Time |
Nick |
Message |
00:36 |
|
paramat joined #minetest-dev |
00:36 |
|
kaeza joined #minetest-dev |
00:43 |
paramat |
RBA changing BS will have to be a MTHD thing, it won't be allowed n MT. i agree camera offset distance is too small, perhaps 1000?, i might make a PR. camera offset is not the main cause of stuttering, that happens all the time not every 200 nodes |
00:48 |
RealBadAngel |
paramat, im not quite sure if it will be worth it, i keep getting mixed results |
00:48 |
RealBadAngel |
but offset indeed could be changed |
00:49 |
paramat |
ok |
00:49 |
RealBadAngel |
i will play with the issue later on, atm i dont have time for something unrelated to deferred rendering |
00:59 |
Fixer |
stutter happens when there is meshgen spike // Captain Obvious |
00:59 |
Fixer |
coincidence? don't think so |
01:04 |
RealBadAngel |
well, whatever is the reason of updating meshes, it has to be limited, its a pure nonsense having mesh updated because time of day changed, player moved a bit and crossed camera offset change line, etc |
01:04 |
|
jordan4ibanez joined #minetest-dev |
01:04 |
RealBadAngel |
makin mesh is heavy duty stuff and have to be avoided at all costs |
01:07 |
RealBadAngel |
atm there are 3 reasons left that are making mesh dirty -> time of day (lighting will fix it), cracks (have to be removed same way as halo), and camera offset (not quite sure how to get rid of it completely) |
01:09 |
RealBadAngel |
well, 4. changing any node meta is causing complete up to 7 mapblock meshes regen |
01:10 |
RealBadAngel |
with all those 4 issues solved we can enable VBO and go further |
01:10 |
RealBadAngel |
at least thats the plan ;) |
01:54 |
VanessaE |
RealBadAngel: the camera offset can only be dealt with properly by splitting the world into many (thousands of) irrlicht Scene Nodes |
01:54 |
VanessaE |
it's there because the map is one big scene node, and the camera gets glitchy once you get more than a few hundred meters from the camera's origin |
01:55 |
VanessaE |
or rather, the rendering precision becomes a problem and it LOOKS glitchy. |
01:55 |
paramat |
would 1000 nodes be okay for the offset distance? it should be maximised |
01:55 |
VanessaE |
paramat: nope. |
01:56 |
VanessaE |
even 1000 is enough to see the glitch, if you're looking for it |
01:56 |
VanessaE |
I don't recall why 200 in particular was chosen, though |
01:56 |
RealBadAngel |
VanessaE, i was looking hard with offset set to 2k and couldnt find any |
01:57 |
VanessaE |
it probably depends partly on the video driver and irrlicht version |
01:58 |
VanessaE |
either way, there is only one proper way to fix it. |
01:58 |
VanessaE |
the proposal at the time was to make each Minetest mapblock a scene node. |
02:14 |
paramat |
i think 200 is a little too low, as it apparently causes mesh updates |
02:16 |
paramat |
need to discuss with nore though |
02:25 |
RealBadAngel |
not really mesh updates but shitload of operations on vertices adding offseted pos to them |
02:27 |
paramat |
ok |
02:29 |
RealBadAngel |
if we consider using VBO it will also mean that mesh have to be reuploaded to GPU VRAM |
02:52 |
VanessaE |
RealBadAngel: but that doesn't answer the scene node problem. |
02:52 |
VanessaE |
adjusting the camera offset is just delaying the inevitable. |
02:54 |
RealBadAngel |
