Good news everyone, thanks to all your comments I finally understand the paint first then cut logic! If I paint first I only need 1 launcher worth of input to output a full space belt of my desired quarter!! It's finally clicked... Thanks for all your persistence, I know I can be slow sometimes! Will do a redesign episode very soon, prior to copying out my design for the 4 layers
If you somehow reuse this concept you could actually make it smaller by having a train with all the colours and outputing it directly into one of these modules in a 1x1 or 1x2 platform and the exit should be just the space paint belt directly so it would be, train ->unloader->modules->output space belt. Painting all shapes first is more efficient anyway and you have to make sure you have paint all the time so the pipe gate is useless for me. Hope it helps👍
I really like the way you think out loud as you design. Thanks for the series. I'm probably not the first to say this (sorry i didn't read all of the other comments) but if you color then chop, you use 1/4 paint and painters. Every quad shape is 4 single shapes condensed.
Please do not cut your videos shorter. I love your explanations. Your external thinking is absolutely amazing. I hate when people are like: ok I cut out the boring and we got this. But like; HOW did you get that?! I love all the explanations.
Haha I don't plan to don't worry. I will skip some boring parts, but anytime I need to logic out anything new you'll get to see exactly how my brain works
Definitely. EP06 - Add pinning and crystalising to 1 layer EP07 - Create a input system using trains since we need 4x the input we need now EP08 - Duplicate our single layer process to make 4 layers and stack
Enjoying this series a lot! To the commenters, it’s not always about perfect efficiency. But to help explain what they’re saying, if you wanted to end up with 4 red quarter circles, you have a choice: Split it into 4 quarters, then paint each one. Or paint it once, then split into quarters. The outcome is the same, but splitting later uses less paint.
The part I don't understand is why I want 4 quarter red circles... What if my shape is 1 red quarter 1 blue quarter 1 white quarter 1 uncolored quarter?
@@DataEngineerPlays in the MAM I’m working on, I rotate the other 3 quarters around so I can 4x the throughput of that one red corner I need. Same thing you’re doing I think, you’re also not throwing away the other 3 quarters :)
@@yohsuzuki But isn't your cutting operation super convoluted? To take a full shape, remove 1 quarter... then have 3/4 of a shape, remove another quarter... half a shape left... split that in half... Feels like alot of work, maybe I'm overthinking... My preference is definitely split into 4 quarters first... then belt spaghetti... But hey I could be wrong, would love to see your MAM one day when I finish my series!
@@DataEngineerPlays Splitting into quarters for me happens in the size of a single small space platform taking 3 belts in on 3 floors and outputting 12 belts out on 3 floors! It’s a split in half, rotate, then split in half again
hey Data? polite reminder, dont be scared to cut/ copy and paste a whole bunch of stuff for your builds if required. if i was building this id have wireless recievers on each input and id have the gate on the beginning of the lauchers. However that is a really good build :)
Honestly have rewrote this comment like 5 times but the way you are doing it works. Basic guise is one way is shape efficient and one way is paint efficient. If you pre paint you wont use as much paint. 4 space belts of each shape minimum for pre painting. Basically a full shape for each quadrant allowing you to use only the 4 painters you use now for 4 times the output. So 16 shape belts 28 paint belts to an output of 4 belts opposed to 4 shape and still 28 paint belts to 1 output. If i did my math right. There is some added time due to rotatons. Not sure if this makes sense but keep doing what you are doing if you are having fun cause that's the main thing. It's a game so play it the way you want to have fun. Tried my best to explain. Legit play how you want will watch. Thanks for reading. If you need and more explanation feel free to ask. Will answer any questions to best of my ability. Tried to make this as readable as possible.
Yeah what you said makes perfect sense, its shape efficiency vs paint efficiency right? If my maths is wrong eventually I'll figure it out! haha I might be a bit slow but I'll get there in the end. And again atm the focus is building a functional mam not a fully 100% optimised mam
I feel like the color array could be compacted a lot more using trains and fluid wagons. Just fill it up elsewhere and route it in. Takes much less space than having to route every belt into every painter array. (IDK I'm watching these in order so LMK if you've already tried this)
I don't think you need to discard 3 forth of your said painted shape, you should just cut/rotate/etc and get 4x quarter throuput (that should match the one you get now). Also I see you implementing the bypass to get your shapes to go in the painter when you have plenty of room on the color selector platform. I know it kinda break the "modular" design but to be honnest thos two modules will be attached to each other everywhere in your mam, it kinda makes sense to merge them on this gigantic platform
12:25 T-shaped one. You won't be able to fit belts with bypass in, though. And you should seriously consider replacing all space pipes with a single train with 7 wagons. And there should be enough space to cut down incoming colors to RGBW and implement basic mixing for the remaining 3 on the same platform.
