latest snapshot for bedrock is great. 1.17.40.21 terrain is so interesting. it has a great cone going down with lots of trees. oak and beech. it's on a fairly large island. world is mostly water that i have found so far. did find few diamonds at -6. hard to find iron though so far. caves are excellent in this snapshot. plenty of enemies but interesting layout. some tight pathways and then open areas. nice balance there. there are several mountains towering above but haven't climbed them yet. don't know how to do the snow yet. anyway, great snapshot on this build.
It doesn't sound super balanced to me. Big caves with high risk and low reward.
Personally, I've really been liking the new terrain- yes it isn't always balanced, but it's certainly looking nicer, (In my totally objective opinion), and on some level, a greater sense of randomness; (That is, occasionally finding situations that don't have as much of a balanced risk/reward ratio), is part of what exploration is lacking right now- when everything is more or less predictable, there's no incentive to look around, and this seems to be one of the main focuses of this update. (Think about the giant ore veins, geodes, cave biomes, and other additions that necessitate and encourage more exploration.)
And in terms of aesthetics, for me the terrain generation is looking astonishingly good. Mojang has somehow managed to toe a very narrow line between vastly improving old terrain, without changing the vanilla feel of it. I still think there's some minor improvements that could be made, (such as better undergrowth for forests, or, say, different sorts of trees based on elevation could be cool); but by and large I don't have much to complain about.
I've really been finding it interesting how Mojang does game development... if only because Minecraft has become much more of a universe unto itself, or a simulation rather than a 'game' in the traditional sense- and they seem to be well aware of this, making design choices that, for the most part, offering new opportunities and cool stuff for players to try, rather than preset goals or overly complex systems. It's almost more worldbuilding than game design, in a very literal sense.
I like how spammers end up helping us revive topics...lol irony.
It doesn't sound super balanced to me. Big caves with high risk and low reward.
I don't get this either; have the developers even tried actually caving? They seem afraid that bigger caves means easier access to ores but in my experience this is not the case at all - in fact, I actually collect resources faster in my first world than I do in my modded worlds, despite vanilla 1.6.4 having only half the overall volume of caves and the largest caves you are likely to encounter in vanilla being less than half the size of what I classify as a "large" cave in TMCW, with the biggest caves being dozens of times larger (one such cave complex had a volume of over 1.2 million blocks and took 5500 torches to light up with around 15000 resources collected; this may seem like a lot but my hourly rate was lower than average for both TMCW and vanilla).
Most of the resources that I collect are also also coal and iron (about 90% of the total) - I average only 3-4 diamond ore per hour, which is not a good rate at all considering that it would take 6-8 hours of caving to get enough for a full set of diamond armor, while branch-mining can get that in half an hour - and for some reason Mojang decided to make many ores only generate if they are not exposed, which will only widen the gap (never mind the differences in safety).
Also, I disagree with making mobs only spawn in a light level of 0; in fact, they need to make it so that a light level of 0 is totally pitch black and editing gamma in options.txt doesn't work, as I did in my own modded versions, then torches (or Night Vision) are actually required to see, not just stop mobs from spawning (and in my experience lighting up caves actually increases the amount of mobs you encounter, or at once, since there is less spawnable area, concentrating them in dark areas); I mainly place torches not to stop mobs from spawning but so I can see with clear visibility (example of a lit-up large cave; note that I also light up the walls and even the ceiling, which requires a lot of pillaring up) and changing spawning levels would have no effect. I also increased the density of mobs by reducing their (de)spawn radius from 128 to 96 blocks as well as the no-spawn radius around the player from 24 to 16 - fighting mobs is a large part of why I find caving to be fun.
Of course, my playstyle is undoubtedly a very small minority; I doubt that the average player does as much caving as much as I do in a single play session over the entire time they play on a world, much less exploring literally entire continents worth of caves, but that just makes the issues mentioned above more significant (granted, you could find diamonds faster if you only explore areas where they can be found; I simply explore everything underground with no regard to what can be found, so long as it is a cave I'll explore it). Naturally, I don't have to worry about any of this myself since I play on my own modded versions and don't even see new updates as updates to the game that I play.
