Greatly improve matrix handling logic in collada code
This commit is contained in:
@@ -11,12 +11,11 @@
|
||||
# * * Collada tooling sucks and even LL is moving away from it
|
||||
# * * Ensuring LLMesh->Collada and LLMesh->GLTF conversion don't differ semantically is easy via assimp.
|
||||
|
||||
import collections
|
||||
import logging
|
||||
import os.path
|
||||
import secrets
|
||||
import statistics
|
||||
import sys
|
||||
from typing import Dict, List, Iterable, Optional
|
||||
from typing import Dict, List, Optional, Union, Sequence
|
||||
|
||||
import collada
|
||||
import collada.source
|
||||
@@ -25,13 +24,27 @@ from lxml import etree
|
||||
import numpy as np
|
||||
import transformations
|
||||
|
||||
from hippolyzer.lib.base.datatypes import Vector3
|
||||
from hippolyzer.lib.base.helpers import get_resource_filename
|
||||
from hippolyzer.lib.base.serialization import BufferReader
|
||||
from hippolyzer.lib.base.mesh import LLMeshSerializer, MeshAsset, positions_from_domain, SkinSegmentDict
|
||||
|
||||
LOG = logging.getLogger(__name__)
|
||||
DIR = os.path.dirname(os.path.realpath(__file__))
|
||||
|
||||
|
||||
def llsd_to_mat4(mat: Union[np.ndarray, Sequence[float]]) -> np.ndarray:
|
||||
return np.array(mat).reshape((4, 4), order='F')
|
||||
|
||||
|
||||
def mat4_to_llsd(mat: np.ndarray) -> List[float]:
|
||||
return list(mat.flatten(order='F'))
|
||||
|
||||
|
||||
def mat4_to_collada(mat: np.ndarray) -> np.ndarray:
|
||||
return mat.flatten(order='C')
|
||||
|
||||
|
||||
def mesh_to_collada(ll_mesh: MeshAsset, include_skin=True) -> collada.Collada:
|
||||
dae = collada.Collada()
|
||||
axis = collada.asset.UP_AXIS.Z_UP
|
||||
@@ -52,7 +65,7 @@ def llmesh_to_node(ll_mesh: MeshAsset, dae: collada.Collada, uniq=None,
|
||||
skin_seg = ll_mesh.segments.get('skin')
|
||||
bind_shape_matrix = None
|
||||
if include_skin and skin_seg:
|
||||
bind_shape_matrix = np.array(skin_seg["bind_shape_matrix"]).reshape((4, 4))
|
||||
bind_shape_matrix = llsd_to_mat4(skin_seg["bind_shape_matrix"])
|
||||
should_skin = True
|
||||
# Transform from the skin will be applied on the controller, not the node
|
||||
node_transform = np.identity(4)
|
||||
@@ -119,9 +132,8 @@ def llmesh_to_node(ll_mesh: MeshAsset, dae: collada.Collada, uniq=None,
|
||||
accessor.set('source', f"#{accessor.get('source')}")
|
||||
|
||||
flattened_bind_poses = []
|
||||
# LLMesh matrices are row-major, convert to col-major for Collada.
|
||||
for bind_pose in skin_seg['inverse_bind_matrix']:
|
||||
flattened_bind_poses.append(np.array(bind_pose).reshape((4, 4)).flatten('F'))
|
||||
flattened_bind_poses.append(mat4_to_collada(llsd_to_mat4(bind_pose)))
|
||||
flattened_bind_poses = np.array(flattened_bind_poses)
|
||||
inv_bind_source = _create_mat4_source(f"bind-poses{sub_uniq}", flattened_bind_poses, "TRANSFORM")
|
||||
|
||||
@@ -142,7 +154,7 @@ def llmesh_to_node(ll_mesh: MeshAsset, dae: collada.Collada, uniq=None,
|
||||
# in SL, with their own distinct sets of weights and vertex data.
|
||||
controller_node = E.controller(
|
||||
E.skin(
|
||||
E.bind_shape_matrix(' '.join(str(x) for x in bind_shape_matrix.flatten('F'))),
|
||||
E.bind_shape_matrix(' '.join(str(x) for x in mat4_to_collada(bind_shape_matrix))),
|
||||
joints_source.xmlnode,
|
||||
inv_bind_source.xmlnode,
|
||||
weights_source.xmlnode,
|
||||
@@ -173,7 +185,7 @@ def llmesh_to_node(ll_mesh: MeshAsset, dae: collada.Collada, uniq=None,
|
||||
node = collada.scene.Node(
|
||||
node_name,
|
||||
children=geom_nodes,
|
||||
transforms=[collada.scene.MatrixTransform(np.array(node_transform.flatten('F')))],
|
||||
transforms=[collada.scene.MatrixTransform(mat4_to_collada(node_transform))],
|
||||
)
|
||||
if should_skin:
|
||||
# We need a skeleton per _mesh asset_ because you could have incongruous skeletons
|
||||
@@ -208,7 +220,8 @@ def transform_skeleton(skel_root: etree.ElementBase, dae: collada.Collada, skin_
|
||||
joint_nodes[skel_node.get('name')] = collada.scene.Node.load(dae, skel_node, {})
|
||||
for joint_name, matrix in zip(skin_seg['joint_names'], skin_seg.get('alt_inverse_bind_matrix', [])):
|
||||
joint_node = joint_nodes[joint_name]
|
||||
joint_node.matrix = np.array(matrix).reshape((4, 4)).flatten('F')
|
||||
joint_decomp = transformations.decompose_matrix(llsd_to_mat4(matrix))
|
||||
joint_node.matrix = mat4_to_collada(transformations.compose_matrix(translate=joint_decomp[3]))
|
||||
# Update the underlying XML element with the new transform matrix
|
||||
joint_node.save()
|
||||
|
||||
@@ -251,7 +264,7 @@ def _create_mat4_source(name: str, data: np.ndarray, semantic: str):
|
||||
|
||||
def fix_weird_bind_matrices(skin_seg: SkinSegmentDict):
|
||||
"""
|
||||
Fix weird-looking bind matrices to have normal scaling
|
||||
Fix weird-looking bind matrices to have normal scaling and rotations
|
||||
|
||||
Not sure why these even happen (weird mesh authoring programs?)
