IRC logs for #openttd on OFTC at 2026-07-31
⏴ go to previous day
00:07:44 *** Flygon has quit IRC (Read error: Connection reset by peer)
00:20:15 *** Rubidium has quit IRC (Ping timeout: 480 seconds)
00:23:23 *** Rubidium has joined #openttd
00:31:15 <brickblock19280> _glx_: Not without my pr iirc
00:31:41 <brickblock19280> At least from the newgrf perspective
00:58:11 *** Rubidium has quit IRC (Ping timeout: 480 seconds)
01:01:57 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
01:01:59 *** WormnestAndroid has joined #openttd
01:02:00 *** WormnestAndroid has quit IRC (Remote host closed the connection)
01:02:16 *** WormnestAndroid has joined #openttd
01:28:39 *** Wormnest has quit IRC (Quit: Leaving)
01:57:07 *** andythenorth_ has quit IRC (Quit: Connection closed for inactivity)
02:26:15 *** Rubidium has joined #openttd
03:16:15 *** Rubidium has quit IRC (Ping timeout: 480 seconds)
03:24:20 *** Rubidium has joined #openttd
03:25:51 *** Philemon has joined #openttd
03:29:15 *** Phileman has quit IRC (Ping timeout: 480 seconds)
05:02:22 *** ChanServ sets mode: +v tokai
05:09:25 *** tokai|noir has quit IRC (Ping timeout: 480 seconds)
06:37:56 *** cu-kai has quit IRC (Quit: Leaving for a bit.)
06:54:07 <peter1138> Hmm, can you have recolour sprites in an action 1... I'm guessing not.
07:03:47 *** wallabra has joined #openttd
07:22:19 <DorpsGek> - Update: Translations from eints (by translators)
07:42:48 <peter1138> Tree clusters generated on start up using a "primary" tree and randomly picking other trees to mix with it, with a 25% chance of using the same tree.
07:43:33 <peter1138> No attempt to make clusters "compatible" with the original data, so trees will be different when loading an old save.
07:45:05 <peter1138> Nothing in place to avoid mixing types of trees yet, only based on climate & tropic zone.
07:46:29 <peter1138> Could perhaps rearrange the tree data so that the "primary" tree order matches better.
07:47:45 *** kuayueren has quit IRC (Ping timeout: 480 seconds)
07:48:07 <peter1138> Matching those up would do most of the work.
08:05:53 <reldred> trees trees trees trees
08:20:19 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
08:20:21 *** WormnestAndroid has joined #openttd
08:20:26 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
08:20:27 *** WormnestAndroid has joined #openttd
08:20:30 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
08:20:31 *** WormnestAndroid has joined #openttd
08:39:46 <peter1138> Hmm, can NML crop images for me? I assumed it did.,
08:40:16 <peter1138> Apparently it should be the default. Hmm.
08:47:51 <peter1138> Oh, I have to pass -c to nmlc.
08:49:05 <peter1138> Seems for building sprites it'll automatically crop without -c, but for direct graphics it won't? Hmm.
09:02:54 <peter1138> Hmm, the intro game's game generation random seed is wiped :o
09:08:50 <peter1138> If we're auto-generating layouts, I can do variations by height as well.
09:56:51 *** Alkel_U3 has quit IRC (Ping timeout: 480 seconds)
09:59:50 *** andythenorth_ has joined #openttd
10:00:21 <andythenorth_> Trees seem like a good application for pre-generated random maps
10:01:10 <andythenorth_> Tables of different entropy patterns which could be selected by lead tree type, or something about clusters of tiles
10:01:27 <andythenorth_> The grf can do the entropy tables, openttd doesn’t need to
10:02:08 <peter1138> That's the opposite of what _zephyris wanted :)
10:34:56 <peter1138> Hmm, how best to do compatibility...
10:40:54 <peter1138> I await _zephyris' spec :)
11:05:25 *** efemert1245 has joined #openttd
11:11:28 *** Smedles has joined #openttd
11:15:15 *** efemert1245 has quit IRC (Quit: Leaving)
11:16:57 *** efemert124 has joined #openttd
11:17:03 <_zephyris> Compatibility with what? Other NewGRFs?
11:17:40 <peter1138> Tree compatibility.
11:18:49 <peter1138> Like vanilla temperate trees tend (always?) to be split into e.g. deciduous and evergreen (of a sort)
11:19:44 <peter1138> That's a big part of clumping.
11:37:43 <_zephyris> I was planning to look at the tables and to try to infer a couple of classes which would give similar behaviour
11:38:19 <_zephyris> But could always keep the old tables for when no tree NewGRF is loaded
11:39:52 <_zephyris> I'm imagining trees having classes (badges?), and have a few lists of them - prefer clustering with these, rarely cluster with these, never cluster with these
11:40:12 <peter1138> It'll be needed with NewGRFs so might as well use it for non-NewGRF.
11:40:42 <peter1138> The built-in tree data will need it to be able to cooperate.
11:42:24 <peter1138> As I mentioned earlier the clusters are built as 1 per "primary" tree type. Some of the vanilla trees are not primary, so those trees can only appear alongside the primary trees in the cluster, never alone.