Also, I disagree with making mobs only spawn in a light level of 0; in fact, they need to make it so that a light level of 0 is totally pitch black and editing gamma in options.txt doesn't work,
Is this something they're actually adding? There is no way they are going to make mobs spawn in only a light level of 0, personally, I prefer a light level of 5-6, 7 is too bright IMO.
About the pitch black thing, they should make it ALMOST pitch black, so you can see around a teeny tiny bit but not really playable. And I'm kinda neutral on the editing gamma bit.
About my opinion on 1.18, I think they should up ore spawning a bit, but not too much. That way, it's not like you can get an average of 7 diamonds in 5 minutes (these are arbitrary numbers by the way), but it doesn't take 34 minutes to come across a single diamond (34 is also an arbitrary number). You don't know how many times I've spent over half an hour exploring caves and stuff only to not encounter a SINGLE diamond. Other than that, well, I have no clue, I haven't really played any of the 1.18 snapshots or even much 1.17 for that matter lol.
the mobs are active in this build. haven't got armor yet. cows are not enough for hides yet and not enough iron for armor yet. maybe not looking in the right places for iron. want to post some screenshots but don't know how to do that on the xbox. i will get the seed and put it on the pc and see if the same world exists. probably time for a new build today for the bedrock.
Is this something they're actually adding? There is no way they are going to make mobs spawn in only a light level of 0, personally, I prefer a light level of 5-6, 7 is too bright IMO.
About the pitch black thing, they should make it ALMOST pitch black, so you can see around a teeny tiny bit but not really playable. And I'm kinda neutral on the editing gamma bit.
Yes, if only for block light and not sky light (which would otherwise mean no mobs at night since it only gets down to 4):
That said, I've implemented a similar change myself, except the threshold was lowered to 5, as this complements what my "cave maps" use to display lit-up caves (at least a light level of 6 within 8 blocks by taxicab from a torch; this actually came first and I got inspiration from 1.18 to change mob spawning as this makes cave maps quite useful to determine if an area has been lit up properly); either way, even a threshold of 0 is unlikely to have much effect when you are placing torches for visibility (and I mean more than just dimly lit, so you can easily recognize blocks at a glance). The main impact would be around lava-filled caves, which are already less likely to spawn mobs anyway, and in structures with torches.
The only concern that I have is how this affects dungeons/mob spawners, which are already far too easy due to their ridiculously slow spawn rates (my solution was to keep the threshold for spawners at 7; either way, in vanilla I can usually just run into a dungeon and break the spawner before it can spawn more than once, if even that, without bothering to light it up).
Also, why "almost" pitch black? You do realize that monitors have a major impact on what people see? On my monitor the default gamma setting is simply awful according to Window's display calibration tool I had to adjust it to nearly as low as it can go to get the correct results (the middle, while the default setting looks like the right), and this was also reflected in-game (even Moody was quite easy to see in, while Bright, which I always use, was just terrible; for comparison, my old computer looked correct and in-game it was nearly totally dark; I added the light level of 0 feature to my own mods after I got a new computer as I saw for the first time how bad an uncalibrated display can be and why so many people don't even use torches when caving (worse, when they record a video and assume that everybody can see just as well; if they do this while I'm watching I just stop watching).
The in-game brightness (gamma) setting is also broken because it is supposed to change the gamma curve between 0% and 100% brightness but since 0% brightness is not actually 0% it is also affected (what it does is change the curve from logarithmic on Moody to linear on Bright, with no effect on the maximum brightness, so it is not really a "brightness" setting. In fact, block light levels 14-15 are visually the same, and even the exact same brightness, on both Moody and Bright, with sky light level 14 being much darker on Moody and about the same as 15 on Bright).
Also, countless bugs and crashes have resulted from Mojang's failure to add proper parsing of values read in from options.txt (for example, 1.7.x will crash if the render distance is set higher than 16, which can also occur if you downgrade from 1.8+ so you don't even need to mess with the file to cause problems). Many people also reinstall the entire game to fix issues when they only had to delete options.txt (the launcher verifies file integrity on every launch with the exception of game data files like options.txt, resource packs, and saves, so only those files can cause issues).