|
||||
Sometimes get enormous inverse bind matrices (each component 10k+) and tiny
|
||||
@@ -259,38 +272,38 @@ def fix_weird_bind_matrices(skin_seg: SkinSegmentDict):
|
||||
with weird scales and tries to set them to what they "should" be without
|
||||
the weird inverted scaling.
|
||||
"""
|
||||
axis_counters = [collections.Counter() for _ in range(3)]
|
||||
for joint_inv in skin_seg['inverse_bind_matrix']:
|
||||
joint_mat = np.array(joint_inv).reshape((4, 4))
|
||||
joint_scale = transformations.decompose_matrix(joint_mat)[0]
|
||||
for axis_counter, axis_val in zip(axis_counters, joint_scale):
|
||||
axis_counter[axis_val] += 1
|
||||
most_common_inv_scale = []
|
||||
for axis_counter in axis_counters:
|
||||
most_common_inv_scale.append(axis_counter.most_common(1)[0][0])
|
||||
scale_fixup = Vector3(1, 1, 1)
|
||||
angle_fixup = Vector3(0, 0, 0)
|
||||
have_fixups = False
|
||||
# Totally non-scientific method of detecting odd bind matrices based on squinting very,
|
||||
# very hard at a random sample of assets.
|
||||
for joint_name, joint_inv in zip(skin_seg['joint_names'], skin_seg['inverse_bind_matrix']):
|
||||
if not joint_name.startswith("m"):
|
||||
# We can't make very good guesses based on collision volume scales and rotations,
|
||||
# skip anything but the "m" joints.
|
||||
continue
|
||||
joint_mat = llsd_to_mat4(joint_inv)
|
||||
joint_scale, _, joint_angle, _, _ = transformations.decompose_matrix(joint_mat)
|
||||
# If the scale component of an mJointName joint isn't roughly <1,1,1>, we likely have
|
||||
# scaling applied to the inverse bind matrices rather than the bind matrix. Figure out
|
||||
# what the fixup should be so that we can reverse it.
|
||||
if abs(3.0 - sum(joint_scale)) > 0.5:
|
||||
scale_fixup = Vector3(1, 1, 1) / Vector3(*joint_scale)
|
||||
have_fixups = True
|
||||
# I wouldn't expect mJointName joints to be rotated at all in their inverse bind matrices.
|
||||
# Is this a rotation that should've been applied to the bind shape matrix instead?
|
||||
# In any event, all joints are likely rotated by this amount, so calculate the inverse.
|
||||
if abs(sum(joint_angle)) > 0.05:
|
||||
angle_fixup = -Vector3(*joint_angle)
|
||||
have_fixups = True
|
||||
|
||||
if abs(1.0 - statistics.fmean(most_common_inv_scale)) > 1.0:
|
||||
if have_fixups:
|
||||
LOG.warning("Detected weird matrices in mesh!", scale_fixup, angle_fixup)
|
||||
# The magnitude of the scales in the inverse bind matrices look very strange.
|
||||
# The bind matrix itself is probably messed up as well, try to fix it.
|
||||
skin_seg['bind_shape_matrix'] = fix_llsd_matrix_scale(skin_seg['bind_shape_matrix'], most_common_inv_scale)
|
||||
if joint_positions := skin_seg.get('alt_inverse_bind_matrix', None):
|
||||
fix_matrix_list_scale(joint_positions, most_common_inv_scale)
|
||||
rev_scale = tuple(1.0 / x for x in most_common_inv_scale)
|
||||
fix_matrix_list_scale(skin_seg['inverse_bind_matrix'], rev_scale)
|
||||
|
||||
|
||||
def fix_matrix_list_scale(source: List[List[float]], scale_fixup: Iterable[float]):
|
||||
for i, alt_inv_matrix in enumerate(source):
|
||||
source[i] = fix_llsd_matrix_scale(alt_inv_matrix, scale_fixup)
|
||||
|
||||
|
||||
def fix_llsd_matrix_scale(source: List[float], scale_fixup: Iterable[float]):
|
||||
matrix = np.array(source).reshape((4, 4))
|
||||
decomposed = list(transformations.decompose_matrix(matrix))
|
||||
# Need to handle both the scale and translation matrices
|
||||
for idx in (0, 3):
|
||||
decomposed[idx] = tuple(x * y for x, y in zip(decomposed[idx], scale_fixup))
|
||||
return list(transformations.compose_matrix(*decomposed).flatten('C'))
|
||||
# TODO: DON'T MESS WITH INVERSE TRANSLATION!!!! Only bind shape gets its translation scaled.
|
||||
# TODO: put this back in, the previous logic was totally wrong-headed..
|
||||
pass
|
||||
|
||||
|
||||
def main():
|
||||
|
||||
Reference in New Issue
Block a user