Just pointing out in case you've missed this feature like I have, but if you click on the Equals gate you can modify the comparator, so if for example it makes your wires more readable to have it act as a NotEquals, you can do that
Haha!😂 totally get the confusion about cutting before or after painting. Let's see if i can explain the process. It hurt my brain too when i got it! For simplicity's sake i assume only 1 floor as it can easily be copy pasted for the others. Goal of the MAM for 1 quarter of 1 layer: get a corner of a colored shape on 4 belts for maximum throughput Method: Step 1: Select the full shape that match this corner. Important: Because you will cut it in 4 later, you only need to select one of your 4 belts. The other 3 belts can keep moving and feed the selector of another corner. Step 2: Color your single belt. Since it is only one belt, you only need 1 paint launcher. Like in the previous step, the other 3 launcher can be used to passthrough color to another corner Step 3: Cut yoir colored shape into corner, realign them so they are all the same corner you want. You know have 4 belts of one colored corner, and it cost 1 paint launcher and 1 belt, instead of 1 space belt of each. This method allow to compress the MAM and limit the pain of feeding multiples inputs!
Thanks for the details. I feel like everyone is saying the same thing.. but what I'm hearing is Step 1: Cost = 4 input belts of shapes - of which we select 1 but the other 3 are also "consumed" Step 2: Cost = 7 input belts of color - of which we consume 1 but the other 6 are also "consumed" Step 3: Now we have 4 quarters one of which we need 3 of which we don't need (am I destroying these or passing them on for recoloring?) Repeat step 1-3 3 more times for the other quarters... totalling 4x4 shape input ?(4x more than I have) 4x7 fluid input (same as what I have) Riddle me this: how many input belts of color and shape do you need to make this shape Top Left: Red Circle Top Right: Red Square Bottom Left: Red Star Bottom Right: Red Emerald You would need at least 4 belts of shapes and 28 belts of color?
Omg wait nvm I just got what everyone is saying, I'm using a full space belt, but you can achieve the same thing with 1/4 of the input.. 1 belt launcher.... I see I see... wtf... why did that take me so long... in answer to my own question to paint that shape I would only need 1 belt of red not 4.... because 1 fluid launcher of red could paint out a full space belt of a quarter shape after I cut it omg thank you finally... I understand
If you pick the shape first and then color and quarter it, you will need only 1 belt to produce 4 belts of quarters, so you could have one space belt containing every shape on one belt each. You will also need only a quarter amount of the paint, since it uses the same amount to paint a quarter then a full shape.
trains can solve your conundrum also when you quarter the painted shape you can just rotate the other quarters to be uniform is it is a full belt of full shapes in for a full 12 lane belt of quarter shapes, this way you dont have to get rid of the rest of anything
I don’t know if I’m looking at this wrong, but I think one of your quarters (the bottom right quarter I think) isn’t being rotated to the correct orientation. I can see 2 rotator modules for the inside 2 quarters, but none for the outside 2. I know you’re probably changing your design anyways, but I thought I’d make you aware of this 😅
So far I find it easier to block the flow on the space belts and space pipes instead of doing it on a single platform. I send the color out on a global signal and just have a 1x1 build with a receiver and the small comparison logic that blocks or open the flow for all 12 belts/pipes, and then just merge the space belts/pipes afterward.
will you run out of signal channels doing that at some point? Think 4 layers x 4 quarters = 16 global signals you need to send? There's only 7 global channels right?
Heyo Allan! Miss me? I heard your comment about me calling you an idiot, which I never have! I've playfully 'taken the piss', but I've never intended to be insulting... First off, your wife must be a saint! So tolerant of all the extension cables running across her living room floor and such! If this were 3D printing, we'd say that you're thinking in 2.5D and not 3D. Wires in the attic, pipes in the basement. You've built lots of stacker and mixer blueprints that require 6 stations on 2 floors. You can break out of the "3 floors of 4 stations" mindset. You'll have a lot more room to make logic circuits and drop the wires down twice. That's a lot more space efficient than remaking the circuits on 2 or 3 floors. You have a 3x3 platform with loads of empty space. Know what's more efficient than running 7 space pipes of color? Running 3 space pipes of color and mixing in all that empty space. If you have a floor dedicated to wires, then you even have room for more complex OR logic. Signal Analyzer Color wire to 4 EQ tests (=Color-b, = -c, = -m, = -w) OR those results and turn on blue pipe. Gate Outpipe to mixers based on NOT -b. One pipe can feed 4 launchers. It's more space efficient to have one gate on the input instead of 4 gates on the output. Extractors have 3 floors too. People have made extractors that do the swaps to recombine the 4 shapes into 4 homogenous ones. It's possible to homogenize the pieces, sort them into the correct space belt, and put them in the correct train car to ship them back to the MAM's staging area. The flying animation plays faster, but it has a little bit of ramp up time before catapulting the piece. The total time it takes the piece to move the 6 lanes (through the launcher, over 4 lanes, and through the catcher) is the same as just riding a conveyor the same distance. You noted that your MAM is only as fast as your slowest lane. Painting and Stacking are the 2 slowest machines. It's faster and less distance to paint the solid once and cut it into quarters than it is to paint the 4 quadrants individually.
Ah I love that color mixing idea... why didn't I think of that, all that empty space I could just run in the primary colors and make the secondary colors on the fly. But yes there is going to be a lot of optimisation work at the end of this. But I don't think my brain has the compute power to optimise at the same time as building the functionality. A lot of the time you get an idea... then you tunnel into that idea when there may be a super obvious and better way to do something. Hence love all your comments, it always opens my mind to new possibilities
You dont waste any quarter of shape if you paint it first. You can easily get same rotated quarters with adding rotaters to cutters output. Cutter have 2 outputs and you have 3 rotater types. Adding one rotater to cutter outputs gives you needed rotated quarter. For examle your random shape has one blue quarter circle at top left, paint whole circle first than cut into quarters and add rotaters to cutter outputs to rotate them to top left. Now you have 4 blue top left quarters, all same rotation and all pre painted. Use them for 4 belts output of final shape.
