Minetest logo

IRC log for #minetest-dev, 2015-06-05

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:32 cib0 joined #minetest-dev
01:44 kaeza_ joined #minetest-dev
01:45 carloslfu joined #minetest-dev
01:50 carloslfu left #minetest-dev
02:00 OldCoder joined #minetest-dev
02:23 zat joined #minetest-dev
02:35 kaeza joined #minetest-dev
02:40 kaeza_ joined #minetest-dev
02:44 jordan4ibanez joined #minetest-dev
03:39 leat joined #minetest-dev
03:53 leat joined #minetest-dev
04:46 VargaD joined #minetest-dev
05:07 paramat joined #minetest-dev
05:14 paramat lighting bug at y = 63 https://github.com/minetest/minetest/issues/2759 which is of course 16 nodes above y = 47 =/
05:14 leat joined #minetest-dev
05:15 paramat http://i.imgur.com/hI0kMSY.png experimenting with a new type of biome blend https://github.com/paramat/minetest/tree/noiseblend
05:17 kaeza_ joined #minetest-dev
05:22 hmmmm much nicer biome blending
05:22 hmmmm excellent in fact
05:22 hmmmm btw what mapgen is that?  it looks really nice imo
05:23 diemartin joined #minetest-dev
05:24 paramat mgv7! hehe
05:24 Wayward_Tab I like it as well
05:24 hmmmm you're kidding
05:24 hmmmm oh yeah i can see it now that i look at the bottom where the river is
05:24 paramat with some crazy surface nodes to show up the borders
05:24 hmmmm wow mgv7 actually turned out to be really nice
05:24 hmmmm i'm surprised
05:25 paramat still experimenting and tuning the biome blend
05:25 hmmmm when i was working on it i always thought it looked like dogshit
05:25 paramat heh
05:25 paramat the new parameters are in
05:25 hmmmm yeah your tweaks helped a lot
05:25 hmmmm it's good to know that it wasn't fundamentally flawed, i guess
05:25 paramat i also doubled the scale of rivers, they really travel now and are quite epic
05:26 paramat er spread
05:49 Hunterz joined #minetest-dev
06:06 paramat left #minetest-dev
06:06 leat joined #minetest-dev
06:37 blaaaaargh joined #minetest-dev
07:07 Darcidride joined #minetest-dev
07:16 Calinou joined #minetest-dev
07:26 OldCoder joined #minetest-dev
07:27 Amaz joined #minetest-dev
07:30 jin_xi joined #minetest-dev
07:31 selat joined #minetest-dev
07:45 OldCoder joined #minetest-dev
07:55 leat joined #minetest-dev
07:56 cib0 joined #minetest-dev
08:00 Yepoleb_ joined #minetest-dev
08:06 leat joined #minetest-dev
08:17 OldCoder joined #minetest-dev
08:27 kaeza joined #minetest-dev
08:48 OldCoder joined #minetest-dev
08:53 chchjesus joined #minetest-dev
09:16 OldCoder joined #minetest-dev
09:30 jin_xi joined #minetest-dev
09:35 OldCoder joined #minetest-dev
10:48 OldCoder joined #minetest-dev
11:48 cib0 joined #minetest-dev
11:48 OldCoder joined #minetest-dev
12:27 LittleJoe joined #minetest-dev
12:44 leat joined #minetest-dev
12:47 OldCoder joined #minetest-dev
13:08 AnotherBrick joined #minetest-dev
13:14 Player_2 joined #minetest-dev
13:29 nore joined #minetest-dev
13:30 Gethiox joined #minetest-dev
13:44 AnotherBrick joined #minetest-dev
13:53 cib0 joined #minetest-dev
14:08 proller joined #minetest-dev
14:15 leat joined #minetest-dev
14:19 est31 joined #minetest-dev
14:25 proller joined #minetest-dev
15:04 hmmmm joined #minetest-dev
15:15 leat joined #minetest-dev
15:36 leat1 joined #minetest-dev
15:46 leat1 joined #minetest-dev
16:18 Hunterz joined #minetest-dev
16:28 OldCoder joined #minetest-dev
16:38 est31 joined #minetest-dev
16:51 Player_2 joined #minetest-dev
16:53 ShadowNinja I'll post a news topic about mod security.  'k?
16:53 VanessaE good idea.
