| Time |
Nick |
Message |
| 00:41 |
|
DeepThgt joined #minetest |
| 00:51 |
|
n_to joined #minetest |
| 00:55 |
erle |
https://mister-muffin.de/p/yMHl.png unwanted horizon tilt is eliminated by my friend li0n |
| 00:56 |
erle |
minimum patch maximum commit message https://git.minetest.land/erlehmann/xcam/commit/768c8b2caa01a0f631eacbe59913e84c4fc7f433 |
| 01:03 |
|
ShadowBot joined #minetest |
| 01:09 |
MTDiscord |
<jordan4ibanez> https://youtu.be/RAxomGVR12E?feature=shared&t=374 minetest sixel support when? |
| 01:09 |
MTDiscord |
<jordan4ibanez> This is of course a joke, but cool either way |
| 01:12 |
|
smk joined #minetest |
| 01:31 |
|
acarrico1 joined #minetest |
| 01:33 |
acarrico1 |
I've got a minetest server logging question. |
| 01:34 |
acarrico1 |
In minetest.conf, I've got debug_log_level = warning |
| 01:35 |
acarrico1 |
And that is correctly logging warnings to /var/lib/minetest/.minetest/debug.txt |
| 01:36 |
acarrico1 |
So far so good, but the systemd logs are getting spammed with action logs. |
| 01:36 |
acarrico1 |
I assume these are written to stdout or stderr and systemd picks them up. |
| 01:37 |
acarrico1 |
Any idea how to stop these ACTION logs? Are there actually two different mechanisms or am I crazy? |
| 02:01 |
acarrico1 |
... It looks like the debian systemd unit sets StandardOutput=null which I guess would fix it. |
| 02:01 |
acarrico1 |
Still does anybody know why the server is verbose logging to standard output? |
| 02:03 |
acarrico1 |
The NixOS systemd unit has StandardOutput=journal which is a problem. |
| 02:10 |
acarrico1 |
Looks like there is a command line option --quiet which may help |
| 02:12 |
|
v-rob joined #minetest |
| 02:14 |
acarrico1 |
Doesn't seem like there is a quiet option in minetest.conf |
| 02:33 |
DeepThgt |
The NixOS systemd unit has StandardOutput=journal which is a problem. |
| 02:33 |
DeepThgt |
ooof |
| 02:33 |
acarrico1 |
... OK so adding StandardOutput=null to the systemd unit stops the minetest server from spamming the journal. |
| 02:34 |
acarrico1 |
DeepThgt: Yup I added it in my confiuration.nix and it stops |
| 02:34 |
acarrico1 |
I don't understand why the server is writing standard out when it has a log file. |
| 02:34 |
DeepThgt |
good, stupid default on nixos's part |
| 02:36 |
DeepThgt |
idk the motivations for why it does, but i am very greatful it does |
| 02:36 |
DeepThgt |
its much easier than going back and forth to a log |
| 02:36 |
DeepThgt |
when setting up a server initially |
| 02:36 |
acarrico1 |
Do you happen to know if --quiet stops that? |
| 02:37 |
DeepThgt |
i can find out, sec |
| 02:37 |
acarrico1 |
If so it is probably a better solution |
| 02:37 |
acarrico1 |
since the server doesn't waste resources writing to /dev/null |
| 02:37 |
DeepThgt |
7a\ |
| 02:38 |
DeepThgt |
ya* |
| 02:38 |
DeepThgt |
it suppressed all but warnings/errs |
| 02:38 |
acarrico1 |
cool. |
| 02:39 |
acarrico1 |
I'll try that and maybe send a patch to NixOS |
| 02:42 |
|
liceDibrarian joined #minetest |
| 02:42 |
acarrico1 |
DeepThgt: thx |
| 03:13 |
|
pellucid joined #minetest |
| 03:31 |
|
v-rob joined #minetest |
| 04:30 |
|
sys4_ joined #minetest |
| 04:53 |
|
sys4 joined #minetest |
| 05:00 |
|
MTDiscord joined #minetest |
| 05:08 |
|
qqq joined #minetest |
| 05:37 |
erle |
rubenwardy very recently CDB started failing to load images on one of my devices. i found the result to be that content negotiation seems to be broken: even with an accept header of image/png;q=0.8,image/jpeg;q=0.7,image/*;q=0.6,*/*;q=0.5 it ALWAYS returns a webp image. |
| 05:46 |
erle |
mostly unrelated: it seems the person who exposed the fact that lossy webp does not compare favourably to jpeg (google just compared it to the reference encoder cjpeg, not the highly optimized mozjpeg) has made a post on lossless webp as well, neat: https://siipo.la/blog/whats-the-best-lossless-image-format-comparing-png-webp-avif-and-jpeg-xl |
| 05:47 |
erle |
> WebP is a good choice for lossless images since it easily wins over PNG when it comes to compression efficiency, having 41% smaller images on average. |