@@DataEngineerPlays the 4 quarters become the same quarter that you needed for that one quarter that was coloured. therefore you 12 belts of squares become 48 belts of quarters of the one quarter you need for that shape on that level the only thing you have to do is rotate the other 3 quarters from that shape to the same orientation that is required for that quarter.
@@DataEngineerPlays You have 3 floors and 4 belts for each shape. That means you need to produce 12 belts of random milestone shape. 12 shapes have (4x12) 48 quarters so you need to produce 12 quarters (having same rotation) of random milestone shape for 12 belts of final output.I am sure you will find a way to use of all top left 4 blue quarters in that example. In fact you need 12 top left blue quarters for 12 belts of final output. Look at your current setup, it is painting 12 same quarters on different platforms. Dont try to produce final shape seperately, focus on quarters. Firts produce 48 quarters (assuming the layer has no gaps). That minimize the possibilities so less belt spagetti. First produce 4x12 quarters of random milestone shape, than stack 12 top right quarters with 12 bottom right quarters to get 12 right half of the shape than stack left quarters to get left half and finally stack two halfes to get 12 belt final shape. I would recomend to use stackers instead of swappers, it is less confusing and doing the same job. Also using stackers simplifies to setup wire logic.
@@gregmeeuwissen6393 Yes it's finally clicked in my head what all of you guys are talking about.... I need to stop thinking of space belts as 1 object that needs to processed together... I can split 1 space belt into 4 units of 3, each unit can be painted THEN quartered to save my paint consumption by 4x I got you guys now!
guess who will hit the 1k follower soon, im watching you since 225 followers (youtube algorithm) startet my abo at 906 or something, i believe you are blessed, i saw you pushed sometimes 2 episodes out, better save them und and do daily one episode at a fixed time, use extra episodes to get a day off, before you suffer in mental damage, you doin great and fast, even your episodes feel sometimes rushed and they are not even short (which is nice), but don't overload yourself your data engineer skills and advise are nice to watch
Haha much love for my early followers! Thanks for viewership. Yeah I think I'll probably do 1 episode a day. I'm still super new at all this so just trying to find the right cadence, a lot of experimenting on what time and how many videos to release etc. PS: I'm blessed to have this amazing audience!
Future me: Did or did he not notice one of the inputlanes (per level) for green and cyan is set to outputs on his colorselecting platform? Awesome project btw!
i dunno if it was mentionend, but after we got the paint gates, wouldnt it be possible to create a dynamik paint mixer ? reducing the painting to the three starting colors, reducing even more paint lanes ?
It only reduces it slightly right? Instead of needing 4 belts of r g b c y m w (28 total belts) You need 4 belts of r g b 2 belts of b 2 belts of g to make c 2 belts of r 2 belts of g to make y 2 belts of b 2 belts of r to make m 2 belts of b 1 belt of r 1 belt of g to make which is still 28 incoming belts right? Just instead of the mixed colors your just importing r g b instead
@@DataEngineerPlays my operator level wasnt that highe jet so i had the assumption, if you shove color into a form analyzer, it could output the colors it need to mix, like the unstacker just with color "the uncolorrizer ?" After testing i noticed that the form analizer can not display on a y that it needs like r/g i would also be happy with a color analizer if we have it for shapes already >.< but that might be a future update. so the belt reduction was under that assumption, you would have like one output pipe and rgb as input, with something like a color analyizer/uncolorizer. p.s. also the MAM series encourages me to keep on gaming to unlock everything.
I am still watching the video so I might not have seen it, but you should need only one switch for each collor output in the platform right? You shouldn't need to switch each paint launcher since the output joins no?
a gate has a fluid throughput of 1800L/m when I merge the 4 launchers into 1 pipe, it becomes a pipe of 7200L/m. So that's the reason to split again and use 4 gates
@@DataEngineerPlays I understand that, but when would you use any of the lauchers individually? You will always either use 4 lauchers or non in you color selection as far as I understand. Anyway, this an exercise... it works either way I believe.
I got a question 🤔 You take 4 inputs of colour and merge in to one pipe. You then split that one pipe in to 4 pipe selectors. And after that you merge the 4 exits from the pipe selectors back in to one pipe. Is it that because of a limit to the capacity of the pipe selector ? If so would it be enough to just split it in to 2 pipe selectors ?
why are you thinking of disposing 3 quarters at the end. using cutters and rotators we can just rotate them in the orientation we want and get all 4 quarters at the end. (Also in the currernt version you can check with belt reader that if a single quarter is neeeded, let's say circle, full circles move at 45/m only or 1 belt is moving and other 3 are stopped, other 3 quarters are stopped at the filter anyway) so even if you divide input shape belts into all 4 platforms we get a full throughput of 180/m at final point of quarters (Let's sayeven if full circle is needed now, now filters perform at 45/m since input belt is divided but at the end we get 4 times the quarters so that is still 180/m) Also painting full and quarters need same amount of paint so we can branch paint also to all four paltforms, paint at 45/m still get quarters at 180/m. Also you made 2 input fluid launchers in the wrong direction in paint selector platform
In my head I might be overcomplicating the process of taking a quarter out of the shape and passing the remaining 3/4 on to the next phase. But it does make total sense. I feel like then it becomes very similar to what I did?