I love the terrain and generation changes I think they are mostly a good addition to what's already a great game. Saying that i would love for 1.19 to be focused more on creating new mobs or even mini bosses.
The minecraft world's just don't seem alive currently. With all the new cave, mountain and generation changes I think we need new hostile mobs in caves and in the overworld. Seeing the same 3 mobs over and over gets stale. Zombies, creepers and Skeletons. If we added maybe different variations maybe Harder versions of those 3 once you are below -y1 would be cool or just adding 3 or 4 new hostile mobs would be even better.
I find the existing mobs to be quite lively, but having cave biome specific mobs would be cool as well as lower level only ones. Perhaps cave spiders for starters and silverfish maybe. Wardens are kind of the only mob doing this right now, if we're talking just hostiles.
My understanding is that for the entirety of 1.18 so far, the rng wasn't being applied correctly. I'm not sure I'm willing to spend hours looking for a good seed, because this doesn't seem like a little change.
My understanding is that for the entirety of 1.18 so far, the rng wasn't being applied correctly. I'm not sure I'm willing to spend hours looking for a good seed, because this doesn't seem like a little change.
Also, if it aint broke, why fix it?
Most world generation used Java's Random, which is ludicrously bad by modern standards - it only uses 48 bits, meaning that for every seed there are 65535 other seeds which produce the exact same terrain and other features; only the biome map, which uses a custom RNG, uses the full 64 bit seed - and this has been the case ever since the game was created (very old versions didn't even use 64 bits for biomes, hence the thread):
In particular, you can find the exact same village down to its loot and even the surrounding tall grass and animals in the seeds "-123775873255737467" and "-556121437483305083"; only the overall biome layout differs (and even then biome borders are similar):
Here is another example, comparing the spawn area of the seed "-123775873255737467" to the same area in "-267891061331593339"; note that the underground is exactly the same, as is non-biome dependent terrain elevation (and even then the only difference is the relative height due to the biome's height values) and even the placement of rivers and biome borders:
Not only that, it is poorly used; the way the "chunk seed" is set causes many seeds to have odd repetition or even fail entirely (aka the famous infinite repeating cave seed). This has less to do with the RNG than the algorithm used to calculate a unique seed for each chunk:
In fact, a third of all seeds have the exact same caves, ravines, and mineshafts at 1/3 of sign-reversed coordinate pairs (villages and other structures can also be affected, though since they use a different placement method it is less visible) - I've even recognized that I explored the same caves in my first world, which has 1/6 of all chunks affected (another 1/3 of seeds, with the remaining third having 1/12, 1/24, etc of chunks matching; only 0.23% of 10,000 seeds checked had no matches within 256 chunks of the origin):
I fixed these in my own mod with my own 64 bit RNG and a better way to calculate chunk seeds (I can't say for sure if it is immune to the bugs mentioned above due to the amount of testing involved); I even changed the random seed on world creation to use it instead of Random (this means that there are effectively only 2^48 possible seeds unless you manually enter a numerical seed; text seeds barely extend the seed space despite there being little overlap as there are only 2^32), as well as allowing "0" to be directly entered (why is this even a thing? I simply check if there is anything at all in the seed textbox, while vanilla checks if it is zero as well, and the seed value itself has no meaning to the game, it is just another number).
