| Time |
Nick |
Message |
| 01:22 |
* sofar |
is just surprised nobody challenged his math skills this afternoon |
| 01:25 |
RealBadAngel |
yeah? telll me what is the very last digit of PI then ;) |
| 01:27 |
jomat |
pi = 3 |
| 01:28 |
sofar |
RealBadAngel: your password is in pi |
| 01:28 |
sofar |
clearly visible |
| 01:28 |
sofar |
to all of us |
| 01:32 |
RealBadAngel |
sofar, you have removed that inventory cube stuff for lava? |
| 01:35 |
RealBadAngel |
but with my patch or without? |
| 01:36 |
sofar |
I removed the inventory_cube lines in minetest_game for those nodes, and ran without your patch |
| 01:37 |
sofar |
just to see if we can remove those lines from minetest_game already preemtively |
| 01:37 |
RealBadAngel |
i see |
| 01:37 |
RealBadAngel |
that means that proper cubes could be here long time before but code was not fixed for it |
| 01:37 |
sofar |
the darkness I think is due to lava being light-emitting |
| 01:38 |
sofar |
normally it would be full lit |
| 01:38 |
sofar |
can we make sure that nodes are fully lit in the inventory mesh? |
| 01:38 |
RealBadAngel |
idk, on my client it rendered properl |
| 01:39 |
sofar |
I'm on nividia/linux |
| 01:39 |
VanessaE |
sofar, also check default:glass |
| 01:39 |
RealBadAngel |
me too |
| 01:39 |
sofar |
VanessaE: as said in the PR, glass was fine (bright, not dark) |
| 01:39 |
VanessaE |
ok. didn't see that. |
| 01:39 |
sofar |
obsidian is just... dark anyway |
| 01:39 |
RealBadAngel |
but whats the point in fixing things which my patch is makin obsolete? |
| 01:39 |
sofar |
VanessaE: well, the picture showed it was fine :) |
| 01:40 |
sofar |
just making sure it doesn't break old clients too much |
| 01:40 |
VanessaE |
I didn't see the picture :) |
| 01:40 |
RealBadAngel |
code which was used to produce that darkened lava is trashed by my PR |
| 01:40 |
VanessaE |
(I only tried the code with otherwise unmodified subgames/mods) |
| 01:42 |
RealBadAngel |
this picture show only what user will see if will connect with old client to server that runs modified minetest_game |
| 01:42 |
RealBadAngel |
server can be old too, it doesnt matter |
| 01:43 |
RealBadAngel |
if newer client will join old server user will see flat lava image |
| 01:44 |
RealBadAngel |
if game will be modified, then real cube |
| 01:44 |
RealBadAngel |
the PR is only side change |
| 01:45 |
RealBadAngel |
*client side |
| 01:45 |
RealBadAngel |
but mods can support it |
| 01:51 |
RealBadAngel |
btw, any comments on performance? does it affect fps or drawtime? |
| 02:02 |
RealBadAngel |
sofar, its kinda same situation like with your stone walls, if you will connect to server that has that mod running, with older clients, do you expect to see the walls at all? |
| 02:04 |
RealBadAngel |
in such cases theres no backwards compability or fixing things |
| 02:08 |
|
lezzy joined #minetest-dev |
| 02:16 |
|
H-H-H joined #minetest-dev |
| 02:17 |
|
lezzy2 joined #minetest-dev |
| 02:49 |
sofar |
RealBadAngel: nah, I was hoping it would be prettier, but I can live with it |
| 03:25 |
|
kaeza joined #minetest-dev |
| 05:50 |
Hijiri |
When I put an item in a custom group, can the associated values be non-numbers, as long as the users of the group aren't expecting numbers? |
| 05:51 |
Hijiri |
and have no bad effects, and be able to retrieve them later |
| 05:51 |
Hijiri |
I'm guessing you would be able to, since it's just put in the node definition, but I want to be sure |
| 06:03 |
sofar |
I wouldn't do that |
| 06:03 |
sofar |
why can't you use numbers? |
| 06:04 |
sofar |
put them in a custom field then |
| 06:08 |
|
jin_xi joined #minetest-dev |
| 06:50 |
|
Hunterz joined #minetest-dev |
| 07:01 |
|
Fritigern joined #minetest-dev |
| 07:12 |
|
kaeza joined #minetest-dev |
| 07:26 |
kaeza |
is there a formal approval process for translations or should I just go ahead and click "accept" on those I deem correct? |
| 07:26 |
kaeza |
on Weblate I mean |
| 07:29 |
kaeza |
there are 426 untranslated strings for es ;_; |
| 07:38 |
|
nrzkt joined #minetest-dev |
| 08:09 |
sofar |
I blow up.... https://youtu.be/ug_Wv6_n3YM have a good night! |
| 08:25 |
Hunterz |
nice... its better than minecraft tnt |
| 08:45 |
|
Obani joined #minetest-dev |
| 08:46 |
|
nrzkt joined #minetest-dev |
| 08:47 |
Hijiri |
sofar: ok, that makes sense |
| 08:48 |
Hijiri |
I don't know why I wanted my data in a group instead of just the node definition |
| 08:48 |
Hijiri |
it didn't even occur to me to put it somewhere else, lol |
| 09:37 |
|
