Time Nick Message 03:01 cheapie So something I've been wanting to look into for a while now in celevator - the jittering/shaking of the car from the point of view of the player when it's moving, even when the speed isn't changing. 03:01 cheapie I threw together a simple "flying dirt node" mod as a test, it does the same thing: https://gist.github.com/cheapie/353b7e585c448ae04ca6b0e94eff570e 03:01 cheapie Video: https://cheapiesystems.com/media/2025-03-29%2021-57-35.webm 03:01 cheapie Am I doing something wrong (this is quite probable) or is this a bug that I can't find/has never been reported? 03:15 swee 404 03:24 cheapie swee: Which one? (both work for me) 03:47 swee cheapie, the video 03:48 MTDiscord works for me 03:48 swee might be HexChat messing up the URL anyways :p 03:48 MTDiscord are you using some esoteric browser? 03:48 swee No, I'm using Chromium 03:49 MTDiscord interesting, could always just wget get the file i suppose if you really cared 03:49 swee oh, HexChat replaced the %20 with %2520 03:55 cheapie Yeah, HexChat tries to be "helpful" like that, it's annoying 03:56 swee ikr 04:07 whosit cheapie: try this PR, it seems to be smooth for me: https://github.com/luanti-org/luanti/pull/15722 04:10 cheapie That didn't help any with celevator, I'll try it with the minimal one later 04:11 whosit meh, sorry 04:12 * cheapie looks over at another monitor with a multiplayer game open, sees lava in the middle of a donut shop, a few nodes on fire, and the fire alarm going SCREEEEE SCREEEEE SCREEEEE in the background 04:27 whosit I guess it's just network jitter? 04:37 cheapie I wouldn't expect network jitter to cause an entity to show up at a different position than the player if the client knows (or should know?) that it's attached to it 04:37 cheapie Higher RTT does seem to make it worse though, and the odd part is the player *does* move smoothly and it's just the entity they're attached to that doesn't 04:38 cheapie That PR did seem to revolve the minimal case but I haven't tested that in multiplayer, seems to maybe be a /little/ smoother with celevator in multiplayer with it, still not great, but the server doesn't have the PR (no idea if that matters on this one) 04:40 cheapie It probably doesn't help that celevator attaches the player to an invisible entity that is then attached to another (visible) one, since it seems like everything involving player attachments is designed to be as much of a pain as possible and this made it a lot easier 04:47 whosit it's pretty weird how I can sometimes get the perfectly smooth ride for quite a while, then teleport back, get new block and it will be jittery... 04:51 whosit maybe somehow you can end up with attachment being processed in different orders or something, I have no idea :p 04:51 cheapie Maybe, I've noticed with two players on one elevator sometimes it's smooth for one of them 04:57 whosit still just retrying and watching: seems like if it starts perfectly smooth, it keeps being smooth, no matter the connection; if it's jittery from the start, then it keep jittering 04:58 whosit (trying with a PC sever/client connected over wifi) 04:59 cheapie Yeah, that's about what I've usually seen too 05:00 cheapie Back when celevator just did a single attachment (and not the double one) I did notice that usually it would be smooth down but not up, with the double one it jitters both ways 05:01 whosit even pulling ethernet cable didn't disrupt the smooth ride :) 05:02 whosit anyway, sorry, can't help much - don't know enough about internals... 05:04 whosit can I try out the celevator anywhere, just out of curiosity? 05:07 cheapie Yes, mod is here: https://content.luanti.org/packages/cheapie/celevator/ or daconcepts.com:37000 has a lot of them set up 05:09 whosit thanks :D I checked the mod page out, but seems like requires some set up to get going X) 05:09 cheapie It does, yeah 05:09 cheapie Not of the "complex installation tag" sort, but more of the "the manual has 66 pages for a reason" sort 05:10 whosit yeah :) 06:09 [MatrxMT] congratulations to VoxeLibre for overtaking Minetest Game in total downloads from ContentDB 06:10 [MatrxMT] it's a better game than Minetest Game, but it does say something that it was and mostly remains a Minecraft clone 06:10 MTDiscord I mean... Luanti is built for Minecraft clones 😉 06:11 [MatrxMT] ultimately, at least in the loose sense 06:11 [MatrxMT] thankfully after the top 3 the other games are quite different from the usual Minecraft-like fare 06:13 [MatrxMT] "If I had asked people what they wanted, they would have said free Minecraft" - Henry Luanti 06:28 MTDiscord Would be cool if we stopped normalizing the idea that "luanti is built for minecraft clones" 06:31 MTDiscord Especially when more than half the games on CDB could be easily classified as not Minecraft clones, and the remaining games could make decent arguments against it as well 06:32 [MatrxMT] as I said, "in the loose sense" - that is, a fixed grid of voxel nodes and entities, nothing more 06:33 [MatrxMT] and for performance reasons, most games should not use too many mesh nodes either 06:35 MTDiscord Thats only if its a sandbox game. If its any other kind of game, by all means, use mesh nodes. Or mesh entities. 