Also, Random is extremely slow due to wrapping its internal state in an "AtomicLong" object for thread safety reasons, which isn't needed - by simply extending the class and replacing all references to AtomicLong with a long field I was able to halve the time taken to generate caves in vanilla and TMCWv4, with substantial improvements elsewhere, with my 64 bit RNG being even faster - generating a random integer in only 1.2 nanoseconds (800 million numbers per second), while Java's Random is 8-14 times slower (times are for nextInt(n) only):
Random.nextInt(n) took 9.9637 nanoseconds
Random48.nextInt(n) took 2.3439 nanoseconds; was 4.2509067 times faster than Random
Random64.nextInt(n) took 1.2762 nanoseconds; was 7.8073187 times faster than Random
Of course, it is still a linear congruential generator, which has some quality issues (mainly due to a too-small state; 64 bits is 65536 times larger than 48 bit), but I mitigate them by using it properly (e.g. using the highest bits possible, which have the most randomness; Java's Random discards as few as 16 bits and nextInt(n) only has maximal period with powers of 2). Even vanilla's biome generator has such issues which I fixed by using the highest bits - not even "shadow seeds" are similar:
"-123775873255737467" and "-556121437483305083" in vanilla 1.6.4 (these have the same lower 48 bits and should produce very different biome maps since the biome generator uses all 64 bits):
"-123775873255737467" and "-556121437483305083" in TMCWv5, which are completely different, including biome borders and rivers:
"-2143500864" and "-7379792618385405355" in vanilla 1.6.4 (these are "shadow seeds", which give the same biome map but different terrain and reduces the number of unique biome layouts to 2^63 since every seed is one of a pair):
"-2143500864" and "-7379792618385405355" in TMCWv5, which are again completely different:
This is how simple my fix for the period is:
Vanilla 1.6.4; note that only the lower 24 bits are discarded and modulo (%) is used, so e.g. nextInt(2) will only return bit 25, whose period is only 2^25:
TMCWv5; note that the highest bits are always used as all bits will be discarded (total shift by 64 places) except those shifted back by multiplying with the input parameter; e.g. nextInt(2) will return bit 64, with a period of 2^64 (one limitation is that n is limited to 1024 instead of Integer.MAX_VALUE but this is the maximum value used anywhere), and removes a unnecessary check for negative values (modulo is also much slower due to being a division):
I like how spammers end up helping us revive topics...lol irony.
It doesn't sound super balanced to me. Big caves with high risk and low reward.
Personally, I've really been liking the new terrain- yes it isn't always balanced, but it's certainly looking nicer, (In my totally objective opinion), and on some level, a greater sense of randomness; (That is, occasionally finding situations that don't have as much of a balanced risk/reward ratio), is part of what exploration is lacking right now- when everything is more or less predictable, there's no incentive to look around, and this seems to be one of the main focuses of this update. (Think about the giant ore veins, geodes, cave biomes, and other additions that necessitate and encourage more exploration.)
And in terms of aesthetics, for me the terrain generation is looking astonishingly good. Mojang has somehow managed to toe a very narrow line between vastly improving old terrain, without changing the vanilla feel of it. I still think there's some minor improvements that could be made, (such as better undergrowth for forests, or, say, different sorts of trees based on elevation could be cool); but by and large I don't have much to complain about.
I've really been finding it interesting how Mojang does game development... if only because Minecraft has become much more of a universe unto itself, or a simulation rather than a 'game' in the traditional sense- and they seem to be well aware of this, making design choices that, for the most part, offering new opportunities and cool stuff for players to try, rather than preset goals or overly complex systems. It's almost more worldbuilding than game design, in a very literal sense.
Cooking with Mindthemoods ~ Biomes ~ Archeology
---
~ My Portfolio ~ Skindex ~ Test ~ Discs ~
I don't get this either; have the developers even tried actually caving? They seem afraid that bigger caves means easier access to ores but in my experience this is not the case at all - in fact, I actually collect resources faster in my first world than I do in my modded worlds, despite vanilla 1.6.4 having only half the overall volume of caves and the largest caves you are likely to encounter in vanilla being less than half the size of what I classify as a "large" cave in TMCW, with the biggest caves being dozens of times larger (one such cave complex had a volume of over 1.2 million blocks and took 5500 torches to light up with around 15000 resources collected; this may seem like a lot but my hourly rate was lower than average for both TMCW and vanilla).
Most of the resources that I collect are also also coal and iron (about 90% of the total) - I average only 3-4 diamond ore per hour, which is not a good rate at all considering that it would take 6-8 hours of caving to get enough for a full set of diamond armor, while branch-mining can get that in half an hour - and for some reason Mojang decided to make many ores only generate if they are not exposed, which will only widen the gap (never mind the differences in safety).
