| Time |
Nick |
Message |
| 00:00 |
erle |
i can also render emojis and hacked in tabs |
| 00:00 |
erle |
a bunch of control characters will show up as placeholder pics though |
| 00:00 |
muurkha |
does Minetest know how to ingest TGAs without writing a Lua thing to read the format? |
| 00:00 |
muurkha |
I mean it's pretty trivial... |
| 00:00 |
erle |
obviously yes |
| 00:00 |
muurkha |
hmm |
| 00:01 |
muurkha |
think about doing the Paeth predictor thing |
| 00:01 |
erle |
not necessary |
| 00:01 |
muurkha |
it might buy you extra crunch |
| 00:01 |
erle |
tga was removed when someone™ did a thing, but it was re-added after it was pointed out that mcl_maps used it for generated textures. |
| 00:02 |
erle |
many games only load tga lol |
| 00:02 |
erle |
it's just simple enough to get it right |
| 00:03 |
erle |
in practice, obv every simple bitmap format would do, but for some reason people who have no idea how these formats look internally always point out that bmp exists and … one look at the wikipedia article is a big NOPE |
| 00:03 |
erle |
i think i implemented a BMP writer once and it was awful (you need padding for each scanline or so) |
| 00:03 |
erle |
anyways, don't get nerdsniped |
| 00:04 |
erle |
i think the predictor thing is not useful |
| 00:04 |
erle |
can you even do it in a single pass? |
| 00:05 |
erle |
then again, my intuitions were wrong and might be wrong again hehe |
| 00:05 |
erle |
muurkha, if you get to anything, tell. but i'll stop at the suggestion to use tga.z and obj.z |
| 00:05 |
erle |
n8n8 |
| 00:07 |
muurkha |
I hacked together a BMP writer in Python one day on a ferryboat |
| 00:07 |
muurkha |
on the way across the river |
| 00:07 |
erle |
yes, it's not that hard, but it's obnoxious |
| 00:07 |
muurkha |
the Paeth predictor is indeed single-pass |
| 00:07 |
erle |
bmp is basically worse than TGA in any case i have seen, so it can go the way of the dodo |
| 00:07 |
muurkha |
it does require you to buffer an entire scan line |
| 00:08 |
erle |
if you can do it fast in lua, i am interested |
| 00:08 |
erle |
because, again, generated textures in minetest mods |
| 00:08 |
muurkha |
probably reasonably fast? |
| 00:09 |
erle |
also the thing must be interoperable ofc |
| 00:09 |
muurkha |
it's mostly explained in https://en.wikipedia.org/wiki/PNG#Filtering |
| 00:09 |
erle |
so are you paeth predicting the entire file and making a custom compressor? |
| 00:09 |
erle |
or inventing a new file format which is based on TGA but incompatible with everything else |
| 00:09 |
muurkha |
well, you can think of it as an image filter |
| 00:09 |
muurkha |
an invertible image filter |
| 00:09 |
erle |
so? |
| 00:10 |
erle |
you make a new file format right? |
| 00:10 |
muurkha |
so you can think of it as taking a TGA as input and producing a TGA as output |
| 00:10 |
muurkha |
of the same size |
| 00:10 |
erle |
yeah i'm not interested |
| 00:10 |
muurkha |
which is usually more compressible |
| 00:10 |
erle |
i get it |
| 00:11 |
erle |
but this is a holographic fracture |
| 00:11 |
muurkha |
I mean Paeth ο zlib is conceptually the same kind of transformation as just zlib, right? |
| 00:11 |
erle |
interpretation of information may differ at system boundaries |
| 00:11 |
erle |
the naive user must not be able to confuse your faul tga with a real one |
| 00:11 |
erle |
faux tga |
| 00:12 |
muurkha |
if you want to put a blob of bytes on the meta property of something in the game, and then use some arbitrary Lua to convert it into a texture |
| 00:12 |
erle |
that sounds bizarre |
| 00:12 |
muurkha |
isn't that what you're doing with zlib? |
| 00:12 |
erle |
i'd of course want the engine to support it in the end |
| 00:13 |
erle |
given the hilarious way [png is implemented (make a temporary file, then read it again), i'm pretty sure a patch for tga.z would be accepted, given it's a much less [redacted] idea |
| 00:13 |
erle |
but i want to *generate* textures in lua |
| 00:14 |
erle |
i mean i am already doing it |
| 00:15 |
erle |