@@DataEngineerPlays When you make quarters, you are not throwing away the other three quarters, you are cutting, rotating, cutting, rotating. Each cutting step takes one lane of input and makes two lanes of output. Each quarter of the input shape becomes a whole lane of output quarter-shape, so you go from one lane of input to four lanes of output. If you paint first, you paint all four quarters at once, with the cost of one lane worth of paint. If you paint second, you have to paint four lanes, each with its own paint consumption.
I think your trouble with understanding the reasoning behind the other order of operations (color first, cut later) comes from you have been focusing on having full space belts at all times. But that's not needed. My suggestion (and some others I've seen) has been 1 space belt of mixed input shapes for the shape selector having 3 plain belts of each shape. 3 belt (plain belts, not space belts) output. 3 belt input to painter, 3 belt output. Painter is tiny, because it only deals with 3 (not 12) belts. 3 belts input to cutter/rotator. One full space belt output. No extras, they are all used for the same corner.
You could of course upgrade to having full space belts of each shape as input. Then you would just deliver 4 full space belts to the vortex. (I would suggest with a train launcher, then.)
@@DataEngineerPlays Fair enough. That particular idea also only works with belts, not trains. The main point was to only have 3 plain belts in the first part of the MaM because that gives an output of one full space belt at the end - and to keep it limited to those 3 belts for as long as possible.
End of the video felt called out cause thats literally what ive done i have 8 space belts of each shape and 4x7 belts of color. Currently losing my mind with the complex of doing 24 belt out.
Sorry didn't mean to call anyone out, it's just not the way I want to build my MAM, but I'm still in the very early stages, so I might run into a reason that makes me want to do it your way! who knows at this stage
@@theinnocenttomato9836 Yeah I think I'm just misunderstanding you guys, because a few people have told me to color the shape first then cut it into quarters.. But I'm not seeing how that helps...
@DataEngineerPlays I'm with you on this the only way I can see painting full shape being better is if you have advanced logic in the beginning that analyzes the entire layer and somehow identifies partial shapes like halves and 3/4 etc.. but even then not sure if it would be more efficient
28 colors? I see input of 4 shapes and 7 colors when you stack 4 layers you`ll have 16 shapes and 28 colors. you`re almost using the minimum amount of inputs. the next best is to input 3 colors and mix dynamicaly inside the mam the correct color.
Your logic is wrong about the 8x4 belts. You don't need to find 8 full belts for each shape, because your end product is not consuming that amount. What I've said all along is that having all the inputs ready, with only a need of picking the correct quarter shape 4 times and then assembly, is much much shorter than what you are doing. You are doing what springs to mind. Your logic is 1. I need a shape. 2. I need to cut it. 3. I need to color it. 4. Then assembly. And there's nothing wrong with it. Your logic will work, it just not efficient at all. You will be spending 20 minutes when there's a new request, just waiting for the new shape to hit the vortex. My logic is how do I supply every possible shape combination in the first step, because here, waiting is not a problem. It doesn't matter how long the belts are or how far they have to travel. As soon as everything there is ready, it will stay filled constantly. So my logic is: 1. Supply all possible quarters in all colors 2. Assemble it and deliver. I know there will be pins and crystals further down the road. But they can have their own MAM, if inputting then becomes a problem. My logic all the time is NOT waiting and wasting time for the "dead shapes" to clear out. And also keeping the travel from the assembly to the vortex at an absolute minimum. Hope this helps. And just to clarify once again. You are not doing it wrong, it's just not what I would do.
I agree with your logic about input belts not mattering, and to reduce the time spent in the MAM itself. But where you're losing me is how are you "1. Supply all possible quarters in all colors" but not needing "8x4 belts"?
@@DataEngineerPlays I mean you need the belts. But how full they are is not important, as you in theory only need as much as you are producing at the end.
Good news everyone, thanks to all your comments I finally understand the paint first then cut logic!
If I paint first I only need 1 launcher worth of input to output a full space belt of my desired quarter!! It's finally clicked... Thanks for all your persistence, I know I can be slow sometimes! Will do a redesign episode very soon, prior to copying out my design for the 4 layers
If you somehow reuse this concept you could actually make it smaller by having a train with all the colours and outputing it directly into one of these modules in a 1x1 or 1x2 platform and the exit should be just the space paint belt directly so it would be, train ->unloader->modules->output space belt. Painting all shapes first is more efficient anyway and you have to make sure you have paint all the time so the pipe gate is useless for me. Hope it helps👍
Your videos are easily the best MAM videos. I love your pace and your voice. Instant sub.
I really like the way you think out loud as you design. Thanks for the series.
I'm probably not the first to say this (sorry i didn't read all of the other comments) but if you color then chop, you use 1/4 paint and painters. Every quad shape is 4 single shapes condensed.
Thanks! I've been told many times about my order of operations, took me a while to understand but I finally do. Implemented in episode 08 stay tuned!
@@DataEngineerPlays It's a great series!