mangeurdenuage left #minetest-dev |
| 09:41 |
|
Calinou joined #minetest-dev |
| 09:42 |
RealBadAngel |
wtf is that?? http://i.imgur.com/h3zeZOU.png |
| 09:42 |
everamzah |
happens often enough. it's related to the fetching of the public server list |
| 09:42 |
everamzah |
doesn't happen if you have that unchecked. |
| 09:43 |
everamzah |
i usually do "t t <enter>" three times to remove it hehe |
| 09:55 |
RealBadAngel |
no matter the error, but it looks like shit |
| 09:55 |
RealBadAngel |
such things should go to console only imho |
| 09:56 |
RealBadAngel |
fps dropped to 5fps with that for quite long moment |
| 10:12 |
|
blaze joined #minetest-dev |
| 10:45 |
kaeza |
the rotating wield items are great |
| 10:46 |
kaeza |
err, inventory items |
| 11:23 |
|
kahrl_ joined #minetest-dev |
| 12:24 |
|
lezzy joined #minetest-dev |
| 12:31 |
|
lezzy joined #minetest-dev |
| 12:34 |
lezzy |
nice tabbed inventory guys :D looks like someone liked my idea and used it in the mainstream |
| 12:36 |
|
Darcidride joined #minetest-dev |
| 12:50 |
|
Taoki joined #minetest-dev |
| 13:09 |
everamzah |
craftingpack did it first, i think |
| 13:27 |
|
Player_2 joined #minetest-dev |
| 13:42 |
|
Taoki[mobile] joined #minetest-dev |
| 14:11 |
|
Fixer joined #minetest-dev |
| 14:48 |
|
DFeniks joined #minetest-dev |
| 14:58 |
|
kaadmy joined #minetest-dev |
| 15:32 |
|
CraigyDavi joined #minetest-dev |
| 15:33 |
|
Taoki[mobile] joined #minetest-dev |
| 15:51 |
|
CraigyDavi joined #minetest-dev |
| 16:05 |
|
hmmmm joined #minetest-dev |
| 16:15 |
|
Obani joined #minetest-dev |
| 16:27 |
|
Terusthebird joined #minetest-dev |
| 16:27 |
|
RealBadAngel joined #minetest-dev |
| 16:29 |
RealBadAngel |
hi |
| 16:29 |
RealBadAngel |
https://github.com/minetest/minetest/pull/3612 |
| 16:31 |
RealBadAngel |
that kind of effect can be done without postprocessing, so here you go |
| 16:32 |
RealBadAngel |
it works unlike others per single pixel, so it doesnt need whole rendered image to work on |
| 16:33 |
RealBadAngel |
imho it enhances the gameplay a lot, world seen through HDR eyes is quite different ;) |
| 16:35 |
kaadmy |
how difficult would SSAO be? |
| 16:36 |
Calinou |
what's the point of SSAO when blocks already have AO? |
| 16:36 |
kaadmy |
it's postprocessing tho, can mt handle that? |
| 16:36 |
Calinou |
good AO, without noise… |
| 16:36 |
kaadmy |
Calinou: nodeboxes, meshes, models? |
| 16:36 |
Calinou |
"uncharted2tonemap" |
| 16:37 |
Calinou |
RealBadAngel, did you extract it straight from disassembly? at least don't state it… |
| 16:37 |
Calinou |
it's illegal |
| 16:37 |
RealBadAngel |
no |
| 16:37 |
RealBadAngel |
its public |
| 16:37 |
RealBadAngel |
everybody around is using it, ask google |
| 16:38 |
RealBadAngel |
its called like that because author of the algorithm made it for this game |
| 16:38 |
sofar |
you still have two unused vars |
| 16:39 |
RealBadAngel |
so we do have tonemaps: Reinhard, Uncharted2 etc |
| 16:40 |
RealBadAngel |
its like you were using Gausian blur and was afraid to be sued because you have used something made by Gauss |
| 16:42 |
sofar |
do I need to compile with ENABLE_TONEMAPPING? |
| 16:42 |
RealBadAngel |
no |
| 16:42 |
RealBadAngel |
use setting in menu |
| 16:43 |
sofar |
ah shaders |
| 16:43 |
sofar |
yes, I did that |
| 16:43 |
sofar |
having a hard time seeing effect, tho |
| 16:43 |
RealBadAngel |
ENABLE_TONEMAPPING is internal shaders flag used when compiling shaders |
| 16:43 |
|
Amaz joined #minetest-dev |
| 16:43 |
RealBadAngel |
open two instances of mt, one by one |
| 16:44 |
RealBadAngel |
you will see the differences in contrast |
| 16:44 |
sofar |
Oh, no |
| 16:44 |
RealBadAngel |
dark and bright things are redistributed equally |
| 16:44 |
sofar |
I had tonemapping enabled but shaders disabled |
| 16:44 |
RealBadAngel |
oops ;) |
| 16:45 |
RealBadAngel |
but, explaining is worth shit |
| 16:45 |
RealBadAngel |
you HAVE to see it |
| 16:46 |
|
Krock joined #minetest-dev |
| 16:48 |
sofar |
it's... very subtle |
| 16:49 |
RealBadAngel |
open two instances side by side |
| 16:49 |
RealBadAngel |
and compare |
| 16:49 |
sofar |
I did |
| 16:49 |
RealBadAngel |
also fly around a bit in valleys mapgen with new effect |
| 16:49 |
sofar |
sorry, maybe it would help if you posted some screenshots as well, so people can see what they're supposed to see |
| 16:49 |
RealBadAngel |
then go back |
| 16:49 |
RealBadAngel |
you will feel something like loosing focus |
| 16:50 |
RealBadAngel |
those effects are about percepting the colors |
| 16:50 |