anyway, good night |
| 00:18 |
muurkha |
I feel like you could set byte 3 of the .TGA to, say, 66 instead of 2 |
| 00:18 |
muurkha |
to indicate that it's Paeth-predicted |
| 00:19 |
muurkha |
looking at http://www.paulbourke.net/dataformats/tga/ |
| 00:24 |
erle |
damn, i'm still not a sleep |
| 00:24 |
erle |
look muurkha you are solving a different problem here |
| 00:25 |
erle |
it's no longer about “how much can i wring out of existing dataformats/primitves” |
| 00:25 |
erle |
but ”if i was not limited, how good could it get” |
| 00:26 |
erle |
interesting problem, but the results are impractical to use, since you would have to patch every single program that uses TGA ig |
| 00:26 |
erle |
meanwhile DEFLATE is available everywhere |
| 00:27 |
erle |
i mean a data format lives and dies by if you can actually use it |
| 00:28 |
erle |
which is also probably the reason why we two wrote BMP encoders, some other software wanted to consume that |
| 00:47 |
|
lemonzest joined #minetest |
| 00:51 |
|
diceLibrarian joined #minetest |
| 01:00 |
|
ShadowBot joined #minetest |
| 01:04 |
|
diceLibrarian joined #minetest |
| 01:18 |
muurkha |
oh, I was thinking "how can I compress textures transmitted between minetest servers and clients" |
| 01:18 |
|
smk joined #minetest |
| 01:37 |
wsor |
just going to drop https://github.com/minetest-mods/skinsdb/pull/91 here for the people noticing that the skins database has been down for a few months. also tagging Krock |
| 02:31 |
|
Lesha_Vel joined #minetest |
| 04:00 |
|
MTDiscord joined #minetest |
| 04:57 |
|
fluxionary joined #minetest |
| 05:10 |
|
calcul0n_ joined #minetest |
| 05:39 |
|
calcul0n joined #minetest |
| 06:00 |
|
Talkless joined #minetest |
| 06:02 |
|
s20 joined #minetest |
| 06:50 |
|
jaca122 joined #minetest |
| 07:05 |
|
olliy joined #minetest |
| 07:22 |
|
definitelya joined #minetest |
| 08:22 |
|
Warr1024 joined #minetest |
| 08:46 |
|
Warr1024 joined #minetest |
| 09:14 |
|
sys4 joined #minetest |
| 10:09 |
|
liceDibrarian joined #minetest |
| 10:18 |
|
imi joined #minetest |
| 10:20 |
|
TheSilentLink joined #minetest |
| 10:27 |
jonadab |
There is a categorical difference in prominence and importance, between PNG and Targa. And it is not in Targa's favor. Software that supports Targa, invariably also supports 30+ other image formats. Whereas it's almost easier to list software that *doesn't* support PNG. |
| 10:32 |
erle |
jonadab it depends on the space you are in. on the web, tga is nothing. in gaming, well, a lot of textures on steam start their life as TGA or so i heard. |
| 10:33 |
erle |
the source engine uses it as base for VTF for example |
| 10:34 |
erle |
in any case, it has far more support than any new format, as it is probably the most basic widely-supported format that gives you the benefits of most of PNG color formats without the complexity |
| 10:34 |
erle |
jonadab what do you think of the tga.z idea btw? |
| 10:34 |
erle |
i literally see no downsides, as having smaller files than optipng can do without the runtime of optipng is pretty good (unexpectedly so) |
| 10:35 |
erle |
and it's not like minetest uses PNG gamma (which leads to some weird blending bugs obv, but no one seems to care) |
| 10:38 |
erle |
btw, if anyone here is rendering text in minetest mods, please tell me if you do bidi stuff |
| 10:38 |
erle |
because if there is no interest at all, you get a broken bidi rendering ig |
| 10:38 |
erle |
as in: if you mix hebrew and english it might look wrong |
| 11:22 |
ROllerozxa |
I'm curious about your text renderer, does it write the text to a texture and send it as dynamic media? |
| 11:46 |
s20 |
Might be a mundane question: Can people modify block textures to make them transparent? |
| 11:47 |
s20 |
Does it depend on node properties which does not enable it for specific ones? |
| 11:50 |
MTDiscord |
<luatic> s20: yes, it depends on the node def - see the use_texture_alpha field (documented in lua_api.md) |
| 11:51 |
s20 |
Ah, thanks for the clarification. Here's your answer @qasdwer |