| 05:47 |
erle |
seems they did not lie in their benchmarks this time! |
| 05:47 |
erle |
> JPEG XL is the new kid on the block and makes a strong impression. The average file size reduction is around 48%, which is slightly better than WebP. |
| 05:47 |
erle |
lol |
| 05:53 |
erle |
ok so the png thumbnails still are accessible if i change the file extension. https://content.minetest.net/thumbnails/2/97de0f72cf.png |
| 05:55 |
erle |
they do have visible compression artifacts though |
| 05:55 |
erle |
meaning it is not lossless webp |
| 05:55 |
erle |
and using mozjpeg is probably a better idea |
| 05:55 |
erle |
i mean the webp images |
| 05:58 |
erle |
rubenwardy visually comparing the png thumbnails to the webp thumbnails i found that the webp thumbnails seem to lose a surprising amount of detail here. in-game images look slightly blurry and have lower contrast. |
| 05:59 |
erle |
like compare https://content.minetest.net/thumbnails/2/97de0f72cf.png and https://content.minetest.net/thumbnails/2/97de0f72cf.webp |
| 06:00 |
erle |
this is really weird |
| 06:04 |
erle |
like the grass and all nodes underwater just have lower contrast and the tree leaves are darker in webp and the area around the chair seems to experience some color bleeding |
| 06:04 |
erle |
given that i can see this at 100% without glasses and with my bad constrast vision i wonder how noticeable this is for others |
| 06:05 |
erle |
i guess using a lossy format for these kind of high-noise screenshots is not the best idea in general. i do not know how to encode a webp image though, but it should probably set to “lossless” in this case. |
| 06:06 |
erle |
ah, now i see it |
| 06:06 |
erle |
the webp image is 10% of the size |
| 06:06 |
erle |
no wonder it is so badly mangled |
| 06:09 |
|
kamdard joined #minetest |
| 06:17 |
erle |
rubenwardy i can see no visual difference between a jpeg at 8k and a webp at 8k, but at those rates of lossy compression, both of them seem to affect screenshots of minetest quite badly (as in: me, someone with less-than-perfect eyesight, notices issues with the thumbnails). |
| 06:19 |
erle |
you should be able to notice these issues in any screenshot that contains minetest_game tree leaves, grass and any detail behind transparent water |
| 06:27 |
erle |
wow, this level of contrast loss is pretty surprising. compare https://content.minetest.net/thumbnails/2/T5VZhncqSg.webp and https://content.minetest.net/thumbnails/2/T5VZhncqSg.png |
| 06:30 |
MTDiscord |
<greenxenith> The contrast is the same. Difference is sharpness |
| 06:30 |
erle |
look at the red |
| 06:30 |
erle |
hearts and red plant |
| 06:31 |
MTDiscord |
<greenxenith> Yep, same contrast, different sharpness |
| 06:31 |
erle |
uh, maybe it is my bad contrast vision, but i perceive VASTLY different contrast here |
| 06:31 |
erle |
how does this work? |
| 06:31 |
erle |
also interestingly enough, converting the png image in my last message to a jpeg using gimp at 65% jpeg quality does NOT reduce the perceived brightness of the red |
| 06:31 |
erle |
(the jpeg is the same size as the webp image approx.) |
| 06:32 |
erle |
so what exactly is it about webp that makes the image lose contrast? |
| 06:33 |
erle |
greenxenith, i can compare this, wait a moment |
| 06:33 |
erle |
i make a picture in gimp |
| 06:34 |
MTDiscord |
<greenxenith> The overall contrast of the images are nearly identical. The reason the reds on the plant and hearts is less bright on webp is purely a difference in sharpness. |
| 06:35 |
erle |
like, color bleeding? |
| 06:35 |
MTDiscord |
<greenxenith> more like blurring |
| 06:35 |
erle |
yeah but then a jpeg should have the same effect at those sizes, right? |
| 06:35 |
erle |
so why doesn't it? |
| 06:36 |
MTDiscord |
<greenxenith> Different algorithm, different decisions on how to handle neighboring colors at small scales |
| 06:37 |
erle |
hmm, i think this gives a hint on why webp fails at the perceptual metric thing against properly encoded jpegs even at small sizes. i would have thought it might have an edge there, since jpeg is blurry as hell at some point. |