im trying to figure out best solution |
02:54 |
RealBadAngel |
and im not sure about scene nodes idea |
02:54 |
* VanessaE |
shrugs |
02:55 |
VanessaE |
I leave it to you guys to find the best way |
02:56 |
RealBadAngel |
fuck, i wrote perfect code for something, then was looking for whole day why it is not working at all |
02:56 |
RealBadAngel |
damn misplaced bracket |
02:57 |
RealBadAngel |
it threw a bit of code out of scope lol |
03:18 |
|
ssieb joined #minetest-dev |
04:44 |
RealBadAngel |
kahrl, here? |
05:05 |
|
Sockbat joined #minetest-dev |
06:36 |
|
Hunterz joined #minetest-dev |
06:48 |
sofar |
why is my character.b3d rotated 90 degrees when I export? |
06:48 |
sofar |
I don't see any b3d export options for axis |
06:57 |
sofar |
anyone know what's the proper B3Dexport.py script to use? |
06:57 |
paramat |
no idea |
06:58 |
sofar |
I tried a different one but that one plainly is broken |
07:00 |
sofar |
there's a small problem with character.b3d that I want to fix |
07:00 |
sofar |
I've managed to fix it |
07:00 |
sofar |
but now I can't export it properly back :( |
07:02 |
paramat |
btw i will merge game#789 game#833 later |
07:02 |
ShadowBot |
https://github.com/minetest/minetest_game/issues/789 -- New mesh door models, and extensive door API by sofar |
07:02 |
ShadowBot |
https://github.com/minetest/minetest_game/issues/833 -- Convert furnace ABM to node timer by sofar |
07:02 |
sofar |
neat |
07:10 |
sofar |
hmmm kilbith was the person who made b3d files out of character.x |
07:11 |
sofar |
I wonder if he has an exporter that doesn't futz the axes |
07:18 |
|
paramat left #minetest-dev |
07:31 |
|
Obani joined #minetest-dev |
08:02 |
|
Darcidride joined #minetest-dev |
08:26 |
|
paramat joined #minetest-dev |
08:29 |
|
Krock joined #minetest-dev |
08:33 |
|
nrzkt joined #minetest-dev |
10:27 |
|
jin_xi joined #minetest-dev |
11:13 |
|
dfelinto joined #minetest-dev |
11:32 |
RealBadAngel |
#3695 |
11:32 |
ShadowBot |
https://github.com/minetest/minetest/issues/3695 -- Use vertices with tangents only when its needed. by RealBadAngel |
11:33 |
RealBadAngel |
^^ some mapblock mesh rework |
11:34 |
* RealBadAngel |
yawns |
12:07 |
|
Ardonel joined #minetest-dev |
12:20 |
|
Ardonel joined #minetest-dev |
12:23 |
|
Amaz joined #minetest-dev |
12:23 |
|
Calinou joined #minetest-dev |
12:26 |
|
nrzkt joined #minetest-dev |
12:35 |
|
Fixer joined #minetest-dev |
13:02 |
|
hmmmm joined #minetest-dev |
13:05 |
|
Obani joined #minetest-dev |
13:32 |
|
turtleman joined #minetest-dev |
13:56 |
|
Player_2 joined #minetest-dev |
14:01 |
|
H-H-H joined #minetest-dev |
14:04 |
|
leat joined #minetest-dev |
14:08 |
|
STHGOM joined #minetest-dev |
14:21 |
|
kaadmy joined #minetest-dev |
14:21 |
|
ShadowBot joined #minetest-dev |
14:22 |
kaadmy |
sofar: i use one that someone pointed me towards, it works fine |
14:22 |
Taoki |
RealBadAngel: Any estimation as to when we can hope for bloom and DoF to make it in master? Just waiting for confirmation that they will be there... especially now that I'm working on a Minetest game designed to have fancy graphics and shaders too. |
14:22 |
kaadmy |
b3d exporter* |
14:23 |
RealBadAngel |
Taoki, dunno, i still have lotsa things to code |
14:23 |
Taoki |
ok |
14:23 |
Taoki |
