| Time |
Nick |
Message |
| 00:28 |
|
YuGiOhJCJ joined #minetest |
| 00:35 |
|
v-rob joined #minetest |
| 01:14 |
|
smk joined #minetest |
| 01:18 |
|
sparky4 joined #minetest |
| 01:29 |
muurkha |
haha |
| 01:31 |
|
Noisytoot joined #minetest |
| 01:41 |
muurkha |
sparky4: I see you got back to your dorm |
| 01:44 |
|
Verticen joined #minetest |
| 02:17 |
|
sys4_ joined #minetest |
| 03:16 |
wsor4035 |
!tell wsor4035 test |
| 03:16 |
MinetestBot |
You can tell that to yourself |
| 03:16 |
wsor4035 |
good |
| 03:17 |
wsor4035 |
!tell shaft assuming that you agree to behave and learn from this, we will unban you one week from now for a second chance |
| 03:17 |
MinetestBot |
wsor4035: I'll pass that on when shaft is around |
| 03:17 |
|
Izaya joined #minetest |
| 04:00 |
|
MTDiscord joined #minetest |
| 04:19 |
|
est31 joined #minetest |
| 04:49 |
|
qqq joined #minetest |
| 05:47 |
|
TomTom joined #minetest |
| 06:12 |
|
est31 joined #minetest |
| 06:36 |
|
calcul0n joined #minetest |
| 08:03 |
copygirl |
So when I define an entity with use_texture_alpha enabled, for some reason it looks like there is still a cutoff at 50% alpha or something? Part of the model ends up being completely invisible, rather than semi-seethrough. |
| 08:04 |
copygirl |
https://i.imgur.com/hsq1tet.png |
| 08:08 |
copygirl |
Here is the full model visible with an opaque texture: https://imgur.com/Rk7gkim.png |
| 08:10 |
MTDiscord |
<rollerozxa> by texture_alpha enabled, do you mean you've set it to "blend"? |
| 08:11 |
copygirl |
That's for nodes. It complained when I tried to use "blend" for an entity. |
| 08:11 |
copygirl |
According to the API, entities have `use_texture_alpha` that's just a boolean. |
| 08:12 |
MTDiscord |
<rollerozxa> oh I see, you're talking about an entity not a node |
| 08:13 |
MTDiscord |
<rollerozxa> hm, I'm unsure if semitransparent textures are even possible in an entity then if use_texture_alpha wouldn't work for it |
| 08:16 |
copygirl |
It *does* work it's just cut off past a certain point for some reason. |
| 08:16 |
MTDiscord |
<rollerozxa> yes, sounds like alpha clipping behaviour |
| 08:18 |
|
est31 joined #minetest |
| 08:19 |
sfan5 |
in other terms: for entities you chave the choice between "blend" (true) and "clip" (false) |
| 08:19 |
sfan5 |
dunno how well semi-transparency even works however |
| 08:24 |
|
mrkubax10 joined #minetest |
| 08:28 |
|
sys4 joined #minetest |
| 08:32 |
|
amfl joined #minetest |
| 08:43 |
copygirl |
Oh it's a year-old bug? #10032 |
| 08:43 |
ShadowBot |
https://github.com/minetest/minetest/issues/10032 -- Entity translucency is still broken |
| 08:47 |
erle |
that sounds like an easy fix |
| 08:48 |
erle |
copygirl looks like a one-line-fix, how about you make a PR? |
| 08:49 |
erle |
like cut it off at 1/256 or so |
| 08:49 |
copygirl |
The cut off should be at 0 for alpha blended entities. |
| 08:50 |
copygirl |
I don't touch C++ though. |
| 08:50 |
erle |
it's a constant WHAT CAN POSSIBLY GO WRONG :D |
| 08:54 |
copygirl |
Me touching C++ is what could go wrong. |
| 08:54 |
copygirl |
I'm not touching game engine code besides my own. |
| 09:33 |
|
Thelie joined #minetest |
| 09:56 |
|
appguru joined #minetest |
| 10:10 |
MTDiscord |
<luatic> sfan5: in theory, semi-transparency works by rendering semitransparent triangles in the correct order (back to front) |
| 10:10 |
MTDiscord |
<luatic> in practice, we don't do that, but I think we render entities that are farther away sooner? (entities are distance sorted) |
| 10:13 |
MTDiscord |
<luatic> so semitransparency across different entities should work, but semitransparency across different faces of the same entity (as in sorcerykid's example), or across entities / nodes / particles won't work |
| 10:14 |
MTDiscord |
<luatic> (it gets even hairier if you want(ed) to properly deal with the "edge case" of intersecting triangles; then you'll have to split up triangles) |
| 10:15 |
MTDiscord |
