the Cursed Belt Tapestry Blueprint Book!

[@MichaelHendriksLIVE]

Alright, after 3 days of murdering my eyeballs by staring at WAY too many overlapping underground belt arrows, I proudly present Michael Hendriks' Cursed Belt Tapestry Book v1.0! www.dropbox.com/scl/fi/lij6w2j4wz8ywshxskfcp/Michael-Hendriks-Cursed-Belt-Tapestry-Book.txt?rlkey=46kqkivkqdlx7jn2a1yrcpnwo&st=v2v2e1qh&dl=1 Includes all the builds from the video, PLUS the FLÆSHLIGHT Spaceship, PLUS an improved, add-10k-asteroids-per-click to the infinite storage cryo plants, PLUS a bunch of blueprint books with orthogonal belt tapestry designs by other people (in the comments and on discord, credits in the book), which I made tileable variants from. The new maximum crossable water gap record: with cursed inserter pulling: 36 tiles (was:32) with full throughput: 32 tiles (full chunk!) (was:30) with normal belt entrances: 28 (was:28 Ha! at least ONE of my designs survived!) Blueprint books not seen in video: 4.4 belts per tileable center (10x10) with lights/lasers for style points 4.5 belts per tileable center (20x20) 4.666 belts per tileable center (48x10) 4.739 belts per tileable center (46x10) 4.756 belts per tileable center (46x10) 4 green belts of throughput while maintaining 3.5-4 belts per tile of storage (including overhead!) There will inevitably be updates made to the book, but I will try to not get sniped by new designs and improvements for at least a couple days! The link will stay the same for new updates. Enjoy! Michael

[@Pystro]

18:52 Correction: The improvement between the 4.666 (in 48x10) and 4.739 (in 46x10) does not come from the mixing of belts. The improvement comes from using a more optimal horizontal belt weave. If you look towards the edges of the 4.666 quilt, you'll see horizontal belts of red,blue,red color next to each other. The reds are needed for this width, but it turns out that you can simply eliminate one of the reds and shorten the belt weave. (Since you're eliminating one occupied position while keeping the number of empties the same, the average density of gaps that can be filled with verticals goes up). The "mixing" of red and blue in the 4.735 design is simply to show that is doesn't matter which of the two colors you eliminate. Also note that it doesn't actually mix the belt colors, it just mixes up the order of their surfacings. The mixing and the improvement do happen in the same place, so I can see how Michael would assume that there's a causal relation.

[@ThatOliveMrT]

The belt weave is a great way to add some color. I should find a way to add style with lamps too

[@samc4741]

I'm thinking of adding long handed inserters to the gaps. I believe that will store an extra one (1) item.

[@dannyboy1350]

@@samc4741 If you move one of the undergrounds forward by two tiles and connect with normal belts then you can do stack inserters instead, which can store up to 16 items each but you'd only have half the space.

[@jmatya]

Hendriks Storage now generally available ❤

[@Thaumogenesis]

You wouldn't download promethium chunks.

[@epicspacetaco7083]

Such a monstrosity, I'm all for it.

[@darkmtbg]

so at 13:07. couldnt you build a timer with a Constant combinator and three decider combiator that would set the recipe on the cryoplants. to switch between batteries and packs by alternating between batteries and packs on the timer reaching two different numbers. Constant combinator outputting 1 Signal T(imer) and descider combinator takes both Constant combinator, its own output. causing signal T to count up. untill it hits an upper bound resetting. the Signal T is then input into 2 decider combinators where when the signal T < 32k one of the output sicence pack and Signal T >= 32k would output battiries. and the end of the cycle would be 33k. then you would have the same as the click and drag method, just automated. of cause there would need some fine tuning

[@Charlie-kc6py]

You can't change recipes with logic, if there are items in output slots.

[@MichaelHendriksLIVE]

yeah Charlie is right, I covered this in the main video on the main channel

[@samc4741]

how does a speed moduled science assembler compare to belts in terms of prometheum density, if you don't do the infinite storage trick? I assume it's bad, but just wondering how bad.