RealBadAngel |
it may vary from one to another |
| 16:50 |
RealBadAngel |
i find this a huge difference |
| 16:51 |
RealBadAngel |
everything gets right contrast, light seems to be spreading further and smoother |
| 16:51 |
sofar |
contrast is improved |
| 16:52 |
sofar |
that's the one thing I can see |
| 16:53 |
RealBadAngel |
thats the main point |
| 16:53 |
RealBadAngel |
to see both dart and light things equally good |
| 16:53 |
RealBadAngel |
wrr |
| 16:54 |
RealBadAngel |
dark and bright |
| 16:54 |
RealBadAngel |
anyway, this one is only the one of the possible to come |
| 16:54 |
RealBadAngel |
there are several other methods to pick from |
| 16:55 |
RealBadAngel |
personally my eyes found it the best one |
| 16:55 |
Amaz |
Both your tonemapping and your inventory meshes pulls are awesome RealBadAngel! |
| 16:55 |
RealBadAngel |
maybe you will find another better fitting your eyes |
| 16:56 |
Amaz |
(I really like the spinning items) |
| 16:56 |
RealBadAngel |
but still, i hardly recomend playing with tonemapping for quite a while |
| 16:56 |
RealBadAngel |
and return to non modified |
| 16:57 |
RealBadAngel |
i can quarantee you will not want to play without it anymore |
| 16:58 |
RealBadAngel |
Amaz, thx |
| 16:59 |
sofar |
I saw you fixed the meshes being too light, too |
| 17:03 |
|
red_001 joined #minetest-dev |
| 17:16 |
Amaz |
A before/after screenshot of the tonemapping: https://i.imgur.com/imE3POs.png (imo, it's soooo much better now!) |
| 17:18 |
Calinou |
difference is not really visible |
| 17:20 |
|
Hunterz joined #minetest-dev |
| 17:29 |
|
jin_xi joined #minetest-dev |
| 17:38 |
RealBadAngel |
Calinou, ive added some shots to the PR to compare |
| 17:39 |
sofar |
Amaz: very, very subtle difference, I can't see them side to side |
| 17:39 |
RealBadAngel |
look at the trees in my post |
| 17:40 |
RealBadAngel |
top is effect on/bottom off |
| 17:40 |
sofar |
I had to overlay the images in gimp and flick in between them to see |
| 17:41 |
RealBadAngel |
best to open both screenshots fullsized and toggle between |
| 17:41 |
Fixer |
very subtle difference |
| 17:41 |
RealBadAngel |
original one is fuzzy |
| 17:41 |
Fixer |
what about caves? |
| 17:41 |
RealBadAngel |
thats the whole point, everything gets equal contrast |
| 17:42 |
RealBadAngel |
theres a series in caves too |
| 17:42 |
|
turtleman joined #minetest-dev |
| 17:43 |
RealBadAngel |
also, theres a paramter i used to compensate snow becoming too bright |
| 17:43 |
RealBadAngel |
gamma |
| 17:44 |
red-001 |
RBA what does #3016 do? |
| 17:44 |
ShadowBot |
https://github.com/minetest/minetest/issues/3016 -- [0.4.13-RC] Bumpy Item Textures |
| 17:44 |
red-001 |
I mean #3612 |
| 17:44 |
ShadowBot |
https://github.com/minetest/minetest/issues/3612 -- Filmic HDR tone mapping by RealBadAngel |
| 17:46 |
Fixer |
https://github.com/minetest/minetest/issues/2350 this issue should probably be closed |
| 17:49 |
|
Player_2 joined #minetest-dev |
| 17:51 |
RealBadAngel |
red-001, see screenshots and preferably just play with the effect on |
| 17:52 |
RealBadAngel |
it changes the perception of the world |
| 17:54 |
|
nrzkt joined #minetest-dev |
| 17:59 |
sofar |
red-001: fyi I sent fixes to mesecons for the new door API. |
| 18:00 |
Fixer |
!tell paramat https://github.com/minetest/minetest/issues/2350 this issue should probably be closed |
| 18:00 |
ShadowBot |
Fixer: O.K. |
| 18:15 |
RealBadAngel |
Fixer, indeed |
| 18:16 |
RealBadAngel |
sofar, btw: http://image.slidesharecdn.com/hablejohnuncharted2hdrlighting-100331075253-phpapp02/95/hable-john-uncharted2-hdr-lighting-144-728.jpg?cb=1270022334 |
| 18:19 |
RealBadAngel |
with the last sentence i can argue, its worth a try |
| 18:30 |
Fixer |
RealBadAngel, i'm not sure why it changes my life completely it is very-very subtle difference |
| 18:32 |
RealBadAngel |
hehe, it changes the feeling |
| 18:33 |
RealBadAngel |
for my eyes, without it, mt is just washed out |
| 18:33 |
RealBadAngel |
fuzzy |
| 18:33 |
RealBadAngel |
hard to focus on details |
| 18:33 |
Fixer |
RealBadAngel, feels more balanced |
| 18:34 |
RealBadAngel |
exactly |
| 18:35 |
RealBadAngel |
i will later on add more filter defined by other folks |
| 18:35 |
Fixer |
RealBadAngel, have you noticed slight difference with snow node rendering behind lone water block here? https://cloud.githubusercontent.com/assets/2177790/12558742/8aa84d7a-c391-11e5-9b10-09c1b5974c18.png |
| 18:35 |
RealBadAngel |
ive picked this one first because its seems to be the most popular |