11:43:28 <peter1138> So we might have probabilities or something for the rest.
11:43:39 <peter1138> Sorry, for the other trees that might appear in a cluster.
11:57:17 *** jinks has quit IRC (Remote host closed the connection)
12:02:53 <peter1138> Maybe ForestClasses, like CargoClasses ;-)
12:04:11 <peter1138> As this version is pre-computing the layouts on game start up, that limits the amount of flexibility NewGRFs have as to tree positioning and things.
13:04:19 <_zephyris> Could set probabilities per tree ID, probabilities per tree class, or just set which tree classes are low, high or disallowed probabilities
13:04:34 <_zephyris> But a tree class table seems useful
13:06:53 <_zephyris> DECI deciduous, CONF conifer, SCRB scrubland, RAIN rainforest, CACT cactus, etc.
13:09:21 <_zephyris> I should look at the vanilla table to try to get an idea of some classes
13:24:08 <peter1138> Looking at the rearranged table in my branch might make that easier.
13:38:17 <rito12_13> _zephyris: Hurray, more labels 🥳
13:41:55 *** efemert124 has quit IRC (Quit: Leaving)
13:54:23 *** Wormnest has joined #openttd
13:58:32 *** MinchinWeb[m] has quit IRC (Ping timeout: 480 seconds)
13:58:47 *** MinchinWeb[m] has joined #openttd
14:07:52 *** andythenorth_ has quit IRC (Quit: Connection closed for inactivity)
14:13:39 <_zephyris> Bool property for whether it can be the primary tree or not
14:15:02 <peter1138> Yes. Well it's a general flags property.
14:15:52 <peter1138> The flag is "MixOnly", so that, so by default trees appear unless you tell it not to. I assume that'll be the default and MixOnly is only really needed for vanilla trees.
14:16:25 <peter1138> Another flag is "Single" which means the tile layout only ever uses the primary tree, nothing else.
14:16:32 <peter1138> This is used for the vanilla Toyland "trees"
14:16:54 <peter1138> And the other is "RandomColour" which fills in the palett remaps with the company colour remaps, also used only for Toyland "trees".
14:16:59 <peter1138> I'm sure there'll be more :-)
14:24:21 <peter1138> Palettes is why I was asking about action 1 palettes earlier :)
14:25:00 <peter1138> Most things use a property or callback I guess, along with GRM to reserve sprite slots for the palettes.
14:52:37 <_zephyris> Yeah, I can't think of any action0 which sets a specific recolour sprite
14:52:49 <_zephyris> Always some kind of action2
14:55:03 <peter1138> Hmm, GUI growth stage?
14:56:10 <_zephyris> What do you mean? Selecting which growth stage to build?
14:56:19 <peter1138> The old GUI had a separate hardcoded list of sprites that didn't necessarily use the same growth stage.
14:56:26 <peter1138> No, just for display the sprite in the UI.
14:56:35 <_zephyris> Oh, never noticed that
14:56:55 <peter1138> You wouldn't, it's hardcoded magic numbers that are SpriteIDs.
15:02:18 *** MinchinWeb[m] has quit IRC (Read error: Connection reset by peer)
15:03:15 *** MinchinWeb[m] has joined #openttd
15:05:04 <peter1138> It doesn't really matter, we can get used to it :)
15:09:24 <_zephyris> I'm returning to civilisation, so I'll have access to a computer soon to poke tables and make test sprites
15:15:04 <peter1138> Well, that bit is easier now.
15:15:09 <peter1138> item (FEAT_TREES) { property { substitute: 0; climates_available: bitmask(CLIMATE_TEMPERATE); } graphics { default: tree_00; } }
15:15:25 <peter1138> Just a regular spriteset now, no sprite layouts.
15:17:34 <peter1138> OpenGFX 1 trees don't stand out very well from original or OpenGFX 2 terrain.
15:32:34 <_zephyris> Simple code for the simple task of adding one tree, nice and tidy...
15:32:53 <_zephyris> "feels" right in that sense.
15:38:46 *** andythenorth_ has joined #openttd
15:59:29 <peter1138> I'm okay with using class labels for a tree UI, but not sure about compatibility rules.
16:09:07 <andythenorth_> _more tail light yak shaving_
16:42:55 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
16:42:57 *** WormnestAndroid has joined #openttd
18:18:22 *** Wormnest has quit IRC (Ping timeout: 480 seconds)
18:22:32 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
18:26:22 *** WormnestAndroid has joined #openttd
18:39:52 *** gelignite has joined #openttd
19:12:47 <will_marshall_> sign of very productive day at work when you take a good break, go excercise, get back to your workstation and realise you had like 300 tabs open that you can close all of
19:12:58 <andythenorth_> were they all agents writing ur code
19:13:15 <will_marshall_> Agents can't write Train Simulators they are too shit.
19:14:27 <andythenorth_> but how else will we sell all the compute?
19:14:34 <andythenorth_> _me writes some Horse_
19:14:55 <will_marshall_> What is a junior engineer if not a kind of agentic intelligence that occasionally you see in person and get shitfaced with?