16:53 ShadowNinja Something like so: http://pastebin.ubuntu.com/11591735/
16:57 VanessaE ShadowNinja: you should mention something in there about mods in the "store" not always being up-to-date, that users should defer to the forum posts and their respective repositories.
16:58 VanessaE also for player_textures, get my fork out of dreambuilder.  it's yours with a few minor patches.
16:59 VanessaE your patch won't work
17:02 VanessaE line comment added.  I think that's the only change needed.
17:03 ShadowBot joined #minetest-dev
17:03 ShadowBot joined #minetest-dev
17:11 proller joined #minetest-dev
17:17 leat1 joined #minetest-dev
17:26 Krock joined #minetest-dev
17:40 Niebieski joined #minetest-dev
17:41 Taoki joined #minetest-dev
17:43 LittleJoe joined #minetest-dev
17:56 cib0 joined #minetest-dev
18:00 Anchakor joined #minetest-dev
18:05 hmmmm joined #minetest-dev
18:12 leat1 joined #minetest-dev
18:13 proller joined #minetest-dev
18:16 RealBadAngel hi guys
18:16 est31 hi RealBadAngel, hmmmm
18:16 RealBadAngel any comments on: http://irc.minetest.ru/minetest-dev/2015-06-04#i_4278019
18:17 est31 I like it
18:18 hmmmm i think it's fine the way it was
18:18 VanessaE hey RBA.
18:18 RealBadAngel and  having 16px default normal textures? https://imgrush.com/w5iCWW1HuC-V.png
18:18 RealBadAngel hmmm, you like it being so dark?
18:18 VanessaE RealBadAngel: the problem with the leaves thing is that it'll get un-done next time someone optimizes the textures
18:19 VanessaE otherwise I agree that they look better your way than with black.
18:19 est31 why does it get undone?
18:19 RealBadAngel VanessaE, honestly i dont understand that optimizing a few hundreds of bytes thing
18:19 VanessaE est31: because image optimizers usually set 100% transparent pixels to black.
18:20 RealBadAngel textures do not have alpha chan, have to be converted to work on them
18:20 VanessaE RealBadAngel: nor do I, in the case of minetest_game anyways.
18:20 RealBadAngel more like pain in the ass than real gain
18:20 VanessaE in a larger project, the differences add up
18:20 RealBadAngel we do count gigabytes nowadays, not bytes for christ sake
18:20 est31 VanessaE, which style of image optimizer does that? I mean then alpha is just simply broken then?
18:21 VanessaE est31: they all do, afaik.
18:21 VanessaE and it's not that the alpha is broken
18:21 RealBadAngel color space is fucked up, its indexed
18:21 VanessaE it's that #xxxxxx00 pixels get set to #00000000
18:21 RealBadAngel so before you can do anything to the image you have to convert it to rgb space
18:21 VanessaE RealBadAngel: yeah, that's a pain in the ass especially when dealing with texture overlays
18:22 RealBadAngel and all that  for saving like 50 bytes on disk
18:22 RealBadAngel while single allocation unit is not less than 4kb
18:22 RealBadAngel sounds like a joke
18:22 kahrl it's also about how long it takes to connect to a server.
18:23 VanessaE RealBadAngel: like I said, it adds up in larger projects.  I think in homedecor it shaved off around 500k the first time I did it
18:23 est31 its bad to design a game only for boxes with > 4GB RAM
18:23 RealBadAngel est31 i do have 4gb and im using 256tp with normal
18:24 RealBadAngel so you have shot god's window with that argument
18:24 est31 hehe
18:24 hmmmm for what it's worth
18:24 hmmmm I designed the optimized perlin noise maps with the explicit speed/space tradeoff in mind
18:24 RealBadAngel we are not making 1kb demo where each byte counts :P
18:25 RealBadAngel in memory those textures are expanded anyways
18:25 RealBadAngel and about sending them, what is the minimal packet size?
18:26 err404 joined #minetest-dev
18:26 troller joined #minetest-dev
18:27 RealBadAngel but wait a sec, as usual starting topic went wild
18:27 RealBadAngel i was asking about something else
18:27 RealBadAngel <RealBadAngel> and  having 16px default normal textures? https://imgrush.com/w5iCWW1HuC-V.png
18:27 RealBadAngel thats 16x16 texture used
18:28 RealBadAngel i can mimic that on the fly, but it will be very costly
18:29 RealBadAngel imho having well hand made normals (such small files!) is better
18:29 sfan5 VanessaE: unknown nodes seem to make minetestmapper veeeeeeery slow (1024x1024: 51,2s instead of 4,9s)
18:30 VanessaE sfan5: ouch
18:30 est31 10x increase...