| 06:37 |
erle |
human eye is sensitive to contrast, much more than to color |
| 06:38 |
erle |
i just wonder which tradeoffs would lead to the thing uniformly losing up contrast AND darkening things |
| 06:38 |
erle |
i mean it happens with the trees in the other screenshot as well |
| 06:38 |
erle |
the leaves i mean |
| 06:39 |
erle |
as if everything that has noise and is kinda bright becomes darker (not only mushed) in webp |
| 06:39 |
erle |
i wonder if this is a gamma thing lol |
| 06:40 |
erle |
in any case, blurry cdb is blurry and weird |
| 06:47 |
erle |
greenxenith, wow, webp is amazingly worse than jpeg. look here: https://mister-muffin.de/p/qpKc.png |
| 06:48 |
erle |
like see how the webp encoder loses all those red dots on the left? |
| 06:49 |
|
Warr1024 joined #minetest |
| 06:49 |
erle |
rubenwardy i go to bed now, but i think you too can just convert some PNG to JPEG here (if you really want lossy compression) and see how much worse whatever your webp encoder does looks at the same filesize https://mister-muffin.de/p/qpKc.png |
| 06:50 |
erle |
but i guess given that webp is a better *lossless* format what was intended here was probably lossless mode |
| 06:50 |
erle |
no idea how that is turned on |
| 06:51 |
erle |
maybe there was no content negotiation after all https://github.com/minetest/contentdb/commit/4578cb157f7fe4612d8a9a3b7c3b56bc86a2e83f |
| 06:55 |
erle |
https://github.com/minetest/contentdb/blob/master/app/blueprints/thumbnails/__init__.py |
| 06:57 |
erle |
i think there may just be a “lossless=true” missing here |
| 07:01 |
|
calcul0n joined #minetest |
| 07:31 |
|
definitelya joined #minetest |
| 08:01 |
|
cation joined #minetest |
| 09:03 |
|
jaca122 joined #minetest |
| 09:07 |
|
Evergreen joined #minetest |
| 09:31 |
|
Thelie joined #minetest |
| 10:12 |
|
appguru joined #minetest |
| 10:17 |
|
Waffelo joined #minetest |
| 10:37 |
|
mrkubax10 joined #minetest |
| 10:44 |
|
Talkless joined #minetest |
| 10:48 |
ROllerozxa |
I believe it's intentional for webp thumbnails to be encoded lossily (though the quality may have been set at too low) |
| 10:51 |
ROllerozxa |
if your browser somehow doesn't support webp then you really should update it or to another one. even netsurf supports webp... doesn't render the rest of CDB particularly well, but it still supports webp |
| 11:01 |
MinetestBot |
[git] grorp -> minetest/minetest: Get rid of hidden settings in settings_translation_file.cpp 8bf2031 https://github.com/minetest/minetest/commit/8bf2031310cd2a9b17f2549484ab974b0caaf0de (2023-11-11T10:59:38Z) |
| 11:09 |
|
s20 joined #minetest |
| 11:19 |
|
frostsnow joined #minetest |
| 11:21 |
|
InFerNo_ joined #minetest |
| 11:35 |
MinetestBot |
[git] waxtatect -> minetest/minetest: Translated using Weblate (French) adf9a39 https://github.com/minetest/minetest/commit/adf9a3953b75b837ebc153d489bcf6d2e9eea86f (2023-11-11T11:06:10Z) |
| 11:35 |
MinetestBot |
[git] BreadT -> minetest/minetest: Translated using Weblate (Japanese) d056bb3 https://github.com/minetest/minetest/commit/d056bb3ee702c310f1956c71dc7e25909b727cd2 (2023-11-11T11:06:10Z) |
| 11:35 |
MinetestBot |
[git] Methro -> minetest/minetest: Translated using Weblate (Spanish) 842b2bb https://github.com/minetest/minetest/commit/842b2bbd3655a6bd000ed6263c02d2bf52635034 (2023-11-11T11:06:10Z) |
| 11:35 |
MinetestBot |
[git] joaquimpuig -> minetest/minetest: Translated using Weblate (Catalan) 2f279d2 https://github.com/minetest/minetest/commit/2f279d24030c7a522a6877b77098e8b0a0314ae3 (2023-11-11T11:06:10Z) |
| 11:35 |
MinetestBot |
[git] (16 newer commits not shown) |
| 12:13 |
erle |
ROllerozxa lossy webp is ALWAYS worse than jpeg lol |
| 12:14 |
erle |
i think you didn't get that memo |
| 12:14 |
erle |
ROllerozxa basically lossless webp is better than png and lossy webp is worse than everything else than cjpeg (which is the worst jpeg encoder you can test against and google did that) |
| 12:14 |
ROllerozxa |
ok |
| 12:14 |
erle |
so if you need some combination of features (like lossy with alpha) that is an advantage |