19:15:47 <will_marshall_> Actually these days junior engineers are all very straight laced and serious and hardly any of them drink, which is probably a sign that the engineering culture that I graduated into is on its way out.
19:16:08 <will_marshall_> My first day of work my new boss took the whole engineering team out to lunch and many people had multiple pints.
19:17:46 <andythenorth_> it was different times
19:27:40 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
20:08:05 <_zephyris> 3 groups for temperate, 1 for arctic, 3 (rainforest, cactus, normal) for tropic
20:08:44 <_zephyris> There's a lot of consistent structure in the current pseudo-random table...
20:09:03 <peter1138> Are they also like NewGRFClasses?
20:09:23 <peter1138> i.e. NewGRFs can define their own.
20:11:30 <_zephyris> Looking at the current table, you could do a decent job copying the temperate tree behaviour by having three classes.
20:13:02 <_zephyris> TEM1, TEM2, TEM3. TEM1 trees often have a second/third/fourth tree from TEM1, occasionally from TEM2. TEM2 trees only have further trees from TEM2. TEM3 trees often have further trees from TEM3, occasionally from TEM2.
20:14:38 <_zephyris> (I realise I'm not quite answering your question)
20:15:20 <_zephyris> I think it'd be powerful for NewGRFs to be able to define new classes, which could then coexist with other grfs if they happen to share them
20:15:37 <_zephyris> eg. defining scrubland tree class, shrub class, etc.
20:26:39 <peter1138> Not sure where to define compatibility with this scheme.
20:27:42 <peter1138> At the tree level means you get to define which classes to use for the 2nd/3rd/4th trees individually.
20:28:30 <peter1138> And also s/get to/have to/
20:33:01 <andythenorth_> _reconsiders FIRS forest industries_
20:52:44 <_zephyris> I'm imagining that each tree has at least one class (eg. TMP1, DECD. Temperate group 1, which the first 4 vanilla temperate trees are in, and NewGRF-defined deciduous class).
20:52:44 <_zephyris> Then, each 'primary' tree has a list of classes and frequencies used to populate the 2nd-4th trees for its tile layouts. There would need to be a special SELF class (another copy of the same tree).
20:52:44 <_zephyris> This might look like SELF 4, TMP1 2, DECD 2.
20:58:28 <_zephyris> Imagine 3 vanilla trees in TEM1 are available (1 disabled), the NewGRF defined 3 in DECD and an additional NewGRF defined a further 1 in DECD.
20:58:28 <_zephyris> For each subsequent tree in the tile layout, pick a random int 0 to 7 (4 SELF + 2 TEM1 + 2 DECD). If it's 0-3, add a another copy of the same tree. If it's 4-5, randomly pick one of the 3 available TEM1 trees. If it's 6-7 pick one of the 4 available DECD trees (from either NewGRF)
21:00:22 <_zephyris> Does that make sense? I could write a proper spec to try to be clearer.
21:06:03 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
21:07:17 <andythenorth_> makes sense to me, although I'd probably just generate a set of random entropy maps into nml switches, then use the primary tree to select a switch to get the entropy from
21:08:24 <andythenorth_> what you've described above is almost exactly how first version of Horse randomised wagons worked, where each wagon in the consist was randomly selected using the previous vehicle's random bits
21:08:55 <andythenorth_> and if the bits were 0-4 or so, the same wagon type was used, and higher than that was a random choice
21:09:05 *** WormnestAndroid has joined #openttd
21:09:33 <_zephyris> How does that work across NewGRFs?
21:10:14 <andythenorth_> but the random effect was quite nice
21:11:09 <andythenorth_> the current version just uses the engine bits to pick a pre-generated random map of length 64, then each wagon is just from the map
21:11:10 <_zephyris> Key aim with this strategy is to allow better mix and match of tree NewGRFs
21:11:29 <_zephyris> Which may be a fools errand, but good to try
21:11:36 <_zephyris> Yeah, horse random is nice
21:11:40 *** Wormnest has joined #openttd
21:12:07 <andythenorth_> I assume if cross-grf support is needed, then OpenTTD is generating the entropy?
21:12:12 <andythenorth_> I didn't look at the PRs :P
21:14:11 <_zephyris> OpenTTD would generate the tile layouts based on which newgrfs are loaded, prng
21:15:11 <_zephyris> Remember it's not linear layout of randomised items, so the eye's not so sensitive to odd patterns
21:15:48 <_zephyris> This is different from PeterN's first version, which had NewGRF-defined trees on each tile
21:16:02 <andythenorth_> repeating patterns are ok in game world anyway
21:16:18 <andythenorth_> Horse eventually repeats, if enough trains are cloned :P
21:18:56 <_zephyris> Clone all the trains!
21:19:24 <_zephyris> Hmm. Overall tree frequency could also be scaled with number of tile layouts generated with that tree as the primary.
21:19:41 <_zephyris> All vanilla primary trees get 4 layouts.
21:44:17 <peter1138> _zephyris, I'm still generation sub-layouts at the moment.
22:30:42 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:49:24 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
23:58:36 *** andythenorth_ has quit IRC (Quit: Connection closed for inactivity)
continue to next day ⏵