18:30 kahrl the python or C++ one?
18:30 sfan5 c++
18:31 est31 is the python one still developed?
18:31 sfan5 no
18:33 RealBadAngel <VanessaE> RealBadAngel: like I said, it adds up in larger projects.  I think in homedecor it shaved off around 500k the first time I did it -----> but it didnt save any disk space both for server and the client. it didnt saved any used memory too because of the allocation size. what about sending them over net? idk, but propably the same
18:34 est31 it did save disk space, didnt it?
18:35 est31 500k
18:35 RealBadAngel if allocation unit is 4kb, 1byte file will occupy the same as 4kb
18:35 est31 ah
18:35 RealBadAngel your calculations wont change that
18:36 est31 yes thats right
18:36 VanessaE between optimizing images and reducing the total number of images in the modpack, close to 1MB saved I think.
18:36 VanessaE but yeah, the 4k point stands.
18:36 RealBadAngel your gain is purely imagination
18:36 sfan5 VanessaE: i looks like it's spending a lot of time on inserting unknown nodes into the list
18:37 sfan5 s/i /it /
18:38 sfan5 326,445,300  /build/gcc/src/gcc-5-20150519/libstdc++-v3/libsupc++/dyncast.cc:__dynamic_cast [/usr/lib/libstdc++.so.6.0.21]
18:38 sfan5 298,935,573  ???:__memcmp_sse4_1 [/usr/lib/libc-2.21.so]
18:38 sfan5 270,585,286  ???:std::_Rb_tree<std::string, std::string, std::_Identity<std::string>, std::less<std::string>, std::allocator<std::string> >::_M_get_insert_unique_pos(std::string const&) [/tm
18:38 sfan5 p/minetestmapper/minetestmapper]
18:38 sfan5 no idea how to make that faster
18:39 VanessaE well it's not critical - mine was a corner-case anyways
18:39 VanessaE t'is what happens when you throw up to 16'000 unknown nodes at it ;)
18:41 kahrl sfan5, perhaps use an unordered_map instead
18:41 est31 unordered_map is only available in c++11
18:41 sfan5 VanessaE: -DUSE_CXX11=1 (unordered_map) makes it better though: 15,8s -> 13,0s
18:41 sfan5 damnit kahrl
18:41 kahrl yeah
18:41 sfan5 can you read minds
18:41 sfan5 like
18:41 sfan5 i was about to type that
18:41 kahrl :P
18:41 sfan5 also we already have that
18:42 RealBadAngel in mapper one should check  if (f->tiledef[0].name != "")
18:42 RealBadAngel air, content ignore and unknown wont have the texture
18:43 kahrl I don't think ContentFeatures are available in the mapper
18:43 sfan5 guess what it already does
18:44 VanessaE like I said, it's not really critical.  I just threw a large map and an empty colors.txt at it in hopes of getting a list of all nodes used in the map.  I don't think too many others are gonna try that ;)
18:44 RealBadAngel anyway i think mapper wont be needed in the future
18:44 RealBadAngel minimap does the same realtime
18:44 VanessaE RealBadAngel: sure it will - for websites/forum pages
18:44 sfan5 ooh
18:44 sfan5 way better
18:44 RealBadAngel we can add just option to save the data
18:45 VanessaE a realtime 100-odd diameter map doesn't tell you that there's an interesting structure 1000 nodes away
18:45 sfan5 VanessaE: 1024x1024 51,2s -> 36,3s (-DUSE_CXX11=1)
18:45 est31 yea, but minimap doesnt map the whole world
18:45 est31 just a small part
18:45 RealBadAngel minimap is using mapblock minimaps cache atm
18:45 est31 btw, whats missing for that?
18:45 RealBadAngel it could just save the cache
18:45 VanessaE RealBadAngel: that doesn't help servers.  only clients.
18:46 kahrl you don't want to have to script the client to automatically produce a map of the world when you're a server admin
18:46 RealBadAngel but the code can be applied on the server side too
18:46 RealBadAngel different output client sinde, different servers
18:46 VanessaE sfan5: that's definitely a lot better.  so why isn't C++11 used by default anyway?
