| Time |
Nick |
Message |
| 00:45 |
|
swift110 joined #minetest |
| 01:05 |
|
ANAND joined #minetest |
| 02:27 |
|
Scotty_Trees joined #minetest |
| 02:37 |
|
frostsnow joined #minetest |
| 02:45 |
|
Hawk777 joined #minetest |
| 03:01 |
|
Ruslan1 joined #minetest |
| 05:39 |
|
illwieckz_ joined #minetest |
| 06:27 |
|
calcul0n joined #minetest |
| 06:49 |
|
jluc joined #minetest |
| 07:06 |
|
CWz joined #minetest |
| 08:30 |
|
illwieckz joined #minetest |
| 08:43 |
|
illwieckz joined #minetest |
| 09:15 |
|
calcul0n joined #minetest |
| 09:45 |
|
illwieckz joined #minetest |
| 09:45 |
|
tune joined #minetest |
| 09:54 |
|
YuGiOhJCJ joined #minetest |
| 09:59 |
|
illwieckz joined #minetest |
| 10:06 |
|
calcul0n joined #minetest |
| 10:16 |
|
Beton joined #minetest |
| 10:37 |
|
illwieckz joined #minetest |
| 10:54 |
|
CWz joined #minetest |
| 11:01 |
|
greeter joined #minetest |
| 11:19 |
|
ChimneySwift joined #minetest |
| 11:36 |
|
illwieckz joined #minetest |
| 11:38 |
|
calcul0n joined #minetest |
| 11:54 |
|
Fixer joined #minetest |
| 12:36 |
|
Fixer joined #minetest |
| 13:11 |
|
Helenah joined #minetest |
| 13:16 |
|
illwieckz joined #minetest |
| 13:38 |
|
meseking joined #minetest |
| 13:39 |
meseking |
hello |
| 13:58 |
|
hecks joined #minetest |
| 14:00 |
|
Fulgen joined #minetest |
| 14:07 |
|
Beton_ joined #minetest |
| 14:23 |
|
scr267 joined #minetest |
| 14:23 |
|
rafalcpp joined #minetest |
| 14:29 |
|
meseking left #minetest |
| 14:40 |
|
illwieckz joined #minetest |
| 14:48 |
hecks |
if I calc lighting in on_generate, do I have immediate access to the lighting data? |
| 14:50 |
hecks |
say I generate my terrain, calc the lighting and then immediately thin out generated plants based on sunlight received |
| 14:50 |
hecks |
to avoid slow abm/lbm |
| 14:54 |
|
scr267 joined #minetest |
| 15:01 |
|
illwieckz joined #minetest |
| 15:14 |
|
jluc joined #minetest |
| 15:19 |
|
Weedy joined #minetest |
| 15:48 |
|
Fixer joined #minetest |
| 16:09 |
hecks |
someone spoonfeed me, how do I use a "biomemap" object? |
| 16:09 |
hecks |
it doesn't seem to 'fit' the chunk that voxelmanip represents |
| 16:18 |
|
illwieckz joined #minetest |
| 16:39 |
|
Orbstheorem joined #minetest |
| 16:41 |
Orbstheorem |
Hello ^^/, I was wondering if there is a minetest-based game comparable to age of engineering. |
| 16:46 |
|
paramat joined #minetest |
| 16:46 |
hecks |
I don't even know what that is |
| 16:47 |
|
Krock joined #minetest |
| 16:47 |
paramat |
"if I calc lighting in on_generate, do I have immediate access to the lighting data?" after 'write to map' yes |
| 16:47 |
hecks |
that'll do |
| 16:49 |
hecks |
maybe you can help me with this too; how can I do anything useful with a biomemap array? |
| 16:49 |
paramat |
if you 'get light data' in bulkas a 2nd vm maybe calc_lighting alters that too, not sure |
| 16:49 |
paramat |
yeah .. |
| 16:50 |
hecks |
say I want to turn water into sand in desert biomes; I call get_mapgen_object( "biomemap" ) |
| 16:50 |
hecks |
and it gives me an array of... 6400 elements? |
| 16:50 |
hecks |
while the voxelmanip's buffer is 1404928 elements |
| 16:50 |
Orbstheorem |
hecks: It's a technical modpack: https://minecraft.curseforge.com/projects/age-of-engineering |
| 16:50 |
paramat |
the 'mapgen object voxelmanip' is the mapchunk plus a mapblock thick shell, so is 80 +16 + 16 in size. however the heat/humidity/heightmaps are only 80x80 |
| 16:51 |
hecks |
so it's a 2D map? |
| 16:51 |
paramat |
those 3 maps are 2D |
| 16:52 |
paramat |
heat and humidity is 2D noise |
| 16:53 |
hecks |
this... sucks, because biomes are not necessarily 2D |
| 16:53 |
paramat |
the engine's own mapgen vm is also 'mapchunk plus mapblock shell' because it also processes an overlap volume, to allow trees to extend 16 nodes up beyond the mapchunk (essential) |