Hope 3D plantlike / torchlike is still among them BTW. I think that's a more urgent visual improvement anyway. |
14:24 |
Taoki |
IIRC you got it working with waving plants last time, so hopefully not much is left there. |
14:25 |
kaadmy |
Taoki: would having the *thin* drawtypes like signlike, torchlike, plantlike extruded look better? |
14:25 |
Taoki |
kaadmy: There's already screenshots and videos of it. They certainly do! At least I think so. |
14:25 |
kaadmy |
in master? |
14:25 |
RealBadAngel |
kaadmy, having flat torches is lookin ANY good to you? |
14:26 |
Taoki |
Not in master yet, but it was coded. |
14:26 |
Taoki |
It going i master is my point :P |
14:26 |
kaadmy |
RealBadAngel: flat torches look retarded :) |
14:29 |
RealBadAngel |
same as plants |
14:30 |
Taoki |
Yep |
14:58 |
|
Darcidride joined #minetest-dev |
15:07 |
|
nrzkt joined #minetest-dev |
15:19 |
|
Calinou joined #minetest-dev |
15:24 |
|
kaeza joined #minetest-dev |
15:33 |
|
Taoki joined #minetest-dev |
15:36 |
|
Hunterz joined #minetest-dev |
15:48 |
|
Gael-de-Sailly joined #minetest-dev |
15:55 |
Fixer |
RealBadAngel, will check it out |
15:55 |
RealBadAngel |
Fixer, ok, let me know |
15:58 |
RealBadAngel |
funny, i just got an attack of stuttering, it was starting a minute or so after starting game, so even walking was choking. played a bit with camera far plane to see if that would help |
15:59 |
Fixer |
and? |
15:59 |
RealBadAngel |
it looked like it helped, so reverted all the changes to see if stuttering will back |
15:59 |
RealBadAngel |
and i cant reproduce stuttering anymore lol |
16:00 |
Fixer |
RealBadAngel, yes, that's the problem, one time i was having very baad stutter with shaders disabled, next day - gone o_0 |
16:00 |
RealBadAngel |
fuck me walking if i know whats the reason |
16:00 |
RealBadAngel |
weather on mars or what? |
16:01 |
Fixer |
mm? i get stutter when new blocks are generated |
16:01 |
Fixer |
simple |
16:01 |
RealBadAngel |
no that not a reason |
16:01 |
Fixer |
est31 said it is uploading to gpu |
16:01 |
RealBadAngel |
atm im walkin around, with debug code to control whats going on with mapblock meshes |
16:02 |
RealBadAngel |
cant see any difference |
16:02 |
Fixer |
walk in a forests %) |
16:02 |
RealBadAngel |
from time to time some meshes got updates |
16:02 |
RealBadAngel |
i do |
16:02 |
RealBadAngel |
propably because of growing grasses etc |
16:02 |
Fixer |
stutter in multiplayer is also present, that is funny |
16:03 |
Fixer |
so it is not mapgen |
16:03 |
|
blaze joined #minetest-dev |
16:03 |
Fixer |
at least partially |
16:03 |
RealBadAngel |
it seems to be tottally random |
16:04 |
RealBadAngel |
i was able to observe it for 15 minutes or so, many client restarts with different settings, was trying to determine whats the reason |
16:04 |
RealBadAngel |
now its completely gone |
16:04 |
Fixer |
your videocard has clock frequency autochange support? |
16:05 |
Fixer |
RealBadAngel, https://i.imgur.com/OdTgVnF.png |
16:06 |
Fixer |
RealBadAngel, three minetest runs |
16:06 |
RealBadAngel |
idk |
16:06 |
RealBadAngel |
havent played with gpu settings |
16:06 |
Fixer |
RealBadAngel, sometimes GPU/MEM clocks drop from 1000 to 750 or even 150, for very short time, few times, not sure if it is the cause |
16:08 |
Fixer |
RealBadAngel, try walking on this test world -> https://github.com/minetest/minetest/files/101456/Just.Test.zip |
