| Time |
Nick |
Message |
| 00:13 |
|
benrob0329 joined #minetest-dev |
| 00:33 |
|
proller joined #minetest-dev |
| 01:11 |
|
Seirdy left #minetest-dev |
| 04:00 |
|
MTDiscord joined #minetest-dev |
| 05:10 |
MTDiscord |
<Liso> Yes, i did some vtune analysis. I can make a more in deep pass and post the results in forums, GitHub, or whatever. But mostly the clientmap update and draw is the bottleneck. Also I did some renderdoc passes and it's the same, thousand of small drawcalls |
| 05:14 |
MTDiscord |
<Liso> But it doesn't mean the shadow shaders are ok and optimized. I'm not even close a GFX expert, this is just for fun. |
| 05:36 |
MTDiscord |
<Liso> Related to this. I have a question to celeron55. I saw you did a try to use texture atlas, but it was removed. I couldn't find the reason in GitHub. What were the reasons to remove it? |
| 05:49 |
MTDiscord |
<Liso> Forget about it, it was in 2013 and by other person. Sorry |
| 06:08 |
celeron55 |
the texture atlas was buggy and other optimizations made it largely irrelevant back then, i suppose because overall less textures were used and rendering distances were smaller |
| 06:13 |
celeron55 |
the tilespec and other classes should still be able to carry texture atlas information with them if some kind of atlasing is added |
| 06:14 |
celeron55 |
the atlas back then was a global atlas generated at startup, a new atlas would probably be something more dynamic |
| 06:14 |
celeron55 |
hecks suggested to have an atlas per mapblock |
| 06:16 |
celeron55 |
those discussions were in 2020-12 if someone wants to browse the logs |
| 06:33 |
|
z812_ joined #minetest-dev |
| 06:38 |
|
z812_ joined #minetest-dev |
| 06:38 |
MTDiscord |
<Liso> Ok ,tnx! ? |
| 06:39 |
MTDiscord |
<Liso> Yes, I saw that discussion. and I feel something can be done ? |
| 07:24 |
|
Noisytoot joined #minetest-dev |
| 07:25 |
|
proller joined #minetest-dev |
| 07:50 |
|
Noisytoot joined #minetest-dev |
| 08:00 |
|
ShadowNinja joined #minetest-dev |
| 08:22 |
|
twoelk joined #minetest-dev |
| 08:26 |
|
hlqkj joined #minetest-dev |
| 08:59 |
|
proller joined #minetest-dev |
| 09:43 |
|
proller joined #minetest-dev |
| 11:39 |
|
hlqkj joined #minetest-dev |
| 13:02 |
|
absurb joined #minetest-dev |
| 13:51 |
|
Fleckenstein joined #minetest-dev |
| 14:06 |
|
silver_est joined #minetest-dev |
| 14:17 |
MTDiscord |
<Jordach> I should point out adding a depth buffer to Irrlicht would solve many problems |
| 14:17 |
MTDiscord |
<Jordach> Because depth buffer shadowing is even cheaper |
| 14:23 |
sfan5 |
Irrlicht has a depth buffer |
| 14:43 |
MTDiscord |
<Jordach> which MT dont use |
| 14:44 |
celeron55 |
of course it's used |
| 14:44 |
|
proller joined #minetest-dev |
| 14:45 |
celeron55 |
MT and irrlicht quite fully depend on it |
| 14:52 |
|
proller joined #minetest-dev |
| 14:53 |
|
benrob0329 joined #minetest-dev |
| 15:01 |
MTDiscord |
<appguru> it is used, but not in a second stage |
| 15:08 |
|
rubenwardy joined #minetest-dev |
| 15:13 |
MTDiscord |
<Jordach> things like reshade and render process hacks report that the z buffer reports the value of 1 of every pixel |
| 15:22 |
MTDiscord |
<Liso> I think jugde dredd is talking about depth pass, without color. And I'm not sure Irrlitch is capable to do it. All the examples I saw at forums use normal shaders |
| 15:23 |
MTDiscord |
<Jordach> what i'm implying is that Minetest discards the depth buffer after the 3d frame is rasterized |
| 15:25 |
|
Fixer joined #minetest-dev |
| 15:26 |
MTDiscord |
<Liso> Are you thinking about deferred rendering? |
| 15:30 |
MTDiscord |
<Yaya MNH48> note: "judge dredd" is jordach (only the username is passed through the bridge to IRC, not the nickname) |
| 15:30 |
celeron55 |
oh yeah it definitely isn't kept for anything else after the geometry drawcalls |
| 15:31 |
MTDiscord |
<Jordach> if that was kept post 3d render it has utility in the fact that depth of field and other similar camera trickery are then possible |
| 15:31 |
celeron55 |
if it's useful for something, then make it so |
| 15:31 |
MTDiscord |
<Jordach> shadows like liso's PR can also be done with a depth buffer stencil |
| 15:31 |
celeron55 |
now even irrlicht is modifiable to almost any extent so there are no excuses |
| 15:31 |
MTDiscord |
<Jordach> basically |
| 15:36 |
|
appguru joined #minetest-dev |
| 15:36 |
|
BrBridge joined #minetest-dev |
| 15:41 |
|
BrBridge joined #minetest-dev |
| 15:45 |
|
BrBridge joined #minetest-dev |
| 16:01 |
|
BrBridge joined #minetest-dev |
| 16:05 |
|
BrBridge joined #minetest-dev |
| 16:06 |
|
BrBridge joined #minetest-dev |
| 16:11 |
|
BrBridge joined #minetest-dev |
| 16:13 |
|
BrBridge joined #minetest-dev |
| 16:22 |
|
BrBridge joined #minetest-dev |
| 16:41 |
|
hlqkj joined #minetest-dev |
| 17:08 |
|
GreenXenith joined #minetest-dev |
| 18:29 |
|
I_am_6r1d joined #minetest-dev |
| 18:55 |
|
Noisytoot joined #minetest-dev |
| 19:21 |
|
BrBridge joined #minetest-dev |
| 19:38 |
|
olliy_ joined #minetest-dev |
| 20:05 |
|
BrBridge joined #minetest-dev |
| 20:51 |
|
bodqhrohro joined #minetest-dev |
| 21:13 |
|
MTDiscord1 joined #minetest-dev |
| 21:15 |
|
proller joined #minetest-dev |
| 21:15 |
|
djenkins__ joined #minetest-dev |
| 21:15 |
|
hlqkj joined #minetest-dev |
| 21:15 |
|
AntumDeluge joined #minetest-dev |
| 21:16 |
|
I_am_6r1d joined #minetest-dev |
| 21:16 |
|
hlqkj_ joined #minetest-dev |
| 22:05 |
|
Taoki joined #minetest-dev |
| 22:23 |
|
BakerPrime joined #minetest-dev |
| 22:24 |
|
BakerPrime joined #minetest-dev |
| 23:36 |
|
ShadowNinja joined #minetest-dev |