<luatic> solving this efficiently is hard. perhaps the easiest solution would be some form of "order-independent transparency" (i think Multicraft does this?). implementing this inefficiently in software rendering is easy (basically go from a single texture + depth buffer to a sorted set of colors + depths per pixel), but I don't know how one would implement it efficiently in GLSL. |
| 10:17 |
MTDiscord |
<luatic> a naive "triangle sorting" approach isn't really an option; it would greatly complicate the code and probably be way too inefficient. for example you might have to break up mapblock meshes. |
| 10:21 |
sfan5 |
don't we have working semi-transparency for mapblocks already? why can't this solution be applied to entities? |
| 10:35 |
MTDiscord |
<luatic> sfan5: for some definition of "working", we do. but AFAIK there's (1) no mesh-interior semitransparency support (2) no support for particle-mapblock / entity-mapblock semitransparency. |
| 10:37 |
MTDiscord |
<luatic> IIRC what it did was (1) sort the mapblocks (as we already do for the entities), (2) change the meshgen to produce different meshes depending on which direction the mapblocks are viewed from, such that (some?) faces are properly sorted |
| 10:38 |
MTDiscord |
<luatic> what we'd have to do for semitransparent entities is effectively sort the triangles of the mesh based on their distance to the camera (z-sorting). special cases (like the cube drawtype) can get simpler solutions. |
| 10:40 |
sfan5 |
ah okay that makes sense |
| 10:55 |
|
Sobinec joined #minetest |
| 10:56 |
|
s20 joined #minetest |
| 11:21 |
|
Alnotz joined #minetest |
| 11:54 |
|
appguru joined #minetest |
| 12:17 |
|
fury999io joined #minetest |
| 12:35 |
|
definitelya joined #minetest |
| 12:42 |
|
Alnotz joined #minetest |
| 12:59 |
|
Sobinec joined #minetest |
| 13:12 |
|
jaca122 joined #minetest |
| 13:40 |
|
Thelie joined #minetest |
| 14:11 |
|
Thelie1 joined #minetest |
| 14:36 |
|
s20 joined #minetest |
| 15:06 |
MTDiscord |
<warr1024> "<copygirl> I'm not touching game engine code besides my own." <-- besides, having to depend on a fork of MT to make your mod work correctly would kinda defeat the whole purpose of using MT in the first place... |
| 15:09 |
erle |
Warr1024 well you can add a PR after that right |
| 15:15 |
|
jaca122 joined #minetest |
| 15:17 |
MTDiscord |
<warr1024> I suppose that PRs are still theoretically a thing, yeah. |
| 15:17 |
|
___nick___ joined #minetest |
| 15:17 |
|
fluxionary joined #minetest |
| 15:23 |
|
Izaya joined #minetest |
| 15:48 |
|
fluxionary joined #minetest |
| 16:05 |
copygirl |
@luatic: To be clear, I'm not running into a sorting issue here. NodeCore has few semi-transparent elements and my arrows would render on top of opaque blocks. Water is unlikely to be involved either. The renderer simply throws away pixels that have <50% transparency. |
| 16:14 |
|
fury999io left #minetest |
| 16:15 |
MTDiscord |
<warr1024> NodeCore has historically avoided use of semi-transparent anything specifically because of the known issues with them at the time I started the project, and I haven't started to rely heavily on the improvements since then either. |
| 16:25 |
|
s20_ joined #minetest |
| 16:37 |
|
erle joined #minetest |
| 16:58 |
|
appguru joined #minetest |
| 17:02 |
|
mrkubax10 joined #minetest |
| 17:05 |
|
Talkless joined #minetest |
| 17:15 |
|
s20_ joined #minetest |
| 17:26 |
|
v-rob joined #minetest |
| 18:02 |
|
appguru joined #minetest |
| 18:08 |
|
rubenwardy joined #minetest |
| 19:03 |
|
Noisytoot joined #minetest |
| 19:07 |
|
TheSilentLink joined #minetest |
| 19:11 |
|
___nick___ joined #minetest |
| 19:14 |
|
___nick___ joined #minetest |
| 19:44 |
|
TheSilentLink joined #minetest |
| 19:51 |
|
TheSilentLink joined #minetest |
| 20:48 |
|
Noisytoot joined #minetest |
| 21:15 |
|
calcul0n joined #minetest |
| 21:41 |
|
Izaya joined #minetest |
| 22:32 |
|
panwolfram joined #minetest |
| 22:37 |
|
v-rob joined #minetest |
| 23:16 |
|
Sokomine joined #minetest |
| 23:36 |
|
Sokomine joined #minetest |