Also, I disagree with making mobs only spawn in a light level of 0; in fact, they need to make it so that a light level of 0 is totally pitch black and editing gamma in options.txt doesn't work, as I did in my own modded versions, then torches (or Night Vision) are actually required to see, not just stop mobs from spawning (and in my experience lighting up caves actually increases the amount of mobs you encounter, or at once, since there is less spawnable area, concentrating them in dark areas); I mainly place torches not to stop mobs from spawning but so I can see with clear visibility (example of a lit-up large cave; note that I also light up the walls and even the ceiling, which requires a lot of pillaring up) and changing spawning levels would have no effect. I also increased the density of mobs by reducing their (de)spawn radius from 128 to 96 blocks as well as the no-spawn radius around the player from 24 to 16 - fighting mobs is a large part of why I find caving to be fun.
Of course, my playstyle is undoubtedly a very small minority; I doubt that the average player does as much caving as much as I do in a single play session over the entire time they play on a world, much less exploring literally entire continents worth of caves, but that just makes the issues mentioned above more significant (granted, you could find diamonds faster if you only explore areas where they can be found; I simply explore everything underground with no regard to what can be found, so long as it is a cave I'll explore it). Naturally, I don't have to worry about any of this myself since I play on my own modded versions and don't even see new updates as updates to the game that I play.
TheMasterCaver's First World - possibly the most caved-out world in Minecraft history - includes world download.
TheMasterCaver's World - my own version of Minecraft largely based on my views of how the game should have evolved since 1.6.4.
Why do I still play in 1.6.4?
Is this something they're actually adding? There is no way they are going to make mobs spawn in only a light level of 0, personally, I prefer a light level of 5-6, 7 is too bright IMO.
About the pitch black thing, they should make it ALMOST pitch black, so you can see around a teeny tiny bit but not really playable. And I'm kinda neutral on the editing gamma bit.
---------------------------------------------------------------------------
About my opinion on 1.18, I think they should up ore spawning a bit, but not too much. That way, it's not like you can get an average of 7 diamonds in 5 minutes (these are arbitrary numbers by the way), but it doesn't take 34 minutes to come across a single diamond (34 is also an arbitrary number). You don't know how many times I've spent over half an hour exploring caves and stuff only to not encounter a SINGLE diamond. Other than that, well, I have no clue, I haven't really played any of the 1.18 snapshots or even much 1.17 for that matter lol.
I think it's pretty dark as is tbh
the mobs are active in this build. haven't got armor yet. cows are not enough for hides yet and not enough iron for armor yet. maybe not looking in the right places for iron. want to post some screenshots but don't know how to do that on the xbox. i will get the seed and put it on the pc and see if the same world exists. probably time for a new build today for the bedrock.
Yes, if only for block light and not sky light (which would otherwise mean no mobs at night since it only gets down to 4):
That said, I've implemented a similar change myself, except the threshold was lowered to 5, as this complements what my "cave maps" use to display lit-up caves (at least a light level of 6 within 8 blocks by taxicab from a torch; this actually came first and I got inspiration from 1.18 to change mob spawning as this makes cave maps quite useful to determine if an area has been lit up properly); either way, even a threshold of 0 is unlikely to have much effect when you are placing torches for visibility (and I mean more than just dimly lit, so you can easily recognize blocks at a glance). The main impact would be around lava-filled caves, which are already less likely to spawn mobs anyway, and in structures with torches.
The only concern that I have is how this affects dungeons/mob spawners, which are already far too easy due to their ridiculously slow spawn rates (my solution was to keep the threshold for spawners at 7; either way, in vanilla I can usually just run into a dungeon and break the spawner before it can spawn more than once, if even that, without bothering to light it up).
Also, why "almost" pitch black? You do realize that monitors have a major impact on what people see? On my monitor the default gamma setting is simply awful according to Window's display calibration tool I had to adjust it to nearly as low as it can go to get the correct results (the middle, while the default setting looks like the right), and this was also reflected in-game (even Moody was quite easy to see in, while Bright, which I always use, was just terrible; for comparison, my old computer looked correct and in-game it was nearly totally dark; I added the light level of 0 feature to my own mods after I got a new computer as I saw for the first time how bad an uncalibrated display can be and why so many people don't even use torches when caving (worse, when they record a video and assume that everybody can see just as well; if they do this while I'm watching I just stop watching).