Please do not cut your videos shorter. I love your explanations. Your external thinking is absolutely amazing. I hate when people are like: ok I cut out the boring and we got this.
But like; HOW did you get that?!
I love all the explanations.
Haha I don't plan to don't worry. I will skip some boring parts, but anytime I need to logic out anything new you'll get to see exactly how my brain works
A train would save so many platform units especially with the paint.
Definitely.
EP06 - Add pinning and crystalising to 1 layer
EP07 - Create a input system using trains since we need 4x the input we need now
EP08 - Duplicate our single layer process to make 4 layers and stack
@@DataEngineerPlays Okay cool. I've been watching these in order and wondering "when's he gonna try trains?" Answer: the very next episode. Cool.
@@ToidiZuzuDiPuts Actually the optimisation process has taken longer than expected so maybe in a few episodes!
Enjoying this series a lot! To the commenters, it’s not always about perfect efficiency.
But to help explain what they’re saying, if you wanted to end up with 4 red quarter circles, you have a choice:
Split it into 4 quarters, then paint each one. Or paint it once, then split into quarters.
The outcome is the same, but splitting later uses less paint.
The part I don't understand is why I want 4 quarter red circles... What if my shape is 1 red quarter 1 blue quarter 1 white quarter 1 uncolored quarter?
@@DataEngineerPlays in the MAM I’m working on, I rotate the other 3 quarters around so I can 4x the throughput of that one red corner I need. Same thing you’re doing I think, you’re also not throwing away the other 3 quarters :)
@@yohsuzuki But isn't your cutting operation super convoluted? To take a full shape, remove 1 quarter... then have 3/4 of a shape, remove another quarter... half a shape left... split that in half... Feels like alot of work, maybe I'm overthinking...
My preference is definitely split into 4 quarters first... then belt spaghetti...
But hey I could be wrong, would love to see your MAM one day when I finish my series!
@@DataEngineerPlays Splitting into quarters for me happens in the size of a single small space platform taking 3 belts in on 3 floors and outputting 12 belts out on 3 floors! It’s a split in half, rotate, then split in half again
@@yohsuzuki yes I can imagine that! Thanks for the insight
hey Data? polite reminder, dont be scared to cut/ copy and paste a whole bunch of stuff for your builds if required. if i was building this id have wireless recievers on each input and id have the gate on the beginning of the lauchers. However that is a really good build :)
First hour club sign in here
by pressing like 😎
👇
No, I don't think I will.
Rawer always driving up my video engagement every video!
@@DataEngineerPlays 😁
Honestly have rewrote this comment like 5 times but the way you are doing it works. Basic guise is one way is shape efficient and one way is paint efficient.
If you pre paint you wont use as much paint. 4 space belts of each shape minimum for pre painting. Basically a full shape for each quadrant allowing you to use only the 4 painters you use now for 4 times the output.
So 16 shape belts 28 paint belts to an output of 4 belts opposed to 4 shape and still 28 paint belts to 1 output. If i did my math right. There is some added time due to rotatons.
Not sure if this makes sense but keep doing what you are doing if you are having fun cause that's the main thing. It's a game so play it the way you want to have fun.
Tried my best to explain. Legit play how you want will watch. Thanks for reading. If you need and more explanation feel free to ask.
Will answer any questions to best of my ability. Tried to make this as readable as possible.
Yeah what you said makes perfect sense, its shape efficiency vs paint efficiency right? If my maths is wrong eventually I'll figure it out! haha I might be a bit slow but I'll get there in the end. And again atm the focus is building a functional mam not a fully 100% optimised mam
~27:00~ pained my heart with so much space on the 3x3 platform 😭
I feel like the color array could be compacted a lot more using trains and fluid wagons. Just fill it up elsewhere and route it in. Takes much less space than having to route every belt into every painter array.
(IDK I'm watching these in order so LMK if you've already tried this)
I don't think you need to discard 3 forth of your said painted shape, you should just cut/rotate/etc and get 4x quarter throuput (that should match the one you get now). Also I see you implementing the bypass to get your shapes to go in the painter when you have plenty of room on the color selector platform. I know it kinda break the "modular" design but to be honnest thos two modules will be attached to each other everywhere in your mam, it kinda makes sense to merge them on this gigantic platform
Yes I've figured all this out by EP08 I believe
12:25 T-shaped one. You won't be able to fit belts with bypass in, though.
And you should seriously consider replacing all space pipes with a single train with 7 wagons.
And there should be enough space to cut down incoming colors to RGBW and implement basic mixing for the remaining 3 on the same platform.
Yep will definitely be doing trains especially when we need 4x this input for our 4 layers
Just pointing out in case you've missed this feature like I have, but if you click on the Equals gate you can modify the comparator, so if for example it makes your wires more readable to have it act as a NotEquals, you can do that
Haha!😂 totally get the confusion about cutting before or after painting. Let's see if i can explain the process. It hurt my brain too when i got it!
For simplicity's sake i assume only 1 floor as it can easily be copy pasted for the others.
Goal of the MAM for 1 quarter of 1 layer:
get a corner of a colored shape on 4 belts for maximum throughput
Method:
Step 1:
Select the full shape that match this corner. Important: Because you will cut it in 4 later, you only need to select one of your 4 belts. The other 3 belts can keep moving and feed the selector of another corner.