| 16:53 |
paramat |
biome y variation is done using y limits |
| 16:54 |
hecks |
I know that, but biomes overlapping in Y means that whatever data biomemap gives me will be bogus |
| 16:54 |
paramat |
so flat biome borders in y |
| 16:54 |
paramat |
ah |
| 16:56 |
paramat |
the biomemap for a mapchunk corresponds to the topmost solid surface in the mapchunk, ie the surface given by the heightmap |
| 16:56 |
hecks |
Orbstheorem: I think all we have is technic, and while complex, it's probably not what you're looking for |
| 16:56 |
paramat |
so yes it lacks information |
| 16:56 |
|
Gael-de-Sailly joined #minetest |
| 16:56 |
hecks |
well if it's just within the mapblock, it may work for something as simple as erasing water |
| 16:57 |
paramat |
however in 5.0.0 we do now have 'get biome data at (3D) point' which is accurate to how things generate |
| 16:57 |
hecks |
we do? |
| 16:57 |
hecks |
is it in the api yet? |
| 16:58 |
paramat |
https://github.com/minetest/minetest/blob/766fb7b46edab47f96b8b940e390daf5319261b7/doc/lua_api.txt#L3937 |
| 16:58 |
paramat |
yeah i coded it |
| 16:59 |
hecks |
hm... would this be much slower than using biomemap? |
| 16:59 |
hecks |
if it's just reading noise, then maybe it's acceptable |
| 17:00 |
paramat |
the 'biomemap' is just a simplified map used for mapgen, since most decorations are only placed on the topmost solid surface in a mapchunk |
| 17:00 |
paramat |
a little slower yes as the biomemap is precalculated |
| 17:00 |
hecks |
...unless you use the all surfaces flag or however it's called |
| 17:00 |
paramat |
indeed |
| 17:02 |
hecks |
I'll work something out |
| 17:02 |
hecks |
one last thing is bothering me; sqrt 6400 is 80, but voxelmanip gives me a 112^3 volume (I think) |
| 17:03 |
|
illwieckz joined #minetest |
| 17:03 |
paramat |
yeah, i explained that above |
| 17:03 |
hecks |
so uh, the difference is a padding of one mapblock, correct? |
| 17:03 |
paramat |
yes |
| 17:03 |
hecks |
and if I correct for that, the coordinates will line up |
| 17:03 |
paramat |
mapchunk is 5x5x5, full vm is 7x7x7 mapblocks |
| 17:03 |
hecks |
okay, thanks for all |
| 17:04 |
hecks |
...unless you fiddle with chunksize in the config |
| 17:04 |
hecks |
I wonder what that does other than changing the dungeon density |
| 17:05 |
paramat |
indeed, whatever chunksize, 1 mapblock padding is added |
| 17:05 |
paramat |
see https://github.com/minetest/minetest/blob/4a2a11262b0821d0bcddd25161f5eb8b172b23fd/minetest.conf.example#L2909 |
| 17:06 |
paramat |
highly recommended to leave it at 5 |
| 17:06 |
hecks |
I've read the desc, but I don't understand the consequences of using smaller values |
| 17:06 |
hecks |
I've used 3 before, the map didn't look much different but there were more dungeons |
| 17:06 |
paramat |
more caves too |
| 17:07 |
hecks |
and I suppose that results in more frequent on_generated calls |
| 17:08 |
paramat |
yes |
| 17:08 |
paramat |
btw i coded my own version of your terrain smoothing mod. you can use the 'heightmap' to immediately find the surface nodes to process them, instead of looping through every node |
| 17:08 |
hecks |
which is much faster, I presume |
| 17:09 |
hecks |
I'll work that in, since I gotta compute 2D coords either way |
| 17:10 |
hecks |
while we're on the topic of cryptic config values, what does "sidelen" in a decoration do? |
| 17:15 |
|
illwieckz joined #minetest |
| 17:20 |
paramat |
oh yeah .. |
| 17:20 |
hecks |
wait, if I use the heightmap to find ground nodes to smooth, smoothing will no longer work with caves, overhangs and ceilings |
| 17:22 |
paramat |
when decorations are placed with density altered by noise, 'sidelen' is the resolution of the density variation. it must divide into a mapchunk, so 80, 16, 8, 4 are the best values |