| 11:51 |
s20 |
Hmm, ping didn't work :shrug: |
| 12:05 |
s20 |
What languages are available for translation/can be translated for the Minetest client? |
| 12:12 |
rubenwardy |
full list: https://hosted.weblate.org/projects/minetest/minetest/ |
| 12:13 |
rubenwardy |
not all of those are enabled in the client because of 1) no RTL support 2) other bugs with certain languages 3) translations not reaching a high enough % to be enabled |
| 12:17 |
s20 |
Ah, I wanted to know it for Hindi. However, I remember the default MT font not supporting it? |
| 12:19 |
s20 |
Checking that once |
| 12:19 |
s20 |
It seems that it does not |
| 12:20 |
erle |
ROllerozxa, what i have created is a thing that fills a pixel array in exactly the format tga_encoder expects it. you would send it as dynamic media, like mcl_maps and xmaps do. |
| 12:22 |
erle |
(tga_encoder obviously expects different formats for indexed, 16bpp, 24bpp, 32bpp) |
| 12:22 |
erle |
(or monochrome) |
| 12:22 |
erle |
so basically i have a pure-lua unicode-text-to-pixels-in-a-texture rendering lib |
| 12:25 |
erle |
before someone asks: no, it's not a good idea to use minetest.encode_png() with the pixel array. first, it only supports RGBA (while you probably want your text to be monochrome or indexed color-format-wise), second it produces ridiculously large files. third, as i have explained above, even optipng (which can take several seconds for a texture) is unable to beat the “throw zlib at the tga” hack filesize- or runtime-wise. |
| 12:26 |
erle |
rubenwardy, can bidi-text be input in formspecs at all? |
| 12:26 |
erle |
like are there issues? |
| 12:26 |
rubenwardy |
not sure but I imagine not |
| 12:26 |
erle |
imagine not input or imagine no issues |
| 12:27 |
rubenwardy |
I imagine big issues |
| 12:27 |
erle |
well i have researched sign libraries and since there is none as ambitious (i.e. full unicode), i.g. we have to try |
| 12:27 |
rubenwardy |
Minetest doesn't support RTL, you can expect a lot of broken stuff when trying it |
| 12:27 |
erle |
ah |
| 12:28 |
erle |
rubenwardy, btw where does the image for tga_encoder on contentdb come from? i certainly did not put it there |
| 12:29 |
erle |
i can provide images, but then i would need to know what to show. it's a library after all. |
| 12:29 |
rubenwardy |
https://content.minetest.net/packages/erlehmann/tga_encoder/audit/ |
| 12:29 |
rubenwardy |
the audit log shows that ROllerozxa added it |
| 12:29 |
ROllerozxa |
hi yes that was me |
| 12:30 |
erle |
maybe i can just make a similar image but generated by the library (and smaller) |
| 12:30 |
erle |
give me a moment ^^ |
| 12:31 |
rubenwardy |
images must be t least 950x517, and only png and jpeg is supported |
| 12:31 |
erle |
give me more of a moment :P |
| 12:32 |
|
appguru joined #minetest |
| 12:56 |
|
Thelie joined #minetest |
| 13:15 |
erle |
rubenwardy ROllerozxa consider the following: https://git.minetest.land/erlehmann/tga_encoder/src/branch/master/logo.lua |
| 13:15 |
erle |
if you wonder why the logo in the source is upside down, i keep to the default scanline order (i do not have to, but i do) |
| 13:16 |
erle |
the upscaled logo is 1200×1200, would that be allowed or even a good logo for tga_encoder? |
| 13:16 |
erle |
i mean it's not much better than the current one lol |
| 13:16 |
erle |
maybe even worse |
| 13:16 |
erle |
who knows |
| 14:14 |
|
emorrp1-irc joined #minetest |
| 14:15 |
|
emorrp1-irc left #minetest |
| 14:32 |
|
sys4 joined #minetest |
| 14:38 |
|
proller joined #minetest |
| 15:06 |
|
Desour joined #minetest |
| 15:23 |
ROllerozxa |
erle: heh neat |
| 15:23 |
ROllerozxa |
might wanna pad out the image so it's close to a 16:9 ratio (also because it will cut off at the edges to fit) |
| 15:23 |
erle |
send a patch lol |
| 15:24 |
ROllerozxa |
@_@ |
| 15:24 |
erle |
do we have a pixelops library? |
| 15:24 |
erle |
for scaling and shit |
| 15:25 |
erle |
for more fun see https://git.minetest.land/erlehmann/tga_encoder/src/branch/master/examples.lua |