Step 2:
Color your single belt. Since it is only one belt, you only need 1 paint launcher. Like in the previous step, the other 3 launcher can be used to passthrough color to another corner
Step 3:
Cut yoir colored shape into corner, realign them so they are all the same corner you want. You know have 4 belts of one colored corner, and it cost 1 paint launcher and 1 belt, instead of 1 space belt of each.
This method allow to compress the MAM and limit the pain of feeding multiples inputs!
Thanks for the details. I feel like everyone is saying the same thing.. but what I'm hearing is
Step 1: Cost = 4 input belts of shapes - of which we select 1 but the other 3 are also "consumed"
Step 2: Cost = 7 input belts of color - of which we consume 1 but the other 6 are also "consumed"
Step 3: Now we have 4 quarters one of which we need 3 of which we don't need (am I destroying these or passing them on for recoloring?)
Repeat step 1-3 3 more times for the other quarters... totalling 4x4 shape input ?(4x more than I have) 4x7 fluid input (same as what I have)
Riddle me this: how many input belts of color and shape do you need to make this shape
Top Left: Red Circle
Top Right: Red Square
Bottom Left: Red Star
Bottom Right: Red Emerald
You would need at least 4 belts of shapes and 28 belts of color?
Omg wait nvm I just got what everyone is saying, I'm using a full space belt, but you can achieve the same thing with 1/4 of the input.. 1 belt launcher.... I see I see... wtf... why did that take me so long... in answer to my own question to paint that shape I would only need 1 belt of red not 4.... because 1 fluid launcher of red could paint out a full space belt of a quarter shape after I cut it omg thank you finally... I understand
If you pick the shape first and then color and quarter it, you will need only 1 belt to produce 4 belts of quarters, so you could have one space belt containing every shape on one belt each. You will also need only a quarter amount of the paint, since it uses the same amount to paint a quarter then a full shape.
Yes that concept has just clicked in my head, I can reduce my paint consumption by 4x
trains can solve your conundrum also when you quarter the painted shape you can just rotate the other quarters to be uniform is it is a full belt of full shapes in for a full 12 lane belt of quarter shapes, this way you dont have to get rid of the rest of anything
I don’t know if I’m looking at this wrong, but I think one of your quarters (the bottom right quarter I think) isn’t being rotated to the correct orientation. I can see 2 rotator modules for the inside 2 quarters, but none for the outside 2. I know you’re probably changing your design anyways, but I thought I’d make you aware of this 😅
Yes I notice this at the start of the next episode, you are indeed correct!
So far I find it easier to block the flow on the space belts and space pipes instead of doing it on a single platform. I send the color out on a global signal and just have a 1x1 build with a receiver and the small comparison logic that blocks or open the flow for all 12 belts/pipes, and then just merge the space belts/pipes afterward.
will you run out of signal channels doing that at some point? Think 4 layers x 4 quarters = 16 global signals you need to send? There's only 7 global channels right?
Heyo Allan! Miss me? I heard your comment about me calling you an idiot, which I never have! I've playfully 'taken the piss', but I've never intended to be insulting...
First off, your wife must be a saint! So tolerant of all the extension cables running across her living room floor and such! If this were 3D printing, we'd say that you're thinking in 2.5D and not 3D. Wires in the attic, pipes in the basement. You've built lots of stacker and mixer blueprints that require 6 stations on 2 floors. You can break out of the "3 floors of 4 stations" mindset. You'll have a lot more room to make logic circuits and drop the wires down twice. That's a lot more space efficient than remaking the circuits on 2 or 3 floors.
You have a 3x3 platform with loads of empty space. Know what's more efficient than running 7 space pipes of color? Running 3 space pipes of color and mixing in all that empty space. If you have a floor dedicated to wires, then you even have room for more complex OR logic. Signal Analyzer Color wire to 4 EQ tests (=Color-b, = -c, = -m, = -w) OR those results and turn on blue pipe. Gate Outpipe to mixers based on NOT -b.
One pipe can feed 4 launchers. It's more space efficient to have one gate on the input instead of 4 gates on the output.
Extractors have 3 floors too. People have made extractors that do the swaps to recombine the 4 shapes into 4 homogenous ones. It's possible to homogenize the pieces, sort them into the correct space belt, and put them in the correct train car to ship them back to the MAM's staging area.
The flying animation plays faster, but it has a little bit of ramp up time before catapulting the piece. The total time it takes the piece to move the 6 lanes (through the launcher, over 4 lanes, and through the catcher) is the same as just riding a conveyor the same distance.
You noted that your MAM is only as fast as your slowest lane. Painting and Stacking are the 2 slowest machines. It's faster and less distance to paint the solid once and cut it into quarters than it is to paint the 4 quadrants individually.
Ah I love that color mixing idea... why didn't I think of that, all that empty space I could just run in the primary colors and make the secondary colors on the fly.
But yes there is going to be a lot of optimisation work at the end of this. But I don't think my brain has the compute power to optimise at the same time as building the functionality.
A lot of the time you get an idea... then you tunnel into that idea when there may be a super obvious and better way to do something. Hence love all your comments, it always opens my mind to new possibilities
You dont waste any quarter of shape if you paint it first. You can easily get same rotated quarters with adding rotaters to cutters output. Cutter have 2 outputs and you have 3 rotater types. Adding one rotater to cutter outputs gives you needed rotated quarter.