| 17:23 |
hecks |
bigger is faster? |
| 17:23 |
paramat |
ah, i didn't realise smoothing altered anything other than surface nodes. it also won't smooth under an overhang |
| 17:24 |
paramat |
yeah bigger sidelen means less noise calculations, but less resolution |
| 17:24 |
hecks |
yeah, I'll have to stick with my old smoother; I gotta iterate over the whole chunk for other things anyway |
| 17:24 |
hecks |
but the heightmap trick could be useful in the "plants die with no light, fungi die with light" round |
| 17:25 |
hecks |
as that will require its own loop |
| 17:25 |
hecks |
also, I understand that it's safe to skip iterating over the chunk padding area? |
| 17:26 |
paramat |
safe yes as that's an overlap |
| 17:30 |
timdorohin |
paramat: what you think, if i add support for loading mapgens as binary plugins (.dll/.so like plugins in browser) will it be accepted to mainline? |
| 17:31 |
rubenwardy |
I'd support such a thing |
| 17:31 |
rubenwardy |
depending on the complexity |
| 17:34 |
hecks |
it's the server's problem, so why not |
| 17:34 |
hecks |
I wouldn't mind having server side native code in general |
| 17:34 |
paramat |
plugin c++ mapgens? |
| 17:34 |
timdorohin |
Yea |
| 17:35 |
Krock |
just Compile It Yourself because binary packages from a random person does not look very trustworthy |
| 17:35 |
Krock |
native code sounds good - especially in terms of speed |
| 17:38 |
timdorohin |
Krock: i can compile my own version with my own mapgen code, but there are many people in community that didn't know how to do it. |
| 17:42 |
paramat |
the concept sounds good but it depends on practicalities and security |
| 17:42 |
|
progysm joined #minetest |
| 17:42 |
paramat |
also the amount of disruption and work needed at a time of lowest dev activity |
| 17:52 |
|
Fulgen joined #minetest |
| 17:57 |
|
Fulgen joined #minetest |
| 18:02 |
|
hecks joined #minetest |
| 18:08 |
|
Scotty_Trees joined #minetest |
| 18:39 |
rocky1138 |
Have you seen Amidst for Minetest? https://www.youtube.com/watch?v=oBs6YrYXlgU |
| 18:54 |
sfan5 |
!title |
| 18:54 |
MinetestBot |
sfan5: Amidst for Minetest - YouTube |
| 19:03 |
hecks |
so let us discuss parallel universes and why minetest doesn't need them |
| 19:04 |
hecks |
though the final word will be, as always, "code it yerself wiseguy" |
| 19:04 |
Hijiri |
I don't think I have enough time, but I have thought about writing it |
| 19:05 |
Hijiri |
It could be done incrementally in a reasonable way so that it doesn't require a mega-PR that conflicts with everything else |
| 19:05 |
hecks |
try making it work with *one* dimension in lua alone, first |
| 19:05 |
hecks |
or, code the rest of the damn game |
| 19:05 |
Hijiri |
"Minetest is an engine, not a game" |
| 19:06 |
Hijiri |
also it's easy to make it work with one dimension, just don't do anything |
| 19:06 |
hecks |
yeah, so... if you want to improve the engine, I suppose you have a game in mind that requires it |
| 19:07 |
Hijiri |
a mystcraft clone would be such a mod, not sure if I actually want to do that though |
| 19:07 |
Hijiri |
Maybe that's also why I'm not actually working on the feature |
| 19:08 |
Hijiri |
also is there a list somewhere of the most critical Minetest features to be implemented |
| 19:08 |
Hijiri |
there is the 5.0.0 project, but it only lists the CSM feature (and some other non-feature improvements) |
| 19:09 |
Hijiri |
Oh, there's the content brower |
| 19:09 |
Hijiri |
browser |
| 19:09 |
Shara |
Personally find it a little depressing when someone wants soemthign just so they can clone something else better. |
| 19:09 |
Shara |
Try and be original instead :P |
| 19:09 |
Shara |
something* |
| 19:10 |
hecks |
CSM is pretty critical, it would fix a lot of nasty issues with custom games |