| 15:29 |
erle |
hmm, the tga test images should probably be generated using tga_encoder |
| 15:29 |
erle |
goals |
| 15:30 |
erle |
ROllerozxa i enjoy the colormap feature so much for putting small images in source code |
| 15:31 |
erle |
ROllerozxa the arrows in xmaps for example are generated like that https://git.minetest.land/erlehmann/xmaps/src/branch/master/generate_textures.lua |
| 15:33 |
erle |
ROllerozxa since you added the image to tga_encoder, you using it for anything? or was it just “that thing is lacking a picture” |
| 15:37 |
ROllerozxa |
well I find it to be a useful library, but I don't actually use it for anything at the moment |
| 15:38 |
ROllerozxa |
anyways I thought just padding out the table could do it |
| 15:38 |
|
bodiccea joined #minetest |
| 15:40 |
|
TheCoffeMaker joined #minetest |
| 15:53 |
erle |
ROllerozxa, wdym padding the table |
| 16:04 |
|
rod_tout_court joined #minetest |
| 16:13 |
|
s20 joined #minetest |
| 16:18 |
ROllerozxa |
increasing the table size with black pixels, for the screenshot image |
| 16:30 |
erle |
rubenwardy, why does contentdb show “recent positive reviews” and not “recent helpful reviews”? |
| 16:30 |
erle |
are the recent reviews not helpful enough because no one upvoted them? |
| 16:31 |
erle |
asking because for some packages there are positive reviews that say nothing and positive reviews that say A LOT |
| 16:34 |
MTDiscord |
<mistere_123> I would imagine that no one cares about negative reviews in this case; its for the purpose of suggesting the next thing to try |
| 16:36 |
erle |
then you could limit it to positive HELPFUL reviews |
| 16:37 |
erle |
i actually think negative reviews are pretty interesting |
| 16:37 |
erle |
i'd love to try out mods by controversial (smallest difference between positive & negative votes, but high number of votes overall) |
| 16:44 |
|
hyvoid joined #minetest |
| 16:46 |
|
proller joined #minetest |
| 16:55 |
|
fling joined #minetest |
| 16:57 |
|
appguru joined #minetest |
| 17:21 |
|
Fleckenstein joined #minetest |
| 17:50 |
jonadab |
erle: It isn't just the web. PNG is supported by things like printers, file managers, email clients, the things that draw wallpaper on the root window / desktop, and flipping widget sets for crying out loud. It's probably the second most widely supported image format in the history of computing, after jpeg. |
| 17:51 |
sfan5 |
I see there are productive discussions being had again |
| 17:51 |
erle |
jonadab so? those environments *mostly* do not have the constraints of game engines. it does not matter at all how widespread something is *if your current toolchain supports it*. |
| 17:52 |
erle |
similarly i could go on how gif is widespread, it would still *probably* be a stupid idea to use it in minetest |
| 17:52 |
MTDiscord |
<luatic> erle: re "pixelops library": best i can do is https://github.com/appgurueu/modlib/blob/master/tex.lua |
| 17:53 |
jonadab |
erle: My point was that ".png is supported with this gross method, so it follows that Targa should be supported" is not sound logic. |
| 17:53 |
erle |
jonadab oh, the argument was more overall one of code quality: if this weird and hackish way of doing was accepted in review, this non-gross and architecturally sound way probably has no issues. |
| 17:54 |
jonadab |
erle: I think you should make the case for Targa on its own merits, without reference to something that is already in the codebase. |
| 17:55 |
erle |
targa is already in the codebase, zlib is already there |
| 17:55 |
jonadab |
Ah. |
| 17:55 |
jonadab |
Didn't realize that. |
| 17:55 |
erle |
always has been |
| 17:55 |
erle |
so i merely showed that tga.z is 30% smaller than *the most optimized png* for a typical minetest texture |
| 17:55 |
jonadab |
zlib makes since, most open-source software of any complexity uses it. |
| 17:55 |
MTDiscord |
<luatic> just make the case for sending some media (targa and wavefront obj, perhaps even .x and blitz3d) zlib-compressed, there hardly is a reason not to |