The in-game brightness (gamma) setting is also broken because it is supposed to change the gamma curve between 0% and 100% brightness but since 0% brightness is not actually 0% it is also affected (what it does is change the curve from logarithmic on Moody to linear on Bright, with no effect on the maximum brightness, so it is not really a "brightness" setting. In fact, block light levels 14-15 are visually the same, and even the exact same brightness, on both Moody and Bright, with sky light level 14 being much darker on Moody and about the same as 15 on Bright).
Also, countless bugs and crashes have resulted from Mojang's failure to add proper parsing of values read in from options.txt (for example, 1.7.x will crash if the render distance is set higher than 16, which can also occur if you downgrade from 1.8+ so you don't even need to mess with the file to cause problems). Many people also reinstall the entire game to fix issues when they only had to delete options.txt (the launcher verifies file integrity on every launch with the exception of game data files like options.txt, resource packs, and saves, so only those files can cause issues).
TheMasterCaver's First World - possibly the most caved-out world in Minecraft history - includes world download.
TheMasterCaver's World - my own version of Minecraft largely based on my views of how the game should have evolved since 1.6.4.
Why do I still play in 1.6.4?
It's weird how the last couple snapshots are mostly just bugfixes and tweaks. We still don't have an entire biome.
-
View User Profile
-
View Posts
-
Send Message
Curse PremiumI bet they won't add the deep dark until the near end.
Praise be to Spode.
Thing is, we ought to be nearly *at* the end. Other than the Deep Dark and the Warden, the only thing missing is goat horns unless I'm mistaken.
Unless they're keeping extra stuff under their hat, I suppose.
I highly doubt both of those.
It seems they're very unsure how to make the new terrain.
simple, just type random numbers in the world gen code and boom new terrain
this is a joke btw
I love the terrain and generation changes I think they are mostly a good addition to what's already a great game. Saying that i would love for 1.19 to be focused more on creating new mobs or even mini bosses.
The minecraft world's just don't seem alive currently. With all the new cave, mountain and generation changes I think we need new hostile mobs in caves and in the overworld. Seeing the same 3 mobs over and over gets stale. Zombies, creepers and Skeletons. If we added maybe different variations maybe Harder versions of those 3 once you are below -y1 would be cool or just adding 3 or 4 new hostile mobs would be even better.
I find the existing mobs to be quite lively, but having cave biome specific mobs would be cool as well as lower level only ones. Perhaps cave spiders for starters and silverfish maybe. Wardens are kind of the only mob doing this right now, if we're talking just hostiles.
I guess small biomes are a loss, yeah.
its quite nice, To have This update
The most recent snapshot feels like a step back in terms of terrain generation.
Only thing changed compared to the previous ones is new RNG for seeds.
It also means this bug is fixed: https://bugs.mojang.com/browse/MC-236650
My understanding is that for the entirety of 1.18 so far, the rng wasn't being applied correctly. I'm not sure I'm willing to spend hours looking for a good seed, because this doesn't seem like a little change.
Also, if it aint broke, why fix it?
Most world generation used Java's Random, which is ludicrously bad by modern standards - it only uses 48 bits, meaning that for every seed there are 65535 other seeds which produce the exact same terrain and other features; only the biome map, which uses a custom RNG, uses the full 64 bit seed - and this has been the case ever since the game was created (very old versions didn't even use 64 bits for biomes, hence the thread):
https://www.minecraftforum.net/forums/minecraft-java-edition/seeds/2229720-can-two-different-seeds-produce-identical-worlds
In particular, you can find the exact same village down to its loot and even the surrounding tall grass and animals in the seeds "-123775873255737467" and "-556121437483305083"; only the overall biome layout differs (and even then biome borders are similar):
Here is another example, comparing the spawn area of the seed "-123775873255737467" to the same area in "-267891061331593339"; note that the underground is exactly the same, as is non-biome dependent terrain elevation (and even then the only difference is the relative height due to the biome's height values) and even the placement of rivers and biome borders:
