How aggressive should I trim the world map before 1.16?

Started by Lazybones, February 09, 2020, 01:27:55 PM

Previous topic - Next topic

How tight should I trim the maps?

Just based on Inhabited time: Less than 3 Min (the most trimmed by time)
2 (66.7%)
Just based on Inhabited time: Less than 2 Min (medium trim)
1 (33.3%)
Just based on Inhabited time: Less than 1 Min (mostly seen but not walked chunks)
0 (0%)
Just based on placed items (haven't got this working yet, but the tightest option not tested)
0 (0%)
Just based on Inhabited time less than or equal 0 (only unwalked and seen chunks)
0 (0%)

Total Members Voted: 3

Voting closed: February 29, 2020, 01:27:55 PM

Lazybones

Hi all I have mcselector working on the current map, it doesn't all fit on one screen but I could visualize some of the differences.. Currently I have not been able to get entity filters to work but time inhabited seems to work perfectly.

https://github.com/Querz/mcaselector#chunk-filter

Unfortunately the nether is very compact so even with fairly aggressive time inhabited value it mostly trims the edges not walked.

The good news is that even when being very aggressive with time no ones base appears to be touched that I have seen.

Thorin

I don't think I can vote until I see what each would look like.  Can you visualize?
Prayin' for a 20!

gcc thorin.c -pedantic -o Thorin
compile successful

Lazybones

I can but I can't get it all on screen at once and each option takes a while to render.

Mostly interested in your base and spawn?

This is the area around your base..

If anyone wants me to show an area I need an X and Z coordinate to locate

You cannot view this attachment.

You cannot view this attachment.

You cannot view this attachment.

Lazybones

I have tested the nether and almost nothing gets trimmed till the 10 min mark so I might have to figure out how to use entity filters to get more aggressive as the big change for 1.16 is new nether stuff.

Lazybones

OK this is what an ULTRA tight trim of the nether would look like.

This ONLY keeps chunks with a torch, wall_torch, stone, or packed_ice. It looks like it would maintain all of our transport paths.. There do seem to be a few dug areas that would be reset however.

I think for the nether, 1.16 ultra tight based on blocks will work best, time stamps didn't really remove anything.. This could result in a very odd hard line between old and new bioms but at least we would not need to travel far.

Any more special blocks I should ID for keeping chunks?

You cannot view this attachment.

Mr. Analog

By Grabthar's Hammer

Lazybones

Quote from: Mr. Analog on February 11, 2020, 05:09:34 AMAre Portal blocks preserved?

I should be able to specifically protect chunks with obsidian/portal blocks


Tom

Quote from: Lazybones on February 11, 2020, 09:15:50 AM
Quote from: Mr. Analog on February 11, 2020, 05:09:34 AMAre Portal blocks preserved?

I should be able to specifically protect chunks with obsidian/portal blocks


Just obisdian may not be ideal. perhaps obidian in chunks that have been in recently? Or if you can mark portal blocks themselves?
<Zapata Prime> I smell Stanley... And he smells good!!!

Mr. Analog

Quote from: Tom on February 11, 2020, 12:02:37 PM
Quote from: Lazybones on February 11, 2020, 09:15:50 AM
Quote from: Mr. Analog on February 11, 2020, 05:09:34 AMAre Portal blocks preserved?

I should be able to specifically protect chunks with obsidian/portal blocks


Just obisdian may not be ideal. perhaps obidian in chunks that have been in recently? Or if you can mark portal blocks themselves?

That may not be ideal as there could be little used portals that people have set up. Just spitballing here
By Grabthar's Hammer

Lazybones

I would prefer to keep the RULES for chunk removal simple or I am going to need to take several passes.

Inhabited time seems to be accurate and ideal for the over world, it trims out nearly all "explored", "traveled", "viewing distance" chunks and keeps ones people have actually spent time in.

The nether is a bit harder since it is so compact, time doesn't seem to help much.. Filtering by blocks that are likely ONLY player placed seems to work well and LOOKS ideal for doing a tight fit trim before the big 1.16 nether update.

Incidentally the amount of disk space savings is interesting, doing a test on the over world the directory size shrunk down more than half which should be good for server memory use.

mcselector is fairly easy to use and 1.16 won't be out for a while yet so we have plenty of time to discuss and run some tests.

An over world shrink might be a good early step then a tight nether shrink Just before / as part of the 1.16 upgrade might be best.

Again if you give me  X and Z coordinates (please note the dimension) I can go in a visualize what the surrounding shrink will look like.

Lazybones

I pulled a copy of the world locally and did a very harsh trim on the nether and loaded the latest snapshot.

Looks like it works very well, timing down to just nether areas where we have placed blocks means NEW nether generates all around the existing nether transport network, it blends in fairly well.

I think once 1.16 is official I am going to need help making a list of unique block IDs to save so I can make a quick accurate trim, however just filtering on torch, wall torch, packed_ice, stone, stone block, glass seems to give good results.. I also tested with nether_brick which saves the already discovered nether fortresses.