| 12:14 |
erle |
but as a standalone format, just look here: ttps://mister-muffin.de/p/qpKc.png |
| 12:15 |
erle |
https://mister-muffin.de/p/qpKc.png |
| 12:15 |
erle |
the webp and the jpeg are the same size |
| 12:15 |
erle |
but the webp loses much more detail |
| 12:15 |
erle |
(this is a detail of a cdb screenshot PNG thumbnail i encoded as jpeg with gimp and the webp served by cdb) |
| 12:16 |
erle |
ROllerozxa here https://siipo.la/blog/is-webp-really-better-than-jpeg |
| 12:16 |
erle |
> I think Google’s result of 25-34% smaller files is mostly caused by the fact that they compared their WebP encoder to the JPEG reference implementation, Independent JPEG Group’s cjpeg, not Mozilla’s improved MozJPEG encoder. |
| 12:30 |
|
s20 joined #minetest |
| 12:41 |
rubenwardy |
I intended to make is lossless |
| 12:42 |
rubenwardy |
fyi, https://content.minetest.net/thumbnails/2/T5VZhncqSg.jpg also exists |
| 12:42 |
rubenwardy |
oh nevermind |
| 12:42 |
rubenwardy |
500 |
| 12:43 |
rubenwardy |
cannot write mode P as JPEG |
| 12:43 |
rubenwardy |
well, it's supposed to exist |
| 12:44 |
rubenwardy |
<rubenwardy> I intended to make is lossless |
| 12:44 |
rubenwardy |
I suppose it explains why webp was 5-10x smaller than png |
| 13:07 |
rubenwardy |
cannot write mode P as JPEG => missing .convert('RGB') |
| 13:07 |
rubenwardy |
what jpeg quality should I use? |
| 13:08 |
erle |
depends on your encoder |
| 13:09 |
rubenwardy |
I'm using pillow |
| 13:09 |
erle |
no idea what gimp uses, but the webp at 80% was matched by jpeg at 65% with no arithmetic coding, but progressive and optimized, full colors |
| 13:09 |
erle |
did you try passing ”lossless=true” to the webp encoding call? |
| 13:09 |
rubenwardy |
yeah, haven't tested yet but that is documented |
| 13:09 |
erle |
i can help you with content negotiation later btw |
| 13:09 |
rubenwardy |
gtg |
| 13:09 |
erle |
then you don't have to manage filenames in templates at all |
| 13:13 |
erle |
btw, for png thumbnails, you can make them a little smaller if you force RGB |
| 13:13 |
erle |
i noticed some are RGBA |
| 13:13 |
erle |
which makes no sense if no alpha channel |
| 13:14 |
erle |
rubenwardy i think it would be a good idea to test manually if you look at the difference between pictures using gimp layer subtract method |
| 13:15 |
erle |
but i have no idea of the perceptual things |
| 13:15 |
erle |
for my own mods i actually do like lossless thumbnails |
| 13:18 |
|
definitelya joined #minetest |
| 13:20 |
rubenwardy |
The reason the thumbnails include the file extension is because they're statically served by nginx once generated |
| 13:24 |
rubenwardy |
You can use a picture element to select an source image based on compatible formats or available space |
| 13:24 |
rubenwardy |
But tbh I'm not interested in supporting super outdated browsers |
| 13:25 |
rubenwardy |
https://caniuse.com/webp |
| 13:26 |
erle |
uh what |
| 13:26 |
|
calcul0n joined #minetest |
| 13:26 |
erle |
i think you misunderstand |
| 13:27 |
erle |
content negotiation is not about outdated browsers, it makes stuff like templates easier to maintain for example |
| 13:27 |
erle |
and you could also have a setting in minetest for lossy or lossless thumbnails in the future if the minetest client would do it |
| 13:28 |
erle |
without changing the url |
| 13:28 |
erle |
(i was a backend dev once hehe) |
| 13:29 |
erle |
rubenwardy do you use nginx as reverse proxy? |
| 13:31 |
erle |
“WebP generally has better compression than JPEG” thanks caniuse hahahaha |
| 13:31 |
erle |
:P |
| 13:32 |
|
Oksanaa joined #minetest |
| 13:33 |
rubenwardy |
Yes, nginx is a reverse proxy to the python application but also serves the thumbnails directly from the filesystem |
| 13:44 |
erle |
rubenwardy btw, did you not notice the webp issues or are you using some metric where it was not an issue somehow? |
| 13:45 |
erle |
(i.e. an image similarity metric that does not care about cloor intensity) |
| 13:45 |
|
appguru joined #minetest |
| 13:45 |
erle |
appguru so about the horizon alignment |
| 13:45 |
erle |