06:35 [MatrxMT] if you want to lean into retro, sprites are an option 06:35 MTDiscord That too 06:35 MTDiscord There is a lot more you can do than just sandbox voxel games 06:37 [MatrxMT] as much as I'm behind people doing other cool things, it's definitely the easiest choice if you're going to use Luanti 06:38 MTDiscord Which is why we should continue making Luanti even easier to use for making games in general 09:16 MTDiscord Yeah I guess it depends on what you count as a clone. In my book, the engine is built around the concept of a 3D, meter-based grid of blocks that players can break and place. Sounds familiar to me! I haven't played many CDB games yet but what I see looks like Minecraft or modded Minecraft to me. Absolutely nothing wrong with that! Luanti as a general game engine is a strange idea, it was certainly built around 3D break-and-place 09:16 MTDiscord mechanics, and that seems like an excellent niche as I haven't found any other engine that handles this well (and I looked a LOT!) 09:23 [MatrxMT] minor correction: Luanti doesn't have to have a 1 node = 1 metre scale, it can be game-defined what node size, player size, stepheight, gravity and so on that it has 09:23 [MatrxMT] *okay not both node and player size, just player size which decreases the relative node size 09:23 Krock numbers without units are like sandals without socks. 09:23 Krock wait no 09:24 [MatrxMT] a ratio is a dimensionless quantity, including scale 09:38 sfan5 and for performance reasons, most games should not use too many mesh nodes either 09:38 sfan5 to some degree that should be solved 09:38 sfan5 generally if you find scenes whose performance is challenging for Luanti I'm interest in investigating them 09:39 [MatrxMT] the mesh cache constructs with mesh nodes to some extent then? 09:42 sfan5 there is a pass that merges identical stuff over the entire scene, now 09:42 cheapie sfan5: I should've had you see that house someone on VE-C built out of meshnodes (facade centerstone) before I got rid of it :P 09:42 sfan5 still mesh nodes won't be able to optimize out neighboring faces 09:43 sfan5 the meshgen could potentially attempt to detect this but would need some research on algorithms first 09:43 [MatrxMT] research? heck, it probably could be a new paper 09:43 cheapie It was killing my framerate (but not as much as these things used to), I ended up just replacing it with solid nodes (baked clay) of the same color 09:45 [MatrxMT] almost all of the rendering improvement techniques I hear about are for pure voxel scenes, or the pure voxel part of the scene only 09:45 sfan5 I think vertex merging (or rather vertex visibility?) is a well-researched subject 09:48 [MatrxMT] oh true, if it's like what blender can do with vertex merging than that is indeed well-established. Blender isn't bound by realtime constraints though 09:49 [MatrxMT] but talking about different nodes is potentially talking about merging multiple materials and so on (though usually not). Testing for that has to be made quick. Anyway I'm out of my depth already 09:50 cheapie I have basically no clue what I'm talking about here, but part of me wonders if some sort of "this mapblock is complex and not changing frequently, spin up a separate thread to try to optimize it" thing is practical 09:53 sfan5 I guess we should actually be working on the texture atlas solution. on most GPUs having a few more or less vertices doesn't really seem to affect performance 09:54 sfan5 but additional drawcalls seem to be a problem 09:54 * cheapie hears "texture atlas", suddenly remembers the early 0.4 days 09:55 [MatrxMT] lt4k uses a texture atlas PNG 10:16 [MatrxMT] I wonder just how small MTG could be made with a texture atlas, asset compression and code compression 10:17 [MatrxMT] the biggest thing would almost certainly end up being sounds 10:54 whosit luanti does not handle well collisions with entities of sizes more than a couple of nodes, so if "player" is about human-sized, then node is about meter sized too... 13:28 MTDiscord I had to turn off collisions on my 50-node-wide entity in subnodeica because it just didn't work except in a few spots 13:29 [MatrxMT] subnodeica? 