16:08 |
RealBadAngel |
if you can observe stuttering check the clocks then |
16:08 |
RealBadAngel |
if thats related |
16:24 |
|
Alduin_ joined #minetest-dev |
16:25 |
Fixer |
RealBadAngel, looks like it is not related |
16:28 |
Fixer |
hold on |
16:28 |
RealBadAngel |
can anybody tell what version of irrlicht travis is using? |
16:28 |
RealBadAngel |
error: ‘EGDS_BUTTON_PRESSED_IMAGE_OFFSET_Y’ is not a member of ‘irr::gui’ |
16:29 |
Fixer |
RealBadAngel, on heavy world, it can stutter when clocks go down to 150MHz for some reason, but there are stutters at FULL clock |
16:29 |
Fixer |
hmmm |
16:29 |
RealBadAngel |
https://travis-ci.org/minetest/minetest/builds/108787153 |
16:29 |
RealBadAngel |
Fixer, so thats not related |
16:30 |
Fixer |
RealBadAngel, hold on |
16:30 |
Fixer |
RealBadAngel, it is partially related |
16:30 |
Fixer |
RealBadAngel, if stutter is really deep it is clocks going to min for 1 sec or so |
16:31 |
RealBadAngel |
Fixer, so related or not? |
16:32 |
Fixer |
RealBadAngel, only for some deep stutters, it is partially related, some stutters are caused by frequencies some are not |
16:36 |
Fixer |
RealBadAngel, pff, why on earth it lower the frequencies when i load heavy MT scene with 30 fps :( |
16:37 |
RealBadAngel |
theres some auto tune code in mt |
16:38 |
RealBadAngel |
i wonder if it doesnt work the opposite as it should |
16:38 |
RealBadAngel |
first of all, why the hell camera far plane is not set when lowering view range? |
16:39 |
RealBadAngel |
and why fog doesnt match view range |
16:54 |
Fixer |
RealBadAngel, clock powersaving can cause some stutters, but not a whole lot to account for everything (most of the time MEM clock is at 75% (lol), GPU is at 100%) |
16:55 |
Fixer |
RealBadAngel, also, some time clocks going down does not cause stutter at all |
16:55 |
Fixer |
RealBadAngel, lots of time actually |
16:55 |
Fixer |
RealBadAngel, so it is not powersaving problem only, other sources are at work |
17:15 |
|
VirusJones joined #minetest-dev |
17:24 |
RealBadAngel |
Fixer, have you tested 3695? |
17:27 |
Fixer |
RealBadAngel, just walked, let me eat something, i will test deeper |
17:27 |
RealBadAngel |
sure |
17:28 |
Fixer |
RealBadAngel, why on earth drop frequencies down to 15% level for very short time in heavy minetest scene? feels like source of the problem is MT, that reduces (?) gpu load for very short time, causing frequinces down and big stutter after |
17:29 |
|
ud1_ joined #minetest-dev |
17:29 |
Fixer |
and frequiences back to normal |
17:29 |
Fixer |
maximum 1 sec for this all |
17:29 |
RealBadAngel |
it may be this auto tune code |
17:31 |
RealBadAngel |
from quick look it works that way: if it takes too long, lets lower wanted fps |
17:31 |
RealBadAngel |
then gpu thinks theres less to do, lower clocks |
17:31 |
RealBadAngel |
then mt again lowers wanted fps |
17:32 |
RealBadAngel |
until it chockes itself and gpu is sitting idle laughing |
17:32 |
RealBadAngel |
pretty funny |
17:33 |
RealBadAngel |
also i like the code for trying to be smarter than hardware |
17:33 |
RealBadAngel |
far clipping plane is actually disabled no matter what |
17:34 |
RealBadAngel |
the code that sets it looks like it could set an actual value |
17:34 |
RealBadAngel |