| 19:10 |
hecks |
there is the content browser, yes |
| 19:10 |
Hijiri |
surely if there's one concept that benefits from infinite dimensions, there is more than one |
| 19:10 |
Hijiri |
the space of mods can't be that small |
| 19:10 |
hecks |
and some people somewhere are thinking of making the fabled VAOs, if they feel like it |
| 19:10 |
|
jluc joined #minetest |
| 19:10 |
Hijiri |
VAEs? |
| 19:10 |
hecks |
in general, minetest's feature schedule highly depends on people willing, able and feeling like doing said features |
| 19:11 |
hecks |
yeah, VAO/VAEs, entities made out of nodes |
| 19:11 |
|
p_gimeno joined #minetest |
| 19:11 |
Hijiri |
ah, ok |
| 19:11 |
hecks |
honestly, I could do VAOs in lua if colliders were not AABBs, and if attachments weren't so flimsy |
| 19:13 |
Hijiri |
I think it would be difficult to get node callbacks and things to work properly though |
| 19:13 |
MinetestBot |
[git] SmallJoker -> minetest/minetest_game: map: Use wood group crafting recipe f42deb4 https://git.io/fAbAO (2018-09-24T19:13:01Z) |
| 19:13 |
hecks |
definitely, they wouldn't be "nodes" in that sense |
| 19:14 |
hecks |
but if you just want to build airships, it would be good enough |
| 19:14 |
* hecks |
fondly remembers building a 3-mapblock-long frameship and losing it due to bugs |
| 19:14 |
|
behalebabo joined #minetest |
| 19:17 |
|
p_gimeno left #minetest |
| 19:17 |
hecks |
I think there are still more pressing concerns regarding minetest |
| 19:17 |
hecks |
like "where's the friggin' gameplay, what's the point" |
| 19:17 |
hecks |
I do believe not even minecraft got that solved |
| 19:18 |
hecks |
but one can dream https://a.uguu.se/xXJC0FKWWbyI_jelly.png |
| 19:18 |
|
AlienCat joined #minetest |
| 19:20 |
|
Lymkwi joined #minetest |
| 19:38 |
|
Taoki joined #minetest |
| 19:53 |
hecks |
I'm having on_generated called several times for the same minp, maxp range |
| 19:55 |
sfan5 |
happens |
| 19:56 |
hecks |
a lot, it appears |
| 19:56 |
hecks |
should I just file the range into some kind of table and skip the chunk if already generated? |
| 19:57 |
|
Scotty_Trees joined #minetest |
| 20:00 |
hecks |
huh, doesn't do it anymore |
| 20:00 |
hecks |
gremlins |
| 20:01 |
paramat |
using multiple mapgen threads? |
| 20:01 |
hecks |
I was |
| 20:01 |
hecks |
I'll recheck |
| 20:01 |
hecks |
also, get_biome_at is *slow* |
| 20:02 |
hecks |
oh yes, threaded emerge causes that |
| 20:03 |
paramat |
'get biome data' is not meant for using in very high numbers, and it's as fast as it can be :) |
| 20:04 |
hecks |
I'll use biomemap... |
| 20:05 |
hecks |
I've got a bigger issue to grind my teeth on now, not knowing if minetest is calling my post-mapgen several times more than necessary |
| 20:06 |
hecks |
thought this would work https://pastebin.com/jGmr2XZV |
| 20:07 |
hecks |
but it just causes some chunks to not get generated ever |
| 20:09 |
paramat |
should be no need if you're not using threaded emerge |
| 20:09 |
hecks |
I'd much, much rather be using threaded emerge |
| 20:09 |
hecks |
if you've seen my mapgen woes |
| 20:10 |
hecks |
could it be that emerge threads use bogus coordinates to avoid collision? |
| 20:11 |
paramat |
i can't see an issue for multiple mapchunk generation |
| 20:11 |
hecks |
https://a.uguu.se/pFoZwbe5xSpn_blah.png |
| 20:11 |
hecks |
this is what's happening, it's thinking that this chunk was already done, and skips it |
| 20:12 |
hecks |
if using the code pastebinned above |
| 20:12 |
hecks |
the only thing that could cause this if threaded emerge was fiddling with minp, maxp |
| 20:12 |
hecks |
in which case, "oh no, I'm getting the same chunk generated 3 times" is a false alarm and I don't need to do this |
| 20:16 |
paramat |
hmm |
| 20:17 |
paramat |