| 18:36 |
Fixer |
RealBadAngel, how much is fps decline? |
| 18:36 |
RealBadAngel |
doubt if any |
| 18:36 |
RealBadAngel |
code is quite simple |
| 18:37 |
RealBadAngel |
i cant see any difference with and without |
| 18:37 |
sofar |
we should really have a demo/minebench |
| 18:38 |
sofar |
fixed seed, fixed settings, start up and run a certain scene and draw frames |
| 18:38 |
sofar |
without vblank |
| 18:38 |
RealBadAngel |
yeah |
| 18:38 |
RealBadAngel |
that would be useful |
| 18:38 |
sofar |
RealBadAngel: great idea! when are you making it |
| 18:38 |
sofar |
it was your idea... |
| 18:38 |
RealBadAngel |
what? the benchmark? |
| 18:39 |
sofar |
yes |
| 18:39 |
RealBadAngel |
lol |
| 18:39 |
RealBadAngel |
good try |
| 18:39 |
sofar |
oh come on, nobody would believe me if I would claim to come up with anything new |
| 18:39 |
Fixer |
mt does have --benchmark iirc |
| 18:39 |
sofar |
everyone knows I steal code ;) |
| 18:40 |
Calinou |
Minetest is a terrible benchmark |
| 18:40 |
Calinou |
Fixer, no? |
| 18:40 |
sofar |
Calinou: not meant to benchmark hardware, meant to test performance difference of patches to minetest |
| 18:41 |
Fixer |
minetest --speedtests |
| 18:41 |
sofar |
that doesn't render |
| 18:41 |
RealBadAngel |
idea is good, but then you will have to face dev who is using freebsd as a gaming platform |
| 18:42 |
sofar |
data is always good, and good data is even better |
| 18:42 |
RealBadAngel |
long time i was wondering why hmmmm is getting such small framerate |
| 18:42 |
Fixer |
hmmm uses freebsd? |
| 18:42 |
RealBadAngel |
yeah |
| 18:42 |
Fixer |
._. |
| 18:42 |
RealBadAngel |
there are no real drivers for it |
| 18:43 |
Fixer |
there are some progress actually |
| 18:43 |
RealBadAngel |
some for nvidia, completely nothing for amd |
| 18:43 |
Calinou |
the NVIDIA proprietary driver for FreeBSD is the same as the one for Linux |
| 18:43 |
Calinou |
it's just updated less often |
| 18:43 |
|
ssieb joined #minetest-dev |
| 18:43 |
sofar |
it's not relevant, even |
| 18:43 |
RealBadAngel |
i wont discuss it, googling about it is an answer |
| 18:44 |
sofar |
there is no way to measure impact of patches right now |
| 18:44 |
sofar |
any way is better than none |
| 18:44 |
RealBadAngel |
yes |
| 18:44 |
RealBadAngel |
for that we could have sample world |
| 18:44 |
sofar |
just pick seed 0 |
| 18:44 |
RealBadAngel |
fixed database with fly around |
| 18:45 |
RealBadAngel |
keep it small |
| 18:45 |
sofar |
you may want to generate terrain during the test |
| 18:45 |
RealBadAngel |
no need for regular world |
| 18:45 |
sofar |
but, a fixed world may be good too |
| 18:45 |
Fixer |
fixed seed, emerge blocks, wait a little, fly around a little, ???, profit! |
| 18:45 |
Calinou |
a regular world should be used, to be close to real-world conditions |
| 18:45 |
RealBadAngel |
it can be pre generated |
| 18:45 |
sofar |
hmm |
| 18:45 |
sofar |
pre-generated is fine |
| 18:45 |
RealBadAngel |
that would be even better |
| 18:45 |
sofar |
we can test emerge separately easy enough |
| 18:45 |
RealBadAngel |
you wont be measuring mapgen or disk |
| 18:46 |
sofar |
can we hard-code a demo path? |
| 18:46 |
RealBadAngel |
and as a bonus you can use it as main menu background |
| 18:46 |
sofar |
speaking of settings, I'd want it to actually use the current settings |
| 18:46 |
sofar |
so you can compare setting performance |
| 18:47 |
sofar |
it should even be a UI button |
| 18:47 |
sofar |
"TEST ME" |
| 18:47 |
RealBadAngel |
put there "fps" stamp and if its ok next one "approved" |
| 18:47 |
RealBadAngel |
;) |
| 18:47 |
sofar |
click it and a 60 second demo runs, then displays results |
| 18:47 |
sofar |
or have it run permanently |
| 18:47 |
sofar |
loop, perhaps |
| 18:47 |
RealBadAngel |
if running as background? why not |
| 18:47 |
sofar |
you don't want a random path |
| 18:48 |
sofar |
it should be hardcoded |
| 18:48 |
RealBadAngel |
anything could be better than clouds |
| 18:48 |
RealBadAngel |
recorded rather |
| 18:48 |
sofar |
first things first, a separate demo mode |
| 18:48 |
RealBadAngel |
pretty nice fly by |
| 18:48 |
sofar |
a simple map can be quite large |
| 18:48 |
sofar |
we may want to package it separately |
| 18:49 |
sofar |
can we embed a path into the world? |
| 18:49 |
sofar |
so that demos can be downloaded separately? |
| 18:49 |
sofar |
e.g. map+path embedded? |
| 18:49 |
sofar |
load demo file? |
| 18:50 |
sofar |
maybe /demos? |
| 18:51 |
sofar |
RealBadAngel: right, scenic hill top flyovers, dive through some river valleys, into a cave |