if far plane could be further than 20000 |
17:35 |
RealBadAngel |
total absurd |
17:35 |
RealBadAngel |
so, no matter what far is always 2000 |
17:35 |
|
DFeniks joined #minetest-dev |
17:35 |
RealBadAngel |
20k i mean |
17:45 |
RealBadAngel |
Fixer, i will make a patch with manual only view range control so you can test it |
18:04 |
|
Darcidride joined #minetest-dev |
18:15 |
RealBadAngel |
Fixer, wanna test it? |
18:19 |
RealBadAngel |
im playing with it right now, it looks like it allows me to set viewing distance at which framerate wont drop below 60fps no matter whats visible |
18:20 |
RealBadAngel |
like shitload of trees with snow nodeboxes on them |
18:22 |
Fixer |
RealBadAngel, yes |
18:22 |
Fixer |
RealBadAngel, i need .diff |
18:23 |
RealBadAngel |
it will be easier, open camera.cpp please |
18:23 |
Fixer |
done |
18:24 |
RealBadAngel |
find void Camera::updateViewingRange(f32 frametime_in, f32 busytime_in) should be line 484 |
18:24 |
RealBadAngel |
and paste this at the start of the function |
18:24 |
Fixer |
yes |
18:24 |
RealBadAngel |
http://pastebin.com/sCedr3Jr |
18:27 |
Fixer |
RealBadAngel, like this? http://pastebin.com/raw/agDEK9YB |
18:28 |
RealBadAngel |
yeah |
18:35 |
RealBadAngel |
Fixer, and? |
18:37 |
Fixer |
RealBadAngel, compiling |
18:37 |
Fixer |
RealBadAngel, not so fast |
18:37 |
Fixer |
RealBadAngel, at 20% |
18:38 |
RealBadAngel |
i thought you have i3? |
18:38 |
Fixer |
RealBadAngel, 3695 fixed mesh lag issue for me, will test performance later |
18:38 |
Fixer |
RealBadAngel, i compile in Debian VM |
18:39 |
Fixer |
RealBadAngel, cross compilation for windows from VM with debian* |
18:39 |
Fixer |
RealBadAngel, it is 2x slower |
18:39 |
RealBadAngel |
aw lol |
18:39 |
Fixer |
it is much easier |
18:42 |
RealBadAngel |
no doubts, compilin on windows is pain in the ass |
18:43 |
Fixer |
i've compiled on it some years ago, on xp, with celeron script |
18:43 |
Fixer |
on msvc |
18:50 |
RealBadAngel |
i have a feeling that game runs much smoother with this patch |
18:50 |
Fixer |
100% |
18:50 |
Fixer |
i mean compiled just now |
18:51 |
Fixer |
15 min %) |
18:52 |
RealBadAngel |
with distance ~100 i can see fixed 60fps |
18:52 |
RealBadAngel |
no fuckin matter what |
18:53 |
Fixer |
starting, shaders enabled |
18:54 |
Fixer |
walking, will start monitoring program |
18:54 |
Fixer |
it feels smooth on just test, walking more |
18:56 |
Fixer |
RealBadAngel, woah |
18:58 |
Fixer |
RealBadAngel, i see fog fluctuation |
19:00 |
RealBadAngel |
fog would need to be fixed, so clouds |
19:01 |
Fixer |
RealBadAngel, it feels smoother, let my try next server |
19:04 |
|
kaeza joined #minetest-dev |
19:06 |
Fixer |
hmm |
19:07 |
Fixer |
RealBadAngel, very interesting, looking linuxgaming now |
19:07 |
Fixer |
RealBadAngel, much less stutters for me, strangely, more smooth |
19:10 |
Fixer |
RealBadAngel, linuxgaming seems slow, not all blocks are loaded, but smoothness is amazing when there are no activity |
19:10 |
Fixer |
super flat |
19:16 |
Fixer |
RealBadAngel, LinuxGaming looks smoother, few deep stutters, also, when packets arrieve, draw curve becames somewhat unsettled |
19:16 |
Fixer |
RealBadAngel, but when no net activity, walking in loaded chunks are very good |
19:18 |
|
ssieb joined #minetest-dev |
19:20 |
|