it was super easy lol |
| 13:45 |
erle |
my friend li0n figured it out |
| 13:45 |
erle |
appguru read this commit and be amused: https://git.minetest.land/erlehmann/xcam/commit/768c8b2caa01a0f631eacbe59913e84c4fc7f433 |
| 13:47 |
appguru |
mhm i don't think this will work if you look perfectly down |
| 13:47 |
erle |
appguru so now it is like this! https://mister-muffin.de/p/yMHl.png |
| 13:47 |
erle |
hahaha |
| 13:47 |
erle |
it works |
| 13:47 |
rubenwardy |
I only have roughly half the images downloaded locally, I tested by inspecting the home page and package list pages but didn't zoom in |
| 13:47 |
erle |
it works perfectly |
| 13:47 |
erle |
rubenwardy well i noticed even without zoom on a dirty screen without glasses, that's why i asked |
| 13:47 |
appguru |
won't you get a zero vector if you look perfectly down or up? (0, 1, 0) and (0, -1, 0) look dir => right of (0, 0, 0) |
| 13:47 |
rubenwardy |
Yeah I can see it very clearly on my phone |
| 13:48 |
erle |
rubenwardy i think maybe it was because my *own* screenshots lost details |
| 13:48 |
erle |
like it just looked weird |
| 13:48 |
erle |
oh yeah on the phone it's prob obvious hehe |
| 13:48 |
rubenwardy |
Would have been helpful to get an up to date backup locally so I could have side by side tested |
| 13:49 |
erle |
well if the lossless webp thing works at least that's no issue. |
| 13:49 |
appguru |
i still think the proper solution is probably something like local angle = girl:get_look_horizontal() + math.pi/2; local right = vector.new(math.cos(angle), 0, math.sin(angle)) |
| 13:49 |
erle |
i kinda doubt lossy is good at all |
| 13:50 |
erle |
appguru why what does that do? |
| 13:50 |
appguru |
+ math.pi/2 may have to be - math.pi/2, the sin and cos may need to be swapped and/or inverted |
| 13:50 |
erle |
and how much is it different than my ghetto solution |
| 13:50 |
appguru |
erle: that gets the horizontal rotation and rotates it by another 90 degrees to get the "right" vector |
| 13:51 |
erle |
uh, so how much does it differ from the one i construct here? |
| 13:51 |
erle |
btw, the player can not look perfectly up or down |
| 13:51 |
erle |
check it out |
| 13:52 |
erle |
i wonder which other issues that prevents |
| 13:52 |
appguru |
"btw, the player can not look perfectly up or down" - interesting |
| 13:53 |
erle |
only ±89.5 degrees |
| 13:53 |
erle |
don't want to get gimbal locked lol |
| 14:23 |
|
Wuzzy joined #minetest |
| 14:27 |
|
kastimonoro joined #minetest |
| 14:32 |
|
calcul0n joined #minetest |
| 14:37 |
|
qqq joined #minetest |
| 14:43 |
|
Guest84 joined #minetest |
| 14:46 |
Guest84 |
hello. If I activate mobs in a world that didn't have mobs before, will that crash it? What if I regret it later? |
| 14:54 |
definitelya |
Mod conflicts would be the reason for a possible crash, make sure the active mods aren't conflicting. In case you remove the mobs mod and don't want to remove the resulting unknown entities, use the mobs mod in a new world you don't care about, I suppose. |
| 14:57 |
|
sparky4 joined #minetest |
| 14:57 |
Guest84 |
so the recommendation is to never activate mobs in worlds that are running |
| 14:59 |
definitelya |
Yes, if I use that world I won't test / try mods on it. |
| 15:00 |
Guest84 |
I understand, thank you |
| 15:36 |
|
Waffelo joined #minetest |
| 15:38 |
|
Sobinec joined #minetest |
| 15:53 |
|
mrkubax10 joined #minetest |
| 16:01 |
|
definitelya joined #minetest |
| 16:17 |
|
Sobinec joined #minetest |
| 16:36 |
|
Leopold joined #minetest |
| 17:03 |
|
Fleckenstein joined #minetest |
| 17:04 |
copygirl |
When defining a node with nodebox drawtype and facedir paramtype2, is there a way to have the textures not rotate with the facedir specified? |
| 17:07 |
MTDiscord |
<rollerozxa> world-aligned textures might be what you're looking for, I think |
| 17:09 |
copygirl |
Ah, `align_style="node"/"world"/"user"` - I'll check it out. Ty |
| 18:28 |
erle |
i get the “i do not want to support old browsers” thing, but i find it extremely frustrating if something that worked fine stops working. guess i have to figure it out. |