does your smoothing code run for every mapchunk, because 'calc lighting' is needed for every mapchunk due to 'nolight' being used |
| 20:18 |
paramat |
i guess it does |
| 20:18 |
hecks |
yes, while using that code snippet above to check if it isn't running on_generated multiple times |
| 20:19 |
hecks |
either way, one emerge thread seems faster than multiple, somehow |
| 20:20 |
hecks |
doesn't even choke globalstep as much |
| 20:23 |
hecks |
a setting of "0" also works fine, but I'm not sure if it's using 1 or some optimal number |
| 20:24 |
hecks |
don't really trust the documentation on this one, and cpu usage seems inconclusive (hovering around 50%) |
| 20:26 |
|
mrchiantos joined #minetest |
| 20:30 |
|
Taoki joined #minetest |
| 21:16 |
|
Cork joined #minetest |
| 21:22 |
|
paramat joined #minetest |
| 21:35 |
|
bigfoot547 joined #minetest |
| 21:35 |
|
bigfoot547 joined #minetest |
| 21:41 |
|
Orbstheorem left #minetest |
| 21:51 |
hecks |
%s, the %s died of dehydration in the middle of a desert https://a.uguu.se/IdxAlVX0hVYA_sand.png |
| 21:52 |
hecks |
and should probably thank paramat for its gruesome death |
| 21:53 |
sfan5 |
the committee of family friendliness has something to say about your player model |
| 21:54 |
timdorohin |
hecks: i want to unsee it |
| 21:56 |
hecks |
the committee can go get eaten by a grue |
| 22:01 |
hecks |
https://a.uguu.se/O3JyGNa5Fs4V_sw.gif I ain't got time for clothes y'know |
| 22:03 |
timdorohin |
hecks: is this minetest or scarlet blade? :D |
| 22:04 |
hecks |
well, the taskbar tells me it's minetest |
| 22:05 |
timdorohin |
hecks: :D |
| 22:37 |
|
YuGiOhJCJ joined #minetest |
| 22:45 |
hecks |
sfan5: the committee may be informed that their complaints have been heard, and the offending art was adjusted https://a.uguu.se/gTmlGub3FjRS_scr.png |
| 22:46 |
sfan5 |
very nice |
| 22:46 |
hecks |
further complaints may be directed to the grue |
| 22:48 |
hecks |
speaking of grues, is there a way to know if a node would be lit by sunlight if it were day, even when it isn't? |
| 22:49 |
paramat |
yes |
| 22:49 |
paramat |
'get light at pos' i think |
| 22:50 |
paramat |
has a parameter for time of day |
| 22:50 |
hecks |
good |
| 22:50 |
hecks |
this will also be useful to tell if rainfall would reach a certain node |
| 22:53 |
|
riff_IRC joined #minetest |
| 22:54 |
paramat |
https://github.com/minetest/minetest/blob/766fb7b46edab47f96b8b940e390daf5319261b7/doc/lua_api.txt#L3847 |
| 22:54 |
|
longerstaff13_ joined #minetest |
| 22:54 |
paramat |
should read: '0 for midnight, 0.5 for midday' |
| 22:55 |
|
exio41 joined #minetest |
| 22:56 |
|
lumidify_ joined #minetest |
| 22:57 |
paramat |
yeah 15 means under open sky |
| 22:58 |
|
ChimneySwift joined #minetest |
| 23:03 |
hecks |
oh right |
| 23:04 |
hecks |
I've had an issue with the model's file size |
| 23:05 |
hecks |
minetest caps it at some network packet size, uncompressed at that |
| 23:05 |
|
Fuchs joined #minetest |
| 23:07 |
|
superfly joined #minetest |
| 23:07 |
|
NathanS21 joined #minetest |
| 23:08 |
|
ssieb joined #minetest |
| 23:09 |
|
Ruslan1 joined #minetest |
| 23:10 |
|
el joined #minetest |
| 23:11 |
paramat |
hm i guess MT is not used to hidef models, maybe the limit can be raised with no problem |
| 23:12 |
sfan5 |
or throw compression on it |
| 23:13 |
sfan5 |
if that's easily doable |
| 23:16 |
|
ircSparky joined #minetest |
| 23:20 |
hecks |
https://github.com/minetest/minetest/issues/7744 |
| 23:20 |
hecks |
it's not the model that kills this limit, it's animation |
| 23:21 |
hecks |
and it will happen even with the blockiest model if it has enough animations |
| 23:22 |
hecks |
60 to 120 seconds of animations will exhaust the file size, depending on skeleton complexity |
| 23:47 |
|
Scotty_Trees joined #minetest |