[@MichaelHendriksLIVE]

Worse than belt weaving storage. without beacons, it takes in 350 chunks in 25 tiles = 14 asteroids per tile =

[@jonasgajdosikas1125]

well with a storage of 675 per it has to contend with 25 tiles of belt weave being generous we'll take 4 belts per tile and get 100 belts worth. this would then store 800 prometheum which is more than the equivalent amount in the cryo plant. if we factor in the loading/unloading and beacons it'll have to contend with 42.5 tiles worth of belts thus skewing the result even more in favor of belts though if you had recursive blueprints+ (cause of update 2.0) installed you could automate the changing of recipes

[@jonasgajdosikas1125]

after further testing you can eat the whole throughput of the 4 belts using the cryogenic plants - you need just enough to handle the inserters, with ~20 inserters handling the throughput 4 cryogenic plants should be able to store as much as you ever would need. the only "problem" is that it's a pain to set the blueprint, but you should be able to use an inserter filtered to blueprints to "flash" the blueprint deployer

[@TedG-fq8kc]

I came up with a single line 32 tile long asymmetrical pattern allowing some stretching of the end entrance points. I suppose mixing colors might allow a longer string. For stating the pattern I will just say the hump pairs: GYRBYRGYBRYR. Figuring the end entrance and exit points is easy. Calculated item storage is 1008.

[@user-he3gn5bb2y]

58 by 58 designs would be a great task for some machine learning algorithm👍

[@Thaumogenesis]

The Intended Way™

[@dannyboy1350]

Surprised the high end designs used scripting, I managed to get a 4.666 design which tiles every 10*36 without scripting, that's a smaller tile area than the 4.666 in the video.

[@Pystro]

I'm one of the people who were involved in scripting. The "problem" is that with many horizontal belt braids, the trivial solution of "just take a horizontal belt braid, and stack 5 of them on top of each other, each with a horizontal offset of 1 more than the previous" has both columns with too many horizontal "surfacings" and with "too little" (I.e. columns with not enough space for 2 colors of vertical belts and columns with too much space). It's pretty easy to manually find one where their numbers are decently low, but there's no way of knowing what the _best_ result is that you can get for a given horizontal belt braid. Especially if you suspect that there should be some way to line them up without any imperfections, then proving that there isn't requires testing _all_ possible ways to shift the horizontal belt braids. And in the process you'll find the maximum length horizontal belt weave with only one vertical belt missing for the longest horizontal belt weave (which is that the 4.753 is). Michael somewhat sold our effort a bit short by only showing the single best result that you can find with scripts. In reality, there's "four-point-seven-something"s for a few widths. Also, given that the current consensus among us scripters is that 4.8 or thereabouts seems to be the best you'd be able to reach or approach (with the 46x46 class of solutions which our scripts can't handle), should put into perspective how much closer a jump from 4.666 to 4.753 out of 4.800 actually takes you.

[@dannyboy1350]

@@Pystro I understand all this and see why the top designs use scripting but if I can do better than the computer for a specific size after only 15 minutes then there seems to be a problem with the computer(an old model sure, so you might have fixed it in the new versions).

[@Pystro]

@@dannyboy1350 Well, let me put it like this: The computer doesn't have any imagination and can't have ideas, so all the ideas have to come from us. And for some reason none of us thought about doing size 36*10, because 36 just isn't a very good horizontal belt baid. [edit: I should probably mention that for each new width, we had to manually find _ALL_ belt braids of that length (of both RGB and GB type); which is why we only looked at the widths with the belt braids that could have allowed us to find a 4.8 density solution.] If I remember the video correctly, then the 4.666 shown was a design with reds in all horizontals. And we found all of those manually. Only the 4.75yada ones are from the scripts. I.e. your 4.666 in 36x10 doesn't match the density in the script solutions, it matched our manually found solutions. Just that yours has a smaller tiling unit (which is not a direction that we were concerned pushing the results into).

[@dannyboy1350]

@@Pystro "the 4.666 shown was a design with reds in all horizontals. And we found all of those manually." ah, so both of the 4.666 designs are human found. I seem to remember Michael saying it was the first design found via script.

[@kingcrimson1631]

Mike you should do a PhD on this haha

[@mnmmar2493]

nice...