| 18:28 |
erle |
(also “supporting” a “browser” is a bit weird, because inversion of control haha) |
| 18:36 |
|
Krock joined #minetest |
| 18:36 |
|
Desour joined #minetest |
| 18:52 |
|
vampirefrog joined #minetest |
| 18:57 |
erle |
what's the correct way to have a wield image some degrees rotated? like, holding a map or photo in the player hand |
| 18:59 |
|
sparky4 joined #minetest |
| 19:07 |
|
sofar joined #minetest |
| 19:10 |
muurkha |
erle: agreed |
| 19:12 |
|
jaca122 joined #minetest |
| 19:13 |
|
Thelie joined #minetest |
| 19:16 |
erle |
muurkha thank you for telling me things about rendering 3d stuff. do you have any idea what i could add to xcam? |
| 19:17 |
erle |
currently it looks like this https://mister-muffin.de/p/yMHl.png |
| 19:18 |
erle |
i wondered if the photos should have metadata, but then that metadata would probably take the fun out of it |
| 19:18 |
erle |
like if there was location information, you could not just hand someone a photo of your secret base |
| 19:19 |
erle |
what if i made a panorama mode where it makes 6 photographs and then sets that as the skybox hehehehe |
| 19:31 |
|
fluxionary joined #minetest |
| 19:33 |
erle |
rubenwardy greenxenith i figured out why lossy webp looks so shitty for minetest screenshots, it only supports 4:2:0 chroma subsampling, which is basically the enemy of pixel art and fine detail. |
| 19:33 |
erle |
something to do with vp8 frame coding stuff it was built on |
| 19:44 |
|
v-rob joined #minetest |
| 20:02 |
|
v-rob joined #minetest |
| 20:11 |
|
Thelie joined #minetest |
| 20:33 |
|
imi joined #minetest |
| 20:51 |
|
liceDibrarian joined #minetest |
| 21:11 |
DeepThgt |
anyone know how to obtain the inv of bones that were placed out of bounds? |
| 21:11 |
erle |
DeepThgt yes. you have to use some mod that can access it server-side. |
| 21:12 |
DeepThgt |
suggestions? lel |
| 21:12 |
erle |
DeepThgt the map is, for various reasons i do not want to go into, surrounded by a mapblock shell that is invisible on the client, but server-side mods can still interact with it |
| 21:12 |
erle |
i.e. they are never sent to the client |
| 21:12 |
erle |
DeepThgt btw what is missing for xcam to be used on your server? also is that server fast enough that the photos do not hang it? |
| 21:13 |
DeepThgt |
i havent been able to toss it on there yet as i have bigger lag issues to address |
| 21:14 |
erle |
what kind of lag issues? |
| 21:14 |
erle |
what mods does it run? |
| 21:14 |
DeepThgt |
http://stealinyour.info/Pernia/.git/tree/ |
| 21:15 |
DeepThgt |
lag issues meaning lua sucks |
| 21:15 |
DeepThgt |
we are maxxing out the one thread it runs on |
| 21:15 |
DeepThgt |
with luajit |
| 21:16 |
erle |
uh no |
| 21:16 |
erle |
lag issues usually mean either your mod sucks or, by extension, the author sucks |
| 21:17 |
DeepThgt |
or in the case of technic, lua sucks |
| 21:17 |
DeepThgt |
issues only started happening when myself and the other admin started really using it hard |
| 21:17 |
erle |
so what makes it “lua sucks” and not “technic sucks” |
| 21:17 |
erle |
i mean yes, lua sucks |
| 21:17 |
erle |
but that's rarely the performance problem in my experience |
| 21:18 |
DeepThgt |
its fine up until you heavily, heavily use it |
| 21:18 |
DeepThgt |
and technic does suck, technic_plus fixed the suck as far as i can tell |
| 21:18 |
DeepThgt |
to where it seems like lua is the bottleneck |
| 21:18 |
erle |
so what do you do then |
| 21:19 |
erle |
that makes it lag |
| 21:19 |
erle |
also consider that making a photo can only ever introduce ONE lag spike per photo taking |
| 21:19 |
DeepThgt |
hv networks with 20k nodes, 30+ quarries running at the same time |
| 21:19 |
erle |
if you have a good idea on how to make that less problematic btw i will take advice |
| 21:29 |
DeepThgt |
ok, so it looks like one of the 3 things would fix my issue |
| 21:30 |
DeepThgt |
moving the bones node above CONTENT_IGNORE node somehow |
| 21:30 |
DeepThgt |
my client ignoring CONTENT_IGNORE |
| 21:30 |
DeepThgt |
or being able to modify the bone node's inventory |