18:46 est31 with all the irrlicht deps?
18:47 leat1 joined #minetest-dev
18:47 est31 celeron55 has found the compromise that c++11 can be used when debian 8 gets released
18:47 est31 its released now, so we can use it?
18:48 RealBadAngel doubt that
18:48 RealBadAngel irrlicht is 1.9 by now and we are still with 1.7
18:48 kahrl I thought the decision was to switch to C++11 when Ubuntu 12.04 LTS went out of support
18:49 VanessaE RealBadAngel: 1.9 is still in nightlies.  it isn't "out" yet.
18:49 RealBadAngel i know
18:50 RealBadAngel but still 1.8 were here ages ago
18:50 RealBadAngel i remember arguing bout it like an yr ago
18:50 VanessaE Irrlicht 1.8.1 released
18:50 VanessaE November-20-2013 – 9:44 pm
18:50 sfan5 <RealBadAngel> irrlicht is 1.9 by now and we are still with 1.7
18:51 sfan5 how are we "with" irrlicht 1.7?
18:51 VanessaE sfan5: because the engine still has to be able to run with 1.7?
18:51 RealBadAngel its a bit ridiculous when each next release of engine and a slight change to the mt game is makin a few weeks older builds not to work
18:51 RealBadAngel but still we do care about libs 2 yrs old
18:51 sfan5 VanessaE: is backwards compatibility a bad thing?
18:52 sfan5 why does the engine have to run with 1.7?
18:52 VanessaE sfan5: no, so long as it doesn't prevent being able to use newer features where available.
18:52 RealBadAngel sfan5, such "compability" sounds a bit like monty python themed
18:53 est31 http://irc.minetest.ru/minetest-dev/2015-01-29#i_4135244
18:53 RealBadAngel youre forced to get up to date game
18:53 sfan5 og
18:53 sfan5 oh*
18:53 sfan5 because of ubuntu 12.04?
18:54 est31 (still talking bout c++11)
18:54 est31 not bout irrlicht
18:54 VanessaE um...
18:54 est31 dunno there
18:54 VanessaE I still use 12.04
18:54 RealBadAngel me too
18:55 R2D3_ joined #minetest-dev
18:55 celeron55 everything is at 1.8 by now i'd think
18:56 celeron55 ...oh, except centos 6, which some people still use
18:56 sfan5 even the windows builds use 1.8
18:56 sfan5 (mingw at least)
18:56 celeron55 centos LTS isn't getting updated still in years and won't get c++11
18:56 celeron55 it's one of the worst offenders of being able to update things
18:57 RealBadAngel celeron55, we still have some code from 1.8 ported
18:57 RealBadAngel because we wanted to be compatible with 1.7
18:57 est31 12.04 has irrlicht 1.7 http://packages.ubuntu.com/precise/libirrlicht-dev
18:57 sfan5 ubuntu 12.04 sucks
18:57 RealBadAngel remember when i merged l-system trees?
18:57 sfan5 snappy and/or leveldb are broken in 12.04
18:57 celeron55 current ubuntu LTS is 14.04; nobody has any excuse to use 12.04
18:57 VanessaE vanessa@rainbird:~$ apt-cache search irrlicht
18:57 VanessaE ...
18:57 VanessaE libirrlicht1.8 - High performance realtime 3D engine
18:58 RealBadAngel some matrices math was ported back from irr 1.8 at that time
18:58 est31 http://packages.ubuntu.com/search?keywords=libirrlicht-dev&amp;searchon=names&amp;suite=precise&amp;section=all
18:58 VanessaE looks to me like 1.8 is available in 12.04 by default (pool/universe/i/irrlicht/libirrlicht1.8_1.8.1+dfsg1-1_amd64.deb)
18:58 RealBadAngel and that was like more than 2 yrs ago
18:58 VanessaE est31: wrong search key
18:59 est31 still not here: http://packages.ubuntu.com/search?suite=precise&amp;section=all&amp;arch=any&amp;keywords=irrlicht&amp;searchon=names
18:59 RealBadAngel we are more conservative than pope :P
18:59 kahrl well, that francis guy is quite progressive for a pope :P
19:00 RealBadAngel i havent seen any commits by him yet ;)
19:00 VanessaE est31: idk, but it's not in a third-party repo for me, it's in 'universe'.
