things with 7 colours is 35 possibilities, 4 locations for them increases that to 140 possibilities, and then you have pins for an extra 4 possibilities making 144 possibilities, so you'd need to have a machine capable of producing 144 belts of output, each platform section has 4 lanes and 3 floors so 12 outputs, 144/12 is 12 platforms of output which coincidentally is 12 space belts full of output
This mam can produce anything it needs and it doesn't need 12 belts of output per layer to do it as it only makes what it needs, it's not a make everything machine it's a make anything machine
I'm feeling dumb now 😅
that % thing isnt accurate
its baed on the progress players made during beta. its not actually comparing your stats to the current player base.
The odd thing is that in my normal difficulty save I started on my first of the two random shapes has red crystal on it. I'm not really sure that pins or crystals are actually gated behind levels of any kind. Could simply be a lower chance of those features. I've yet to unlock the access to those shapes but it's there all the same.
I'm up to level 57 of the first random shape. Pins only started showing (randomly) up at lvl 50+, and crystals have never shown up. I'm pretty sure only the 2nd of the 2 random shapes will ever have crystals, at least according to shapez2.wiki.gg/wiki/Operator_Level. If you are seeing red crystal on your first random shape, it could be a bug or something...
@@HairymuppetGamingAh...I bet is is the 2nd shape. I'm not at my pc...it's late and I'm dealing with covid as well...so memory was likely foggy.
@@bagofholding ouch sorry for you
I came out of covid with my legs partially atrophied from inactivity
Instead of doing it manually do you have space to just put the checker up there so that it doesnt have to send the signal? I cant remember what the space is like over there.
One tiny nitpick is there's really no reason to dump full pins
At 31:15 is it at all possible that the cutters could back up further down the line from overflow variance?
It's possible that one set of cutters will be delayed more than others, at it's outputs take longer to get to the stackers in the next block, but once the shapes from the other side of the stacker show up the whole thing syncs up nicely. It also helps that all 4 cutter blocks are fed by a single belt, so there's very little that the cutters can do to cause problems further up the line.
Audio too soft again mate :(
Yeah, it's a constant process to get it right...