| 18:51 |
kaadmy |
hmm |
| 18:52 |
kaadmy |
maybe a flyby of the world to-be with the mapgen? |
| 18:52 |
kaadmy |
with the current seed? |
| 18:52 |
sofar |
RealBadAngel: could be fun adding manually crafted blocks and making buildings and structures to demo what can be done in minetest and to test various node types |
| 18:52 |
kaadmy |
would be interesting but hard to implement afaik |
| 18:53 |
sofar |
kaadmy: generating is faster than making another way to to "play" it |
| 19:07 |
Fixer |
make benchmark subgame ._. |
| 19:11 |
sofar |
Fixer: hmmmm, but subgame can't load a predefined map, that still requires the user to 'play game demo with map demo' |
| 19:11 |
sofar |
but yes, that is one way to hack it together |
| 19:12 |
Fixer |
will be nice to have option to generate map chanks like in game, to test stutters |
| 19:12 |
sofar |
to test that, just test the game |
| 19:13 |
sofar |
I really want to test frame rate in an environment where I can eliminate mapgen and network issues |
| 19:15 |
Fixer |
problem is that mapgen and network issues have big effect on fps :) |
| 19:16 |
sofar |
problem is that we can test those already |
| 19:16 |
sofar |
but we have no way to test framerate |
| 19:18 |
Fixer |
two options: with mapgen and pregenerated |
| 19:18 |
sofar |
I'm also convinced that the buggy minetest.after() implementation currently is causing havoc massively everywhere |
| 19:18 |
sofar |
but it seems my PR's to fix it are not getting the right review |
| 19:19 |
|
hmmmm joined #minetest-dev |
| 19:29 |
|
louiloui joined #minetest-dev |
| 19:30 |
RealBadAngel |
taklkin bout reviews, what about film filter? |
| 19:42 |
Calinou |
what about Instagram filter? |
| 19:42 |
Calinou |
turns everything blue-ish |
| 19:42 |
sofar |
hahahaha |
| 19:42 |
sofar |
do eeeet |
| 19:47 |
hmmmm |
btw |
| 19:47 |
RealBadAngel |
you may laugh, but with irrpp theres a filter called I See Red |
| 19:47 |
kaadmy |
is it called D3D? |
| 19:47 |
RealBadAngel |
guess what is it doing ;) |
| 19:48 |
hmmmm |
if you're seriously going to make minetest use a deferred shading strategy, check out far cry 2 |
| 19:48 |
hmmmm |
they used deferred shading and it works o.k. for the most part, if you're fine with indoor and underground areas looking bad and pixelated shadows |
| 19:50 |
hmmmm |
i think, ultimately, we're going to need better heuristics of whether or not something is underground/inside |
| 19:50 |
RealBadAngel |
hmmm, thats quite easy and almost done |
| 19:50 |
RealBadAngel |
see lightbanks |
| 19:50 |
hmmmm |
static landscape games can do this rather trivially by allowing the level designers to set a region around what is supposed to be considered indoors |
| 19:51 |
RealBadAngel |
night holds everything artificial |
| 19:51 |
hmmmm |
the best we can do, though, is somehow optimize a flood fill algorithm to determine if something is indoor/outdiir |
| 19:51 |
RealBadAngel |
day - both |
| 19:51 |
Calinou |
hmmmm, shadows being pixelated has nothing to do with deferred rendering |
| 19:51 |
Calinou |
some forward renders have shit shadows, like Godot's |
| 19:51 |
hmmmm |
shadow mapping? |
| 19:51 |
RealBadAngel |
3rd bank is needed that would hold only sun affected areas |
| 19:51 |
hmmmm |
shadows being pixelated has to do with the performance tradeoff necessary |
| 19:51 |
Calinou |
good shadow mapping (like Variance Shadow Mapping), when blurred + anti-aliasing, will look great |
| 19:52 |
Calinou |
it's fully doable on today's mid-end systems |
| 19:52 |
hmmmm |
in any case |
| 19:52 |
RealBadAngel |
i can blur + fxaa the final result |
| 19:52 |
RealBadAngel |
i do have postrpocessing by now |
| 19:52 |
hmmmm |
I'm thinking maybe this flood fill I'm talking about would be done on map generation, and the results would be updated per setnode |
| 19:52 |
hmmmm |
maybe we can steal a bit from somewhere |
| 19:52 |
hmmmm |
node_is_indoor/outdoor |
| 19:53 |
RealBadAngel |
hmmmm, baw is using floodfill too |
| 19:53 |
Calinou |
FXAA would be nice to have, though SMAA is better |
| 19:53 |
hmmmm |
man I don't care what BAW is doing |
| 19:53 |
hmmmm |
a few years later, where are they right now? |
| 19:53 |
RealBadAngel |
i know but they are using irr |
| 19:53 |
hmmmm |
a heavily modified version |
| 19:53 |
Calinou |
BAW? |
| 19:53 |
hmmmm |
build-a-world |
| 19:53 |
RealBadAngel |
they were in similar situation like us |
| 19:54 |
hmmmm |
they modified irrlicht. |
| 19:54 |
hmmmm |
that is a lot of work you know |