19:01 est31 can you run lsb_release -a ?
19:01 VanessaE oh fuck me
19:01 VanessaE I'm on 14.04
19:01 VanessaE :)
19:01 est31 :)
19:01 VanessaE I totally forgot that I'd updated :D
19:03 RealBadAngel celeron55, https://imgrush.com/w5iCWW1HuC-V.png what do you think about havin such default shaders effects?
19:03 VanessaE one can see the layering effect in that screenshot
19:04 VanessaE (look at the edge of the brick that's just above the dark block at the right end of the hotbar HUD)
19:05 celeron55 RealBadAngel: i don't like that
19:05 RealBadAngel celeron55, http://i.imgur.com/qZ6Wp2G.png this neither?
19:06 Krock wow, what an eye candy!
19:06 RealBadAngel VanessaE, that because of parallax and too few iterations, relief mapping doesnt have such flaws
19:07 celeron55 well that really just adds noise without adding information; i like plain blocks
19:07 RealBadAngel ive already implemented relief mapping, effects are way nicer than with parallax
19:08 VanessaE RealBadAngel: how about a screenshot without the "generate normalmaps" /edge-detect feature?  showing JUST the parallax maps
19:08 celeron55 i would be kind of okay with parallax maps IF there was dynamic lighting; but without dynamic lighting there's no point
19:09 celeron55 and IF they were made properly and not just generated from existing textures
19:09 est31 and about unordered_map, what about making a typedef to map if the compiler doesnt support c++11?
19:09 RealBadAngel you can already compare hand made ones and autogenerated
19:10 RealBadAngel https://imgrush.com/w5iCWW1HuC-V.png hand made map
19:10 celeron55 there's beauty in just blocks with textures; it combines a 3d world with traditional pixel art in a way that doesn't usually happen
19:11 RealBadAngel https://imgrush.com/LCf8h8L6otDY.png autogeneration
19:11 sfan5 est31: this is what i do in minetestmapper https://github.com/minetest/minetestmapper/blob/master/TileGenerator.h#L46-54
19:12 est31 hrmm should be done in minetest too
19:12 RealBadAngel celeron55, as you said world is 3d
19:12 proller joined #minetest-dev
19:12 RealBadAngel why objects in it have to be plain flat?
19:13 est31 thats still pixel-like, and also just a setting, doesnt have to be default on
19:13 RealBadAngel https://imgrush.com/hz-gt8FqGPOA.png
19:13 RealBadAngel take a look on HD version of it
19:14 RealBadAngel with dynamic shadows and lightsource (still at (0,900,0)
19:14 RealBadAngel waiting to be moved ;)
19:24 selat joined #minetest-dev
19:29 troller joined #minetest-dev
19:40 VanessaE celeron55: that said, just how "dynamic" do shadows/lighting need to be?  moves with the sun?  reacts to other light sources too?
19:40 celeron55 dunno
19:41 celeron55 but my point is, anything that you do with the current lighting could be just baked into the textures
19:41 celeron55 which is why it is pointless
19:41 VanessaE this is true (to a degree).
19:55 leat1 joined #minetest-dev
19:59 cib0 joined #minetest-dev
20:00 diemartin joined #minetest-dev
20:00 kaeza joined #minetest-dev
21:40 paramat joined #minetest-dev
21:43 paramat sfan5 or anyone, any advice on the best way to add a common group to the water buckets as requested by PilzAdam here https://github.com/minetest/minetest_game/pull/528 ? any comments to add?
21:48 paramat guess i'll try to add a group field to the registration that is backwards compatible
21:51 paramat hmmmm i assume it's okay to remove the deprecated ore generation code https://github.com/minetest/minetest_game/pull/527
22:03 blaise joined #minetest-dev
22:14 est31 joined #minetest-dev
23:02 ShadowNinja ~tell RealBadAngel Your non-fancy-tree-leaves tweak and custom normalmaps look great!
23:02 ShadowBot ShadowNinja: O.K.
23:06 paramat left #minetest-dev
23:10 rickmcfarley joined #minetest-dev
23:26 Player_2 joined #minetest-dev
23:35 cib0 joined #minetest-dev
23:39 jin_xi joined #minetest-dev

| Channels | #minetest-dev index | Today | | Google Search | Plaintext