| 21:30 |
DeepThgt |
none of the 3 i have an off the shelf solution either |
| 21:31 |
DeepThgt |
so, would the bug that i just encountered be in minetest, or minetest_game? |
| 21:31 |
DeepThgt |
idk which handles bones |
| 21:34 |
|
v-rob joined #minetest |
| 21:40 |
erle |
DeepThgt look the thing is, the nodes are NEVER sent to the client |
| 21:40 |
erle |
DeepThgt they exist, they can be handled by server sent mod |
| 21:40 |
erle |
DeepThgt “ignore” nodes are just parts of the map that are not loaded. i.e. the (visible) map border is permanently unloaded on the client. |
| 21:40 |
DeepThgt |
ok |
| 21:41 |
DeepThgt |
so, either move the node or modify its inv |
| 21:41 |
erle |
if you have a mod that moves nodes and also their node meta (i.e. the inventory) you can easily hide stuff in the map border |
| 21:42 |
erle |
(and it would be extremely pointless to special case the map border for those) |
| 21:42 |
erle |
i did it |
| 21:42 |
erle |
so the trick is to have some kind of hopper node |
| 21:42 |
erle |
that can take inventory from a node above it |
| 21:42 |
erle |
and spit it out to the side |
| 21:43 |
erle |
btw, if you want to see what kind of “extremely pointless attempts to fix stuff at the map border” i mean, go visit the map border in mineclone2 haha |
| 21:43 |
erle |
and spawn some mobs there |
| 21:44 |
DeepThgt |
i have some mobs at the bottom in mt-game |
| 21:44 |
erle |
DeepThgt how did the bones get there in the first place btw? |
| 21:44 |
DeepThgt |
its fun the cats fall into the shadows and hop back out |
| 21:44 |
DeepThgt |
.... TNT |
| 21:44 |
DeepThgt |
lol |
| 21:45 |
erle |
wait but how did you get out of bounds |
| 21:46 |
erle |
i mean i know you can explode tnt out of bounds |
| 21:46 |
DeepThgt |
the node in bounds was apparently occupied wheni died |
| 21:46 |
DeepThgt |
so the bones went to the next unoccupied node |
| 21:46 |
DeepThgt |
as its want to do |
| 21:46 |
erle |
i guess the fix would be to chose the next unoccupied node in the current mapblock probably |
| 21:47 |
DeepThgt |
or have all beyond the barrier be CONTENT_IGNORE |
| 21:47 |
DeepThgt |
instead of 1m deep |
| 21:47 |
DeepThgt |
with empty space |
| 21:54 |
erle |
DeepThgt not again this bullshit |
| 21:54 |
erle |
DeepThgt everyone who hears about this for the first time has like SUPER NEAT ideas about how the map border should be changed |
| 21:54 |
erle |
DeepThgt even coredevs |
| 21:54 |
erle |
(in the past) |
| 21:54 |
DeepThgt |
so, there should be documentation fo whats been discusssed lol |
| 21:54 |
erle |
it's a chestertons fence thing |
| 21:55 |
erle |
as i said, the visible map border is just an area that is permanently unloaded on the client |
| 21:55 |
erle |
EVERYTHING else follows from that |
| 21:56 |
erle |
i mean everything that concerns you ig |
| 21:56 |
erle |
as to why the shell is needed, imagine someone writing a mod that generates structures and such a structure gets placed at a visible boundary |
| 21:56 |
erle |
as such that it transgresses the boundary |
| 21:57 |
erle |
you can either have what you have right now, i.e. “it just works” |
| 21:57 |
erle |
or you have to do with a ton of cases where random code special cases the map border |
| 21:57 |
erle |
other games do this too btw |
| 21:58 |
erle |
like when in command & conquer you have units spawning out of bounds and driving into the map |
| 21:58 |
erle |
the real map is bigger than what you see |
| 21:58 |
erle |
pokemon (gameboy) too |
| 21:58 |
erle |
that what you see on screen is not the entire map is a common and useful idiom |
| 21:59 |
erle |
DeepThgt btw is that on your server? |
| 22:01 |
DeepThgt |
ya |
| 22:04 |
erle |
DeepThgt i can give you a way funnier solution then |
| 22:05 |
DeepThgt |
im all ears |
| 22:05 |
DeepThgt |
i just got why my idea wouldnt work. duh |
| 22:05 |
DeepThgt |
low sleep lol |
| 22:05 |
erle |
add 1000 to MAX_MAP_GENERATION_LIMIT in src/constants.h and then recompile minetest and start the server with that |
| 22:05 |
erle |
i mean you can add a lower number |
| 22:05 |
erle |