For examle your random shape has one blue quarter circle at top left, paint whole circle first than cut into quarters and add rotaters to cutter outputs to rotate them to top left. Now you have 4 blue top left quarters, all same rotation and all pre painted. Use them for 4 belts output of final shape.
But what if I don't need 4 blue quarters... what if my shape is 4 diferent colored quarters?
@@DataEngineerPlays the 4 quarters become the same quarter that you needed for that one quarter that was coloured. therefore you 12 belts of squares become 48 belts of quarters of the one quarter you need for that shape on that level the only thing you have to do is rotate the other 3 quarters from that shape to the same orientation that is required for that quarter.
@@DataEngineerPlays You have 3 floors and 4 belts for each shape. That means you need to produce 12 belts of random milestone shape. 12 shapes have (4x12) 48 quarters so you need to produce 12 quarters (having same rotation) of random milestone shape for 12 belts of final output.I am sure you will find a way to use of all top left 4 blue quarters in that example. In fact you need 12 top left blue quarters for 12 belts of final output. Look at your current setup, it is painting 12 same quarters on different platforms.
Dont try to produce final shape seperately, focus on quarters. Firts produce 48 quarters (assuming the layer has no gaps). That minimize the possibilities so less belt spagetti. First produce 4x12 quarters of random milestone shape, than stack 12 top right quarters with 12 bottom right quarters to get 12 right half of the shape than stack left quarters to get left half and finally stack two halfes to get 12 belt final shape. I would recomend to use stackers instead of swappers, it is less confusing and doing the same job. Also using stackers simplifies to setup wire logic.
@@gregmeeuwissen6393 Yes it's finally clicked in my head what all of you guys are talking about....
I need to stop thinking of space belts as 1 object that needs to processed together...
I can split 1 space belt into 4 units of 3, each unit can be painted THEN quartered to save my paint consumption by 4x
I got you guys now!
guess who will hit the 1k follower soon, im watching you since 225 followers (youtube algorithm) startet my abo at 906 or something, i believe you are blessed, i saw you pushed sometimes 2 episodes out, better save them und and do daily one episode at a fixed time, use extra episodes to get a day off, before you suffer in mental damage, you doin great and fast, even your episodes feel sometimes rushed and they are not even short (which is nice), but don't overload yourself your data engineer skills and advise are nice to watch
Haha much love for my early followers! Thanks for viewership. Yeah I think I'll probably do 1 episode a day. I'm still super new at all this so just trying to find the right cadence, a lot of experimenting on what time and how many videos to release etc.
PS: I'm blessed to have this amazing audience!
Future me: Did or did he not notice one of the inputlanes (per level) for green and cyan is set to outputs on his colorselecting platform? Awesome project btw!
Thanks I'll fix that!
Hexagonal MAM next?
haha let's get a functional regular MAM then we'll talk!
another point, why don't you copy your painting array on to your paint selector, you do have the space for it, and it would compact it even further.
It's true, could definitely do that, not a bad idea
i dunno if it was mentionend, but after we got the paint gates, wouldnt it be possible to create a dynamik paint mixer ? reducing the painting to the three starting colors, reducing even more paint lanes ?
It only reduces it slightly right?
Instead of needing 4 belts of r g b c y m w (28 total belts)
You need 4 belts of r g b
2 belts of b 2 belts of g to make c
2 belts of r 2 belts of g to make y
2 belts of b 2 belts of r to make m
2 belts of b 1 belt of r 1 belt of g to make
which is still 28 incoming belts right?
Just instead of the mixed colors your just importing r g b instead
@@DataEngineerPlays my operator level wasnt that highe jet so i had the assumption, if you shove color into a form analyzer, it could output the colors it need to mix,
like the unstacker just with color "the uncolorrizer ?" After testing i noticed that the form analizer can not display on a y that it needs like r/g
i would also be happy with a color analizer if we have it for shapes already >.< but that might be a future update.
so the belt reduction was under that assumption, you would have like one output pipe and rgb as input, with something like a color analyizer/uncolorizer.
p.s. also the MAM series encourages me to keep on gaming to unlock everything.
I am still watching the video so I might not have seen it, but you should need only one switch for each collor output in the platform right? You shouldn't need to switch each paint launcher since the output joins no?
a gate has a fluid throughput of 1800L/m
when I merge the 4 launchers into 1 pipe, it becomes a pipe of 7200L/m. So that's the reason to split again and use 4 gates
@@DataEngineerPlays I understand that, but when would you use any of the lauchers individually? You will always either use 4 lauchers or non in you color selection as far as I understand. Anyway, this an exercise... it works either way I believe.
@@yuriferes I don't believe you would ever use the launchers individually
I got a question 🤔
You take 4 inputs of colour and merge in to one pipe. You then split that one pipe in to 4 pipe selectors. And after that you merge the 4 exits from the pipe selectors back in to one pipe.
Is it that because of a limit to the capacity of the pipe selector ?
If so would it be enough to just split it in to 2 pipe selectors ?
