nan
dc000e50b1
- Added a new proposal for boundary smoothing in the weight cutout process, differentiating between Split edges and Outline smoothing requirements. - Introduced a new capability `cutout-boundary-smoothing` to control smoothing strength and default values. - Updated specifications to define the behavior of boundary-specific smoothing strength and default settings. - Created tasks to verify the implementation of boundary behavior and asset updates. - Developed comprehensive tests for depth symmetry and normal scaling to ensure expected behavior and performance. - Introduced new attributes for normal mapping to facilitate persistent face attributes sharing between depth symmetry and image materials. - Built a new node group for O Image Depth Symmetry, integrating the new smoothing capabilities and ensuring compatibility with existing workflows.
81 lines
3.2 KiB
Python
81 lines
3.2 KiB
Python
"""Identify connected calculations by their nearest geometry or shader consumers."""
|
|
|
|
from collections import defaultdict
|
|
from graphlib import TopologicalSorter
|
|
|
|
|
|
def calculation_blocks(nodes, edges):
|
|
"""Group fields with equal consumers, then attach their local interface reads."""
|
|
ids = {node: i for i, node in enumerate(nodes)}
|
|
parents = {i: set() for i in range(len(nodes))}
|
|
children = {i: set() for i in parents}
|
|
for edge in edges:
|
|
parents[ids[edge.to_node]].add(ids[edge.from_node])
|
|
children[ids[edge.from_node]].add(ids[edge.to_node])
|
|
order = list(TopologicalSorter(parents).static_order())
|
|
core = {
|
|
i
|
|
for i, n in enumerate(nodes)
|
|
if n.bl_idname == "NodeGroupOutput"
|
|
or n.bl_idname in {"ShaderNodeNormalMap", "ShaderNodeBump", "ShaderNodeDisplacement"}
|
|
or any(s.type in {"GEOMETRY", "SHADER"} and s.is_linked for s in n.outputs)
|
|
or getattr(n, "paired_output", None) is not None
|
|
}
|
|
inputs = {i for i, n in enumerate(nodes) if n.bl_idname == "NodeGroupInput"}
|
|
core -= inputs
|
|
fields = set(range(len(nodes))) - