@SurtenSoita @PlodTweets The main reason for the disparity is not technical, it is because the two editions were originally built by different teams in different locations. Now we are one team working together, so we are trying to reduce the gap.
Tech note: In this snapshot we ditched the thing about humidity and temperature varying depending on y level (for now at least), and made them 2d noises. 3d was causing too many weird issues, and wasn't strictly necessary in order to place cave biomes underground anyway.
Note that this only affects biome placement, not terrain shape. So biomes that are heavily tied to terrain shape (such as the peak biomes) still occupy the same amount of space as before. While terrain-independent biomes like forest are larger and more spread apart.
@Vextrove Barely. The difference between that and savanna is very small. But there are some differences so we kept it. We mainly removed/merged biomes that only were a different terrain shape and didn't have any other difference.
We needed to make a tech change to aquifers to ease future configurability. As a result we needed to retune them. So they are all in difference places now compared to last snapshot. I hope the overall balance (flooded / non-flooded) is still OK.
In general, caves that are near river/coastline should use sea level, inland should have less aquifers (but not be completely dry), and oceans should have more aquifers beneath them, so you can enter cave systems through the ocean sometimes.
Important: If you are testing & comparing this snapshot with last (which we really appreciate) please don't draw too many conclusions from one seed or location. That could be just be random. Test on multiple seeds or locations and look for recurring patterns.
About dripstone cave spawn changes: We kind of liked the thing about zombies not spawning in dripstone caves, but reverted that for now since we don't have time to test & balance that properly for this update. Might revisit in the future #nopromisesthough
@Phimic1 We break seeds sometimes during the snapshot cycle. That is important for the end result. Otherwise we can't experiment and get feedback, we would be stuck with terrain & biome distribution from the first snapshot.