tynet-lichat/shirakumo - IRC Chatlog
Search
Thursday, 30th of March 2023, 8:22:21 UTC
8:27:49
shinmera
https://github.com/Shirakumo/trial/pull/39
8:27:54
shinmera
As always, reviews welcome.
8:44:49
shinmera
mood: selwyn: |3b|: gingerale: pinging the crew
9:40:28
selwyn
no time to look at it in detail rn
9:40:46
selwyn
but i notice from the commit logs you have your own convex-mesh
9:41:18
shinmera
yeah, I did tell you about that :y
10:23:32
selwyn
ill see about using ik for some hand poses in vr
10:23:45
selwyn
but i don't anticipate it to be really useful in the near future
10:24:04
selwyn
how do you even test something like this
10:26:10
|3b|
your shared-initialize for ik-solver looks like it sets the reader instead of the accessor for joints?
10:26:13
shinmera
you construct a skeleton, set some targets, see it looks ok.
10:26:34
shinmera
|3b|: there's a defmethod later down
10:28:03
|3b|
and does REPLACEing constraints there drop elements if the vector in the slot is shorter (or still #() )?
10:28:38
shinmera
it'll drop elements, yeah.
10:28:55
|3b|
describe-object does TERPRI on wrong stream too
10:38:22
|3b|
when (<= 0 last) in solve-for vec3 ccd-solver possibly should be < ?
10:38:56
|3b|
ACTION can't count though, so maybe not
10:40:01
|3b|
ACTION tries to math a few more times, and still thinks it should be <, so maybe
10:40:22
|3b|
(not correctness, just wasting time testing and then doing nothing a bunch of times in a loop)
10:41:27
|3b|
alternately, maybe it should do something for the case with 1 element in chain... not sure what that would actually mean
10:42:46
shinmera
There's no harm to it being <=, but yeah.
10:44:49
|3b|
ACTION also wonders if threshold squared is big enough to be useful when comparing distances in single float, if it started at 0.00001 before squaring
10:45:27
shinmera
I just used the values that the book had in there. It seems to solve fine :shrug:
12:38:58
shinmera
Constraints seem to be working fine.
12:39:07
shinmera
(after a fix to 3D-quaternions)
12:52:54
shinmera
A bit stuck on the last code feature, which is the non-local targets.
12:53:19
shinmera
The IK system doesn't "know" about the entity, so it doesn't know how to translate the target into the local space
12:53:32
shinmera
I could just add another slot but that feels... messy?
14:02:26
shinmera
I think I'll add a transform slot which can be shared with the entity (if there is one)
17:53:06
shinmera
Okey, docs have been added.
17:59:27
selwyn
not sure how i feel about these openxr structure pointer chains
17:59:37
selwyn
will they cause problems?
18:59:35
shinmera
Aight with IK done, I guess next step would beeeeee
18:59:49
shinmera
a) dual quaternion skinning
18:59:57
shinmera
b) raycasts over 3d-spaces
19:00:22
shinmera
c) some kinda rendering system
19:00:38
shinmera
d) figuring out how to export custom metadata from blender to gltf
19:01:00
shinmera
e) fixing libmixed's spatial audio
19:01:24
shinmera
f) fixing the particle system
19:01:50
shinmera
g) figuring out a splatmap system
19:02:08
shinmera
h) other stuff I can't think of right now
19:08:05
shinmera
i) writing a generic IO system to base SF3 on, which can parse from/to memory, octet-array, stream, file-stream, and also do so ain "pointer mode" rather than having to copy all the data
19:08:18
shinmera
j) continue the 3d-* rewrite
19:18:44
shinmera
really don't know what to do
Thursday, 30th of March 2023, 20:22:21 UTC