| 17:55 |
erle |
so you merelly need to put together the pieces together differently |
| 17:56 |
erle |
luatic well it was not all intuitive that the savings would be that big |
| 17:57 |
erle |
i merely found it when i did the unicode text rendering library |
| 17:57 |
erle |
and was pretty amused |
| 17:57 |
ROllerozxa |
.x text format really could use some compression... |
| 17:57 |
sfan5 |
we could really use a better model format |
| 17:57 |
ROllerozxa |
yes of course |
| 17:57 |
jonadab |
30% size savings is probably worth making for something that has to be sent via the network, if there isn't a major perf difference in some other way (e.g., CPU cycles) against it. |
| 17:58 |
sfan5 |
on that note I haven't had time to review the static gltf loading PR |
| 17:58 |
erle |
nah there is not |
| 17:58 |
erle |
what i think can be done and reduce traffic by quite a bit is, add support for tga.z and obj.z and then send it to older clients uncompressed |
| 17:58 |
jonadab |
Many users won't notice a 30% savings by itself, but these things can add up. |
| 17:58 |
erle |
what should *not* be done is automatically compress all tga or obj |
| 17:59 |
erle |
just in case some rando thinks it's a good idea, it's a shitty one and don't do it |
| 17:59 |
erle |
jonadab, several projects spend a ridiculous amount of time to run optipng on everything |
| 17:59 |
erle |
it pollutes the git history and generally sucks |
| 18:00 |
erle |
also because png has metadata, you regularly get mods where stuff is *not* optimized and the entire script of the bee movie is there in some png chunk or so |
| 18:00 |
erle |
at least mineclone2 had that several times |
| 18:00 |
erle |
a format that only has raw pixels does not offer this failure mode ;) |
| 18:01 |
MTDiscord |
<luatic> hah, imagine not scrubbing metadata |
| 18:01 |
erle |
and tbh the main pull for me is the encoding speed (as i generate textures dynamically) |
| 18:01 |
jonadab |
So we're talking about having the game server automatically convert/compress the images, what, as needed, when a client needs them? |
| 18:01 |
erle |
no |
| 18:01 |
erle |
that is wrong |
| 18:01 |
erle |
i already wrote that |
| 18:02 |
jonadab |
When would it happen then? At startup? |
| 18:02 |
MTDiscord |
<goodclover> PBM would be nice. |
| 18:02 |
erle |
i have spent a lot of time on figuring out what works and what not and “throw everything into zlib” does not work |
| 18:03 |
MTDiscord |
<luatic> I think erle is proposing burdening the mod authors with this |
| 18:03 |
MTDiscord |
<luatic> which I'm not sure is a good idea |
| 18:03 |
erle |
luatic not necessarily |
| 18:03 |
MTDiscord |
<luatic> throwing everything into zlib certainly does work |
| 18:03 |
MTDiscord |
<luatic> just make sure to only use zlib if the size savings are worth it |
| 18:03 |
erle |
yes that's the point |
| 18:03 |
erle |
it sucks |
| 18:03 |
jonadab |
Requiring module authors to convert images to Targa would be insane. |
| 18:03 |
erle |
and you don't *always* save stuff |
| 18:03 |
erle |
jonadab, you would not require that |
| 18:04 |
erle |
but if so, it's like 20 lines of shell to convert a mod (have done that with minetest game and mineclone2 to see if the savings are imaginary) |
| 18:04 |
MTDiscord |
<luatic> erle: just have the the server try compression at startup and fall back to sending media files uncompressed if it turns out to not be worth it |
| 18:05 |
erle |
why is it that EVERY SINGLE TIME i say “there is an obvious but subtly wrong solution to this problem DO NOT DO THE THING” people come out and immediately insist on DOING THE THING? |
| 18:05 |
jonadab |
Because you haven't given an alternative yet, in terms that folks can understand. |
| 18:05 |
erle |
do you think i have some strange ulterior motive for keeping you from doing greedy optimizations or so |
| 18:06 |
erle |
what i propose is simply to start *small*, support tga.z and obj.z as file formats with backwards compatiblity, server-side. everything else can follow afterwards. |