13:32 MTDiscord an unfinished game of mine, basically a partial remake of subnautica in luanti 13:32 rubenwardy the idea list turns into the unfinished project list. How relatable 13:40 rubenwardy congratulations to VoxeLibre for overtaking Minetest Game in total downloads from ContentDB 13:40 rubenwardy VL has had more total downloads than MTG on CDB for years as MTG was shipped with Luanti, so didn't get many CDB installs. So the total downloads is a fairly corrupted measure. The thing to look at is downloads in the last 30 days. VL is 13945 and MTG is 19966 13:41 rubenwardy VL ranks higher on CDB due to reviews, not downloads 13:42 rubenwardy maybe I should put downloads in the last 30 days on the package page 13:43 [MatrxMT] if you wanted to lean that way you could have trending sections rather than just top. Never was that into virality personally 13:45 [MatrxMT] or to really gamify it, a plot of daily downloads of most trending packages, so people can try to push their numbers up 13:45 rubenwardy "top packages" is a heuristic that estimates downloads in the last 30 days and adds average review scores. Total downloads isn't taken into account 13:45 sfan5 do you have downloads in last year 13:45 [MatrxMT] but you're not making money from that unlike a tech startup who need to juice the analytics 13:45 rubenwardy https://content.luanti.org/packages/Wuzzy/mineclone2/stats/?start=2024-01-01&end=2024-12-31 13:45 rubenwardy filter by last year, it shows a sum 13:46 rubenwardy last year - 245,725 for VL, 530,760 for MTG 13:46 sfan5 https://content.luanti.org/packages/sfan5/minetest_classic/stats/?start=2024-01-01&end=2024-12-31 13:46 sfan5 stonks 13:47 rubenwardy there are also CSV downloads and a rest API for those interested 13:50 rubenwardy another thing is that the "unique download" thing depends on IP addresses in a 14 day period. So if you release an update every 14 days, you'll increase your total downloads by quite a lot 13:50 rubenwardy which if the updates actually contain changes is not a bad thing necessarily, promotes actively developed packages 13:52 [MatrxMT] gotcha, time to organise all my packages to run in 14 day sprint development cycles :D 15:52 MTDiscord there are lots of ways to game the cdb ranking if you really watch it 15:52 MTDiscord just dont abuse it 15:53 ireallyhateirc what? you mean Subway Miner isn't the best and most complete game in the galaxy? 15:56 MTDiscord that game isnt really gaming much of anything, its barely updated 16:13 ireallyhateirc one thing about the ranking is that some games (looking at Subway Miner) got positive reviews as jam games due to people having lower expectations than they would for ordinary games 16:13 ireallyhateirc well we also have games that surpassed expectations as jam games and deserve their high place 16:15 MinetestBot 02[git] 04cx384 -> 03luanti-org/luanti: Improve hand override documentation 13af1ffce https://github.com/luanti-org/luanti/commit/af1ffce084ff9a37e46649ab693e9f924159d606 (152025-03-30T16:15:11Z) 16:15 MinetestBot 02[git] 04cx384 -> 03luanti-org/luanti: Document special items 13882f132 https://github.com/luanti-org/luanti/commit/882f132062498b6019e8177fca1d53df875a2e72 (152025-03-30T16:15:11Z) 16:15 MinetestBot 02[git] 04SmallJoker -> 03luanti-org/luanti: Formspec: fix clamped scroll offset of scroll_containers larger than … 13309c0a0 https://github.com/luanti-org/luanti/commit/309c0a0cb6420d13495a003d07b49d126eda80bf (152025-03-30T16:15:38Z) 16:18 MinetestBot 02[git] 04y5nw -> 03luanti-org/luanti: Refactor input handler (#15933) 134cd2273 https://github.com/luanti-org/luanti/commit/4cd22733499b3ad19424e28ba6399a36ef592927 (152025-03-30T16:16:20Z) 16:18 MinetestBot 02[git] 04cx384 -> 03luanti-org/luanti: Document server texture pack in texture_packs.md (#15951) 1341d43e8 https://github.com/luanti-org/luanti/commit/41d43e8d95a4dd1b983b877b51b04e5ed7983198 (152025-03-30T16:16:34Z) 16:18 MinetestBot 02[git] 04wrrrzr -> 03luanti-org/luanti: Refactor createShadowRenderer 134bc366b https://github.com/luanti-org/luanti/commit/4bc366b984aece91fa95a0315586eaa1700994f7 (152025-03-30T16:16:45Z) 16:18 MinetestBot 02[git] 04wrrrzr -> 03luanti-org/luanti: Set StartupWMClass in desktop file 13d7edf0b https://github.com/luanti-org/luanti/commit/d7edf0b229874e8f9f2f5ca3e03df5fe73a53a85 (152025-03-30T16:17:10Z) 16:18 MinetestBot 02[git] (9 newer commits not shown) 21:55 cheapie New mod idea (as in shitpost, not for actual use) - mesecons pistons but they extend smoothly/slowly and make this noise when they do: https://www.youtube.com/watch?v=h8c2N0axAOI&t=6 21:55 cheapie Or possibly this noise, one of the two anyway: https://www.youtube.com/watch?v=Rrn3a_-Y-5w&t=69 21:56 ireallyhateirc reminds me that I need to model a faithful replica of yandev toothbrush 21:57 ireallyhateirc with a spawner