kaadmy joined #minetest-dev |
19:27 |
RealBadAngel |
fixer, so, overall how it feels? |
19:27 |
sofar |
what did you add RBA? |
19:27 |
RealBadAngel |
rather what ive trashed ;) |
19:28 |
Fixer |
RealBadAngel, smoother, but in singleplayer stutters are present, but when no activity it is goood |
19:28 |
Fixer |
RealBadAngel, doing more testsw |
19:28 |
RealBadAngel |
sofar, http://irc.minetest.ru/minetest-dev/2016-02-12#i_4531932 |
19:29 |
Fixer |
RealBadAngel, wait, do i need to specify _max range? |
19:29 |
RealBadAngel |
use +- |
19:31 |
Fixer |
RealBadAngel, hmm, removed back max range and now it feels much better, not sure why |
19:31 |
RealBadAngel |
ive set i menu also wanted fps = 60, min 60 |
19:31 |
RealBadAngel |
im not sure if it counts at all |
19:32 |
Fixer |
RealBadAngel, singleplayer looks seriously better |
19:34 |
Fixer |
RealBadAngel, feels really smooth and good most of the time, hmm |
19:35 |
Fixer |
RealBadAngel, i will do automated tests, need 20 min or so |
19:36 |
RealBadAngel |
when you start game with creating server instead of famous singleplayer, it works indeed better |
19:36 |
Fixer |
RealBadAngel, i start singleplayer as always |
19:38 |
RealBadAngel |
btw, when you now set a viewing distance, its a real one |
19:39 |
RealBadAngel |
no fancy gimmicks with drawing blocks or not, its a circle around you |
19:40 |
RealBadAngel |
thats why fog doesnt fit |
19:41 |
Amaz |
RealBadAngel, with the latest git I'm sometimes getting problems with node highlighting. The box sometimes, seemingly randomly, turns opaque for a random period of time... https://i.imgur.com/SUleY1t.png |
19:41 |
RealBadAngel |
before it was drawn on things theoretically on things outside viewing range |
19:41 |
|
steven2 joined #minetest-dev |
19:41 |
RealBadAngel |
Amaz, i know about it, havent found the reason for it yet |
19:42 |
Amaz |
Ah, okay |
19:42 |
Amaz |
Thanks! |
19:42 |
Fixer |
doing some test, need 20 min |
19:42 |
RealBadAngel |
ok |
20:08 |
|
Warr1024 joined #minetest-dev |
20:10 |
|
Warr1024 joined #minetest-dev |
20:19 |
|
kaeza joined #minetest-dev |
20:21 |
Fixer |
RealBadAngel, tested in no mapgen environment, some liquid flows were possible on a test, here my results |
20:21 |
Fixer |
RealBadAngel, https://i.imgur.com/194Yxxp.png |
20:22 |
Fixer |
RealBadAngel, in general, with your changes you get little performance boost, but stutters are the same |
20:23 |
Fixer |
RealBadAngel, as you can see both curves on graph #3 became one somewhere near stutter zone (which is at the end - highest spikes) |
20:23 |
ShadowBot |
https://github.com/minetest/minetest/issues/3 -- Furnace segfault |
20:24 |
* Fixer |
>_> ShadowBot |
20:25 |
|
iqualfragile joined #minetest-dev |
20:28 |
|
iqualfragile joined #minetest-dev |
20:31 |
|
iqualfragile joined #minetest-dev |
20:40 |
|
technics joined #minetest-dev |
21:41 |
Fixer |
RealBadAngel, this one with mapgen v6 test: https://i.imgur.com/zGKTue0.png |
21:42 |
Fixer |
RealBadAngel, very slight fps boost, but stutters are the same, so it is not fixing the stutters unfortunately |
22:33 |
|
Player-2 joined #minetest-dev |
22:47 |
|
iqualfragile joined #minetest-dev |
23:50 |
|
RealBadAngel joined #minetest-dev |
23:55 |
RealBadAngel |
Fixer, pity :( im quite suprised because it really does feel smoother |