| 18:06 |
jonadab |
Ah. |
| 18:07 |
jonadab |
So you're proposing not to _require_ mod authors to convert to Targa, but to _allow_ them to do so? |
| 18:07 |
erle |
yes |
| 18:07 |
erle |
specifically because i have written a text rendering library, for which the savings are *very large* |
| 18:07 |
jonadab |
I mean, that's harmless, but module authors aren't going to do it. |
| 18:07 |
erle |
as i said, people are willing to wait hours to get 3000 files through optipng |
| 18:08 |
erle |
if you really want small transfer size, converting them to another file format and using deflate is MUCH faster |
| 18:08 |
erle |
and just to spell it out: the reason why you do not want to automatically recompress everything before sending is a) CPU cycle waste b) it depends on the content of the image if you get any space savings at all |
| 18:09 |
erle |
that's trivially true for every ccmpressed image format lol |
| 18:09 |
erle |
just fill it with random noise |
| 18:10 |
erle |
jonadab, the unique value proposition is a) smallest transfer size b) fast creation of textures |
| 18:10 |
erle |
so it is probably the best method for dynamically creating textures |
| 18:10 |
erle |
like the best, even in theory |
| 18:11 |
erle |
like even if you fix minetest.encode_png() to use whatever optipng does, it could not ever beat it in runtime and filesize |
| 18:11 |
erle |
because a) it would be noticeably slow for generated textures b) regardless you don't get a file that small unless you get to larger texture sizes than minetest typically uses |
| 18:12 |
erle |
my ultimate goal is fast and efficient dynamic texture generation and transfer |
| 18:12 |
erle |
e.g. unicode on signs |
| 18:12 |
erle |
and by noticably slow i mean “optipng may take several seconds *per texture* even on the fastest hardware that my employer allows me to use” |
| 18:12 |
MinetestBot |
[git] SmallJoker -> minetest/minetest: Inventory: Fix assertion caused by a no-op stack movement 0ba899e https://github.com/minetest/minetest/commit/0ba899e2390a397b2c68e33e96f8ba6c48739013 (2023-08-27T18:12:10Z) |
| 18:12 |
MinetestBot |
[git] SmallJoker -> minetest/minetest: Inventory: skip redundant stack movement bf9f831 https://github.com/minetest/minetest/commit/bf9f831cb2a7977c16fe146abdd41c5ee85d0bb9 (2023-08-27T18:12:10Z) |
| 18:12 |
MinetestBot |
[git] savilli -> minetest/minetest: Fix potential freeze in core.check_for_falling 852d6a7 https://github.com/minetest/minetest/commit/852d6a7976d05e5ab58d6b8a685b1b488f914840 (2023-08-27T18:12:53Z) |
| 18:13 |
erle |
jonadab you can try all that out yourself, install mtpaint or so, draw something and then save it in different formats. |
| 18:14 |
erle |
basically, what quake etc. did (tga in zip) is a low-complexity, high-throughput local optimum that you probably can't beat at the texture sizes we are looking at |
| 18:14 |
erle |
and 30% is 30% |
| 18:15 |
erle |
now you can certainly make up something more complicated involving custom dictionaries for zstd or something |
| 18:16 |
erle |
but my experience has taught me that this never pans out |
| 18:17 |
erle |
see minetest.compress() with zstd (which could be better than zlib, but so far the only cases i have seen where it performed better than zlib were when you gave it several seconds to compress stuff or had more than about 4MB to 8MB of data) and minetest.encode_png(), which creates PNGs that are about 20 times the size that they should be |
| 18:18 |
erle |
in both cases you have a complex solution which could be tuned, but it would put a *huge* burden on mod authors to, say, provide zstd dictionaries |
| 18:19 |
MinetestBot |
[git] grorp -> minetest/minetest: Small setting-related fixes (#13755) 7b56daa https://github.com/minetest/minetest/commit/7b56daa2368be8de8514487f6c937f342d2d6ac5 (2023-08-27T18:18:41Z) |
| 18:19 |
erle |
so back to the original point: i think just slapping already supported primitives together is a good idea |
| 18:19 |
erle |