A pipe carries 7200L (4x 1800L)
A fluid launcher is 1800L
A pipe gate is also 1800L
So you need 4 pipe gates per fully merged pipe
@@DataEngineerPlays ahh i figured it was something like that 🙂 thank you for clearing that up for me 🙂
why are you thinking of disposing 3 quarters at the end. using cutters and rotators we can just rotate them in the orientation we want and get all 4 quarters at the end. (Also in the currernt version you can check with belt reader that if a single quarter is neeeded, let's say circle, full circles move at 45/m only or 1 belt is moving and other 3 are stopped, other 3 quarters are stopped at the filter anyway) so even if you divide input shape belts into all 4 platforms we get a full throughput of 180/m at final point of quarters (Let's sayeven if full circle is needed now, now filters perform at 45/m since input belt is divided but at the end we get 4 times the quarters so that is still 180/m) Also painting full and quarters need same amount of paint so we can branch paint also to all four paltforms, paint at 45/m still get quarters at 180/m.
Also you made 2 input fluid launchers in the wrong direction in paint selector platform
In my head I might be overcomplicating the process of taking a quarter out of the shape and passing the remaining 3/4 on to the next phase. But it does make total sense. I feel like then it becomes very similar to what I did?
and thanks for the tip about the fluid launchers I'll fix that
@@DataEngineerPlays When you make quarters, you are not throwing away the other three quarters, you are cutting, rotating, cutting, rotating. Each cutting step takes one lane of input and makes two lanes of output. Each quarter of the input shape becomes a whole lane of output quarter-shape, so you go from one lane of input to four lanes of output. If you paint first, you paint all four quarters at once, with the cost of one lane worth of paint. If you paint second, you have to paint four lanes, each with its own paint consumption.
I think your trouble with understanding the reasoning behind the other order of operations (color first, cut later) comes from you have been focusing on having full space belts at all times.
But that's not needed. My suggestion (and some others I've seen) has been
1 space belt of mixed input shapes for the shape selector having 3 plain belts of each shape. 3 belt (plain belts, not space belts) output.
3 belt input to painter, 3 belt output. Painter is tiny, because it only deals with 3 (not 12) belts.
3 belts input to cutter/rotator. One full space belt output. No extras, they are all used for the same corner.
You could of course upgrade to having full space belts of each shape as input. Then you would just deliver 4 full space belts to the vortex. (I would suggest with a train launcher, then.)
Oh man the thought of having mixed input belts... that doesn't sit right with me!
@@DataEngineerPlays
Fair enough. That particular idea also only works with belts, not trains.
The main point was to only have 3 plain belts in the first part of the MaM because that gives an output of one full space belt at the end - and to keep it limited to those 3 belts for as long as possible.
End of the video felt called out cause thats literally what ive done i have 8 space belts of each shape and 4x7 belts of color. Currently losing my mind with the complex of doing 24 belt out.
Sorry didn't mean to call anyone out, it's just not the way I want to build my MAM, but I'm still in the very early stages, so I might run into a reason that makes me want to do it your way! who knows at this stage
@@DataEngineerPlays it's not in a bad way it was just funny to me as on my second screen that's literally what was on there
@@theinnocenttomato9836 Yeah I think I'm just misunderstanding you guys, because a few people have told me to color the shape first then cut it into quarters.. But I'm not seeing how that helps...
@DataEngineerPlays I'm with you on this the only way I can see painting full shape being better is if you have advanced logic in the beginning that analyzes the entire layer and somehow identifies partial shapes like halves and 3/4 etc.. but even then not sure if it would be more efficient
First hour club
28 colors? I see input of 4 shapes and 7 colors
when you stack 4 layers you`ll have 16 shapes and 28 colors.
you`re almost using the minimum amount of inputs.
the next best is to input 3 colors and mix dynamicaly inside the mam the correct color.
That's sounds like a good idea at first the dynamic mixing, but wouldn't that just require more belts of red/blue/green?
Your logic is wrong about the 8x4 belts. You don't need to find 8 full belts for each shape, because your end product is not consuming that amount. What I've said all along is that having all the inputs ready, with only a need of picking the correct quarter shape 4 times and then assembly, is much much shorter than what you are doing.
You are doing what springs to mind. Your logic is 1. I need a shape. 2. I need to cut it. 3. I need to color it. 4. Then assembly. And there's nothing wrong with it. Your logic will work, it just not efficient at all. You will be spending 20 minutes when there's a new request, just waiting for the new shape to hit the vortex. My logic is how do I supply every possible shape combination in the first step, because here, waiting is not a problem. It doesn't matter how long the belts are or how far they have to travel. As soon as everything there is ready, it will stay filled constantly. So my logic is: 1. Supply all possible quarters in all colors 2. Assemble it and deliver.
I know there will be pins and crystals further down the road. But they can have their own MAM, if inputting then becomes a problem. My logic all the time is NOT waiting and wasting time for the "dead shapes" to clear out. And also keeping the travel from the assembly to the vortex at an absolute minimum.
Hope this helps. And just to clarify once again. You are not doing it wrong, it's just not what I would do.
I agree with your logic about input belts not mattering, and to reduce the time spent in the MAM itself. But where you're losing me is how are you "1. Supply all possible quarters in all colors" but not needing "8x4 belts"?
@@DataEngineerPlays I mean you need the belts. But how full they are is not important, as you in theory only need as much as you are producing at the end.