| 19:54 |
RealBadAngel |
not really |
| 19:54 |
hmmmm |
maintaining a new package? |
| 19:54 |
hmmmm |
of our own |
| 19:54 |
RealBadAngel |
for per pixel lighting to work i just need that 3rd exclusive for sun(moon) bank to make it all work |
| 19:55 |
hmmmm |
what do you mean 3rd exclusive |
| 19:55 |
RealBadAngel |
we do have 2 |
| 19:55 |
RealBadAngel |
night one is torches |
| 19:55 |
hmmmm |
you mean you want to add a third light bank? |
| 19:55 |
RealBadAngel |
2nd is sun + torches |
| 19:55 |
RealBadAngel |
i NEED one only SUN |
| 19:55 |
hmmmm |
I was thinking of getting rid of the concepts of light banks entirely |
| 19:56 |
RealBadAngel |
look |
| 19:56 |
hmmmm |
I don't see the value in it |
| 19:56 |
RealBadAngel |
problem with HW light ray is that its an fucking xray |
| 19:56 |
RealBadAngel |
it shines on everything |
| 19:56 |
RealBadAngel |
no mater indoor, surface or cave |
| 19:57 |
hmmmm |
that doesn't sound quite right |
| 19:57 |
RealBadAngel |
tell me how not to shine the caves with sun then |
| 19:57 |
RealBadAngel |
above is my idea |
| 19:57 |
hmmmm |
look I don't know, but hardware dynamic lights shouldn't do that as long as the overhead mesh is there |
| 19:58 |
RealBadAngel |
no |
| 19:58 |
RealBadAngel |
it goes through |
| 19:58 |
hmmmm |
why |
| 19:58 |
RealBadAngel |
thats how it works |
| 19:58 |
hmmmm |
i don't buy it |
| 19:58 |
hmmmm |
that's not how it's supposed to work |
| 19:58 |
RealBadAngel |
hehe |
| 19:59 |
RealBadAngel |
believe it or not, but thats how it works |
| 19:59 |
hmmmm |
are you continuously changing the light's position or something..? |
| 19:59 |
RealBadAngel |
no matter if i do that or not |
| 19:59 |
hmmmm |
ehh.. |
| 19:59 |
hmmmm |
I'm gonna try it myself |
| 20:00 |
RealBadAngel |
only thing that counts is the surface normal |
| 20:00 |
RealBadAngel |
when you need to stop the ray, you have to do it on your own |
| 20:01 |
RealBadAngel |
thats why i wanted lightlevels |
| 20:01 |
RealBadAngel |
if i pass them on vertices i can allow lightsource to act (or deny) |
| 20:01 |
RealBadAngel |
btw, HW lighting is about using shaders |
| 20:02 |
RealBadAngel |
almost all advanced materials are made with shaders |
| 20:02 |
hmmmm |
i'm talking about hardware dynamic lights |
| 20:02 |
RealBadAngel |
all the calculations are done in shaders |
| 20:03 |
RealBadAngel |
irrlicht support those shaders |
| 20:03 |
hmmmm |
really |
| 20:03 |
RealBadAngel |
if you or anybody turns off shaders in mt, you think irr is not using them? |
| 20:03 |
hmmmm |
so irrlicht has a set of default shaders that it uses when you enable a hardware dynamic light |
| 20:03 |
RealBadAngel |
what a suprise ;) |
| 20:03 |
RealBadAngel |
yes |
| 20:04 |
RealBadAngel |
if you decide to use your own you have to do calculations on your own |
| 20:04 |
RealBadAngel |
btw, bumpmapping, parallax are native irrlicht materials |
| 20:05 |
RealBadAngel |
those shaders are just built in |
| 20:06 |
RealBadAngel |
if you let material to be lit by hw light and dont enable mt shaders, built in engine shaders will do that for you |
| 20:06 |
RealBadAngel |
if you enable shaders for your materials, you have to do the math |
| 20:07 |
RealBadAngel |
and then lightsource becomes just a moving point, and you need just its position |
| 20:07 |
RealBadAngel |
i do have all the math in shaders for the long time already |
| 20:08 |
RealBadAngel |
what i dont have (working reliably) is when i should let the sun/moon shine in and when i should deny it |
| 20:08 |
RealBadAngel |
in fact current system is ok, until i lit a single torch |
| 20:09 |
sofar |
I wish the moon wasn't in the L2 lagrangian point in minetest |
| 20:10 |
RealBadAngel |
i can split them, using Fiat Lux.py lib |
| 20:10 |
RealBadAngel |
thats no problem |
| 20:10 |
hmmmm |
RBA, are you sure that's how dynamic lights work in irrlicht? |
| 20:11 |
RealBadAngel |
yes |
| 20:11 |
RealBadAngel |
do you need code to test? screenshots? |
| 20:11 |
hmmmm |
http://sourceforge.net/p/irrlicht/code/HEAD/tree/trunk/source/Irrlicht/COpenGLDriver.cpp#l3023 |
| 20:12 |
RealBadAngel |
yes |
| 20:12 |
hmmmm |
so where is irrlicht's dynamic light shader code? |
| 20:12 |
RealBadAngel |
that just is passing light source pos to shaders |
| 20:12 |
RealBadAngel |
then you can read pos ambient etc in shaders |
| 20:13 |
RealBadAngel |
grep sources |
| 20:13 |
RealBadAngel |
shaders are partially compiled |
| 20:13 |
RealBadAngel |
similar to assembly |
| 20:13 |
hmmmm |
yes but where are they |