while trying to invent new image formats is not |
| 18:21 |
erle |
jonadab also, to say it again: if you put images in item meta, you CERTAINLY want them to be as small as possible |
| 18:21 |
erle |
because huge item meta = lag |
| 18:25 |
erle |
and i'm pretty sure that the same kind of clowns who gave us this gem will put them in item meta |
| 18:25 |
erle |
https://git.minetest.land/MineClone2/MineClone2/pulls/1939#issuecomment-30134 |
| 18:25 |
erle |
and you can't run optipng on item meta |
| 18:25 |
erle |
surely you see where i am going with this |
| 18:25 |
erle |
after all, i got a 97% filesize reduction in the linked comment |
| 18:25 |
erle |
and no one before me even noticed |
| 18:26 |
erle |
>4 kilobytes for a 16x16 texture, just by saving it using GIMP lol |
| 18:26 |
erle |
that makes like 16 bytes for each pixel hehehehe |
| 18:32 |
erle |
jonadab, i think the only lesson here is: if you make a NEW engine and you have small textures (below prob 256x256 or so) mainly, use tga.z and don't bother with any other format |
| 18:32 |
erle |
it eliminates entire categories of png bloat errors :P |
| 19:00 |
|
Fleckenstein joined #minetest |
| 19:11 |
|
proller joined #minetest |
| 20:18 |
|
kyle56 joined #minetest |
| 20:28 |
kyle56 |
erle: just read the chat log and saw what you did with the goatse. thats funny AF. reminds me of what i did to make sure links to my website were removed from the MT forum since they were on a page i could not access after i had asked my account be deleted. i like to look under the tail -F of my web server. saw that months after my account was deleted, someone found an old page through the forum. decided to have some fun. >:) |
| 20:45 |
|
proller joined #minetest |
| 20:47 |
|
kyle56 left #minetest |
| 20:52 |
muurkha |
ego puto in horto meo. |
| 20:53 |
|
x_4 joined #minetest |
| 20:53 |
|
Lunatrius` joined #minetest |
| 20:54 |
|
pounce_ joined #minetest |
| 20:54 |
|
m42uko_ joined #minetest |
| 20:55 |
|
TheCoffeMaker_ joined #minetest |
| 20:56 |
|
rubywarden joined #minetest |
| 20:56 |
|
fluxionary_ joined #minetest |
| 21:01 |
|
sys4_ joined #minetest |
| 21:28 |
MTDiscord |
Command sent from Discord by not luatic: |
| 21:28 |
MTDiscord |
!tell TenPlus1 have you considered moving (or mirroring) your mods from notabug to a more reliable service such as codeberg? (notabug has been down again today) |
| 21:28 |
MinetestBot |
MTDiscord: I'll pass that on when TenPlus1 is around |
| 22:35 |
|
panwolfram joined #minetest |
| 23:02 |
|
sparky4 joined #minetest |
| 23:19 |
|
Lesha_Vel joined #minetest |
| 23:45 |
|
sparky4 joined #minetest |
| 23:54 |
jonadab |
erle: Pretty sure the last time I used mspaint, it was running on Windows 95 OSR2, and did not support formats other than .bmp |
| 23:54 |
erle |
jonadab, mtpaint, not mspaint |
| 23:54 |
jonadab |
Oh, I thought that was a typo. |
| 23:55 |
erle |
great, you are one of these google people hehe |
| 23:55 |
jonadab |
Never heard of mtpaint. |
| 23:55 |
erle |
PGP? did you mean: PHP? we helpfully searched for PHP instead |
| 23:55 |
erle |
:P |
| 23:55 |
jonadab |
No, I know what PGP was. |
| 23:55 |
jonadab |
Though I haven't heard much about it lately. |
| 23:55 |
erle |
i think google did that once |
| 23:55 |
erle |
it still works and is still used |
| 23:56 |
erle |
many people who say they “don't use PGP” use it to sign commits and for pass(1) the standard unix password manager (ed is the standard editor) |
| 23:56 |
erle |
anyways, mtpaint is a pretty nice program to edit textures, especially colormapped ones |
| 23:57 |
erle |
https://en.wikipedia.org/wiki/MtPaint |
| 23:57 |
cheapie |
I'm assuming you don't mean this MT Paint: https://cheapiesystems.com/media/mtpaint-startup.webm |
| 23:57 |
jonadab |
How on earth could people be using software that costs thousands of dollars per seat and not know it? |
| 23:57 |
erle |
Mark Tyler's Painting Program |
| 23:58 |
erle |
jonadab, you mean “they use the PGP protocol, not PGP, but GPG”? ;) |
| 23:58 |
jonadab |
Ah. |