but at least 80 |
| 22:06 |
DeepThgt |
that would work |
| 22:06 |
erle |
i can not promise it, but the fact is, this constant is entirely arbitrary |
| 22:06 |
DeepThgt |
thanks, i at least now have 1 solution i know how to implement |
| 22:06 |
erle |
the only purpose is to keep away far from the bounds of 16 bit signed integer space |
| 22:06 |
erle |
see the comment in that file |
| 22:06 |
DeepThgt |
i wonder what would happen if i then loaded the world on a standard build |
| 22:07 |
DeepThgt |
cause i just built a buncha stuff at the bottom :| |
| 22:07 |
DeepThgt |
i can always find out |
| 22:07 |
DeepThgt |
thats why we have dev and staging lol |
| 22:07 |
DeepThgt |
problem with this solution is downtime |
| 22:08 |
DeepThgt |
well, wouldnt be too bad |
| 22:08 |
DeepThgt |
hrm\ |
| 22:08 |
erle |
look i only suggested it because it is the funniest one |
| 22:08 |
erle |
the arguably less funnier solution is to have some server side mod that access that inventory |
| 22:09 |
erle |
and either spits it out into the visible part of the map or just shows you a formspec |
| 22:09 |
erle |
though hmm |
| 22:09 |
erle |
i think there is a boundary over which nodes can not have meta |
| 22:09 |
erle |
i wonder where it was |
| 22:09 |
erle |
i *think* it was further out |
| 22:09 |
erle |
but i am not entirely sure |
| 22:09 |
erle |
how do you even know the bones got placed there in the first place? log message? |
| 22:09 |
muurkha |
erle: maybe some vignetting and sepia would be cool |
| 22:10 |
DeepThgt |
ya logged |
| 22:10 |
DeepThgt |
2023-11-11 21:04:37: ACTION[Server]: NASTYNATE dies at (-2462,-30913,3653). Bones placed |
| 22:10 |
erle |
muurkha just because i have a toy camera i don't have to make it less useful right? |
| 22:10 |
muurkha |
Mina points out that a timer for a group photo would be a cool feature too :) |
| 22:10 |
erle |
muurkha with effects i mean more practical ones like depth of field |
| 22:11 |
erle |
i shall make xcam an API i guess |
| 22:11 |
erle |
so you can make all your things yourself |
| 22:11 |
erle |
like timers and stuff |
| 22:11 |
erle |
i still have to figure out how to raycast so much without hanging the server for a moment |
| 22:12 |
DeepThgt |
.i wish the 3ds was more powerful |
| 22:12 |
DeepThgt |
id snag one and try to port to it |
| 22:12 |
DeepThgt |
then see if i could get stereoscopy working |
| 22:12 |
DeepThgt |
but the 3ds is weak |
| 22:12 |
muurkha |
vignetting and sepia are very practical inded |
| 22:12 |
erle |
muurkha in what ways? |
| 22:13 |
muurkha |
I was reluctant to suggest depth of field for computational reasons |
| 22:13 |
muurkha |
I mean sepia is just multiplying the pixels by a constant rgb vector for example |
| 22:14 |
muurkha |
and vignetting is almost as cheap |
| 22:14 |
erle |
i already do the depth of field for supersampling reasons anyway |
| 22:14 |
erle |
it is just very subtle |
| 22:14 |
erle |
or maybe i do it wrong |
| 22:18 |
|
calcul0n joined #minetest |
| 22:26 |
erle |
muurkha i want honest materials, that is the thing |
| 22:26 |
erle |
muurkha every ornament has to do work to not be considered a crime |
| 22:34 |
|
acarrico1 joined #minetest |
| 22:38 |
muurkha |
erle: oh, I like ornament |
| 22:40 |
erle |
muurkha it should not be pointless, that's my assertion |
| 22:41 |
erle |
for example, i see a difference in category between the dithering shader that reduces banding and the waving leaves shader |
| 22:41 |
erle |
the latter looks funny and makes holes in geometry |
| 22:41 |
erle |
(does it still do that?) |
| 22:42 |
muurkha |
what would it mean for ornament to have a point? |
| 22:42 |
muurkha |
it affects your perception, that's what it's supposed to do |
| 22:43 |
muurkha |
people have a subjective experience of it, which they may like or dislike |
| 22:44 |
muurkha |
what higher purpose can something in a video game or in an art photo serve, other than provoking such a subjective experience? |
| 22:46 |
|
acarrico1 left #minetest |
| 22:56 |
|
YuGiOhJCJ joined #minetest |
| 23:32 |
|
panwolfram joined #minetest |