| 20:14 |
RealBadAngel |
i will find that for you, but gimme half an hour, cooking now :) |
| 20:14 |
hmmmm |
really cause i can't find them |
| 20:15 |
RealBadAngel |
make that: |
| 20:15 |
RealBadAngel |
enable lighting |
| 20:15 |
RealBadAngel |
disable final color blend on both |
| 20:15 |
RealBadAngel |
cpu and gpu |
| 20:15 |
hmmmm |
i think you're talking out of your ass |
| 20:15 |
|
Robert_Zenz joined #minetest-dev |
| 20:15 |
RealBadAngel |
really? |
| 20:15 |
hmmmm |
yeah sorry man |
| 20:15 |
RealBadAngel |
i will just upload my branch |
| 20:15 |
RealBadAngel |
you will see on your own |
| 20:16 |
|
PenguinDad joined #minetest-dev |
| 20:19 |
|
Wayward1 joined #minetest-dev |
| 20:27 |
RealBadAngel |
hmmm, the code yuo have pasted from irr, is making this possible: https://www.opengl.org/sdk/docs/tutorials/ClockworkCoders/lighting.php |
| 20:27 |
|
AnotherBrick joined #minetest-dev |
| 20:28 |
RealBadAngel |
whatever refers to light, like vec3 L = normalize(gl_LightSource[0].position.xyz - v); |
| 20:28 |
RealBadAngel |
vec4 Idiff = gl_FrontLightProduct[0].diffuse * max(dot(N,L), 0.0); |
| 20:28 |
RealBadAngel |
is because of the code you have highlighted |
| 20:29 |
RealBadAngel |
theres no black hardware magic for lights |
| 20:29 |
RealBadAngel |
there are just shaders :P |
| 20:30 |
RealBadAngel |
you have set the material, have moving source |
| 20:30 |
RealBadAngel |
its coordinates are passed to shaders |
| 20:30 |
RealBadAngel |
be it one, or many, they are processed one by one |
| 20:31 |
RealBadAngel |
per each pixel |
| 20:31 |
RealBadAngel |
or vertex if you like (that would be very similar to our smooth lighting) |
| 20:44 |
hmmmm |
like I said; I don't buy it |
| 20:44 |
hmmmm |
you have no idea what's going on at the driver level |
| 20:44 |
hmmmm |
on AMD cards this is probably true. on nVidia, probably not so much. |
| 20:44 |
RealBadAngel |
https://github.com/Paulus/Irrlicht/blob/1.7/source/Irrlicht/CD3D8ParallaxMapRenderer.cpp#L21 |
| 20:44 |
RealBadAngel |
shut up and read example |
| 20:45 |
hmmmm |
OpenGL is mostly implemented on the silicon with nvidia cards |
| 20:45 |
RealBadAngel |
this is parallax shader with lights |
| 20:45 |
RealBadAngel |
and i expect you will say sorry at least |
| 20:45 |
hmmmm |
yeah, for d3d8 |
| 20:46 |
RealBadAngel |
https://github.com/Paulus/Irrlicht/blob/1.7/source/Irrlicht/COpenGLParallaxMapRenderer.cpp |
| 20:46 |
RealBadAngel |
here you go |
| 20:46 |
RealBadAngel |
wheres the "sorry" |
| 20:46 |
hmmmm |
nowhere |
| 20:47 |
hmmmm |
you're showing me something completely unrelated and telling me it proves your point |
| 20:47 |
RealBadAngel |
"# transform the light vector 2 with U, V, W \n"\ |
| 20:47 |
hmmmm |
yes, lots of fragment shaders transform light |
| 20:47 |
RealBadAngel |
do you think what are the lights here? |
| 20:47 |
RealBadAngel |
those are your HW lights |
| 20:48 |
hmmmm |
maybe if you're using the parllax map renderer |
| 20:48 |
RealBadAngel |
shaders are GPU code, your hardware |
| 20:48 |
RealBadAngel |
awww, cmon |
| 20:48 |
RealBadAngel |
i thought youre smarter |
| 20:48 |
RealBadAngel |
go dig for yourself |
| 20:49 |
RealBadAngel |
when go back, we can talk again |
| 20:49 |
|
VanessaE left #minetest-dev |
| 20:49 |
hmmmm |
so you admit you can't find the thing you were telling me exists |
| 20:49 |
|
VanessaE joined #minetest-dev |
| 20:49 |
hmmmm |
in other words you were full of bullshit |
| 20:49 |
hmmmm |
it's just that nobody else ever calls you out on it |
| 20:49 |
RealBadAngel |
you want regular material |
| 20:49 |
RealBadAngel |
ok |
| 20:50 |
hmmmm |
I don't think it's too farfetched to believe that hardware lights are implemented in the hardware on at least some cards |
| 20:50 |
hmmmm |
others, maybe the driver has the shader code |
| 20:51 |
ssieb |
hmmmm: are you saying that "lights" are going to be automatically blocked by geometry without the shader having to do anything? |
| 20:51 |
hmmmm |
on at least some nvidia cards there really is dedicated hardware light hardware thogh |
| 20:53 |
RealBadAngel |
hmmm, why you cant you just enable lights in minetest and see how it works? |
| 20:53 |
kaadmy |
hmmmm: afaik no gfx cards support hw lights yet |
| 20:53 |
hmmmm |
that's what I'm doing |
| 20:53 |
RealBadAngel |
ok |
| 20:53 |
RealBadAngel |
nothing better than find things on your own |
| 20:54 |
kaadmy |
hmmmm: so you want mt to have lighting only on nvidia? |
| 20:54 |
hmmmm |
huh? |
| 20:54 |
hmmmm |