Not only that, it is poorly used; the way the "chunk seed" is set causes many seeds to have odd repetition or even fail entirely (aka the famous infinite repeating cave seed). This has less to do with the RNG than the algorithm used to calculate a unique seed for each chunk:
https://www.minecraftforum.net/forums/minecraft-java-edition/seeds/2929490-the-chunk-seed-algorithm-used-by-caves-and-some
In fact, a third of all seeds have the exact same caves, ravines, and mineshafts at 1/3 of sign-reversed coordinate pairs (villages and other structures can also be affected, though since they use a different placement method it is less visible) - I've even recognized that I explored the same caves in my first world, which has 1/6 of all chunks affected (another 1/3 of seeds, with the remaining third having 1/12, 1/24, etc of chunks matching; only 0.23% of 10,000 seeds checked had no matches within 256 chunks of the origin):
https://www.minecraftforum.net/forums/minecraft-java-edition/discussion/197499-random-world-generation-not-so-random-smallest
https://www.minecraftforum.net/forums/minecraft-java-edition/seeds/3004110-repeating-decorations-trees-ores-etc-seeds
I fixed these in my own mod with my own 64 bit RNG and a better way to calculate chunk seeds (I can't say for sure if it is immune to the bugs mentioned above due to the amount of testing involved); I even changed the random seed on world creation to use it instead of Random (this means that there are effectively only 2^48 possible seeds unless you manually enter a numerical seed; text seeds barely extend the seed space despite there being little overlap as there are only 2^32), as well as allowing "0" to be directly entered (why is this even a thing? I simply check if there is anything at all in the seed textbox, while vanilla checks if it is zero as well, and the seed value itself has no meaning to the game, it is just another number).
Also, Random is extremely slow due to wrapping its internal state in an "AtomicLong" object for thread safety reasons, which isn't needed - by simply extending the class and replacing all references to AtomicLong with a long field I was able to halve the time taken to generate caves in vanilla and TMCWv4, with substantial improvements elsewhere, with my 64 bit RNG being even faster - generating a random integer in only 1.2 nanoseconds (800 million numbers per second), while Java's Random is 8-14 times slower (times are for nextInt(n) only):
Of course, it is still a linear congruential generator, which has some quality issues (mainly due to a too-small state; 64 bits is 65536 times larger than 48 bit), but I mitigate them by using it properly (e.g. using the highest bits possible, which have the most randomness; Java's Random discards as few as 16 bits and nextInt(n) only has maximal period with powers of 2). Even vanilla's biome generator has such issues which I fixed by using the highest bits - not even "shadow seeds" are similar:
"-123775873255737467" and "-556121437483305083" in TMCWv5, which are completely different, including biome borders and rivers:
"-2143500864" and "-7379792618385405355" in vanilla 1.6.4 (these are "shadow seeds", which give the same biome map but different terrain and reduces the number of unique biome layouts to 2^63 since every seed is one of a pair):
"-2143500864" and "-7379792618385405355" in TMCWv5, which are again completely different:
This is how simple my fix for the period is:
Vanilla 1.6.4; note that only the lower 24 bits are discarded and modulo (%) is used, so e.g. nextInt(2) will only return bit 25, whose period is only 2^25:
protected int nextInt(int par1) { int var2 = (int)((this.chunkSeed >> 24) % (long)par1); if (var2 < 0) var2 += par1; this.chunkSeed *= this.chunkSeed * 6364136223846793005L + 1442695040888963407L; this.chunkSeed += this.worldGenSeed; return var2; }TMCWv5; note that the highest bits are always used as all bits will be discarded (total shift by 64 places) except those shifted back by multiplying with the input parameter; e.g. nextInt(2) will return bit 64, with a period of 2^64 (one limitation is that n is limited to 1024 instead of Integer.MAX_VALUE but this is the maximum value used anywhere), and removes a unnecessary check for negative values (modulo is also much slower due to being a division):
protected final int nextInt(int n) { this.chunkSeed *= this.chunkSeed * 6364136223846793005L + 1442695040888963407L; this.chunkSeed += this.worldGenSeed; return (n * (int)(this.chunkSeed >>> 42)) >>> 22; }(this alone doesn't fix "shadow seeds", which were fixed by adding the initial world seed to the final "worldGenSeed" value)
TheMasterCaver's First World - possibly the most caved-out world in Minecraft history - includes world download.
TheMasterCaver's World - my own version of Minecraft largely based on my views of how the game should have evolved since 1.6.4.
Why do I still play in 1.6.4?