no i said that nvidia probably has support for hardware dynamic lighting right in the silicon, whereas amd cards implement it with a shader built into the driver |
| 20:54 |
RealBadAngel |
btw, i do have lighting working flawlessly, with sun/moon cycle |
| 20:55 |
hmmmm |
but it certainly does NOT happen at the level of abstraction that rba is saying it does |
| 20:55 |
VanessaE |
RealBadAngel: link to the code? :P |
| 20:55 |
RealBadAngel |
just cant block reliably the ray to not shine interiors or caves |
| 20:55 |
hmmmm |
yeah you keep telling us about it, and you told me that you were going to upload it |
| 20:55 |
hmmmm |
but where is it |
| 20:56 |
hmmmm |
ssieb, yes, isn't that the point of a dynamic light? |
| 20:56 |
RealBadAngel |
its a huge mess with shitload of pieces here and here, gimme hour or two and i will upload essential parts to get it working |
| 20:57 |
VanessaE |
translation: <RealBadAngel> I have a lot of polishing to do, expect it in 1-2 weeks |
| 20:57 |
VanessaE |
:P |
| 20:57 |
RealBadAngel |
now im going to eat with my wife or she is going to kill me |
| 20:57 |
hmmmm |
she'll grind you up and stuff you into a kielbasa |
| 20:57 |
RealBadAngel |
VanessaE, please dont mark something partially workin as rdy for PR |
| 20:58 |
RealBadAngel |
hmmmm, or worse |
| 20:58 |
VanessaE |
RealBadAngel: I'm not. compileable, runnable code or it's all just a bunch of mockups. :) |
| 20:58 |
RealBadAngel |
im outta here, be back a bit later |
| 21:08 |
ssieb |
hmmmm: I'm pretty sure that there is no triangle intersection code build into the hardware like that... |
| 21:08 |
ssieb |
can you show any evidence for it existing? |
| 21:13 |
ssieb |
and assuming that information did get created, how would it get passed to the shader? |
| 21:46 |
RealBadAngel |
VanessaE, problem is my playground is a mess, theres irrpp, irrrenderer and some other small things merged |
| 21:46 |
VanessaE |
yeah yeah sure ;) |
| 21:46 |
RealBadAngel |
im trying to get them all to work |
| 21:46 |
RealBadAngel |
but |
| 21:46 |
RealBadAngel |
ive sent hmmm a pm with steps to reproduce what i was talking about |
| 21:46 |
RealBadAngel |
quite simple ones |
| 21:47 |
RealBadAngel |
http://pastebin.com/54W9ABiP |
| 21:49 |
RealBadAngel |
this way you will see when you will modify default irrlicht materials by adding own shaders to them, we do disable irrlicht ligthing |
| 21:50 |
RealBadAngel |
also, you will notice that every surface is lit, no matter if something is between |
| 21:54 |
RealBadAngel |
Re: Build A World - new massive game, using Irrlicht |
| 21:54 |
RealBadAngel |
Postby sodandk » Fri May 02, 2014 6:11 pm |
| 21:54 |
RealBadAngel |
Hi. The look of the game is a mix of different things. This is my first 3D game, so at start I thought that each light would just be a hardware light and shadows would automatically be shown... but got wiser :-) |
| 21:54 |
RealBadAngel |
First of all, hardware light just goes through everything and shadows are not automatic, and neither is AO (Ambient occulusion) , and the game needed thousands of lights... in case of for example a volcano, each blocks is a light block and there are maybe 50.000 lights ... |
| 21:54 |
RealBadAngel |
So, instead this is how its done: |
| 21:54 |
RealBadAngel |
Each block and the sun can emit light, and a specialized software floodfill fills around with light and shadow and light are applied to the vertices on the go, and also AO is computed on the fly, so basically most of the light is made by software (maybe we move that code to the GPU with OpenCL). DevSH (user here on Irr) has made a number of Shaders for water, reflections, fire, smoke etc. For the player's headlamp and the rotator lights we d |
| 21:54 |
RealBadAngel |
o use hardware lights... |
| 21:55 |
RealBadAngel |
This is a fast rough simple description of how light is done in BAW. It might change radically later on, but the real challenge is to have 10.000 lights or 100.000 lights in a scene, and still have it running in realtime. |
| 22:02 |
|
Taoki joined #minetest-dev |
| 22:25 |
|
troller joined #minetest-dev |
| 22:46 |
|
Taoki joined #minetest-dev |
| 22:51 |
Calinou |
https://github.com/minetest/minetest.github.io/issues/57 |
| 22:51 |
Calinou |
:/ |
| 23:06 |
celeron55 |
eh what |
| 23:06 |
celeron55 |
how did that guy find his way there without finding my email address or something first |
| 23:06 |
celeron55 |
that's literally the worst place to post that question |
| 23:13 |
|
Taoki joined #minetest-dev |
| 23:49 |
|
red-001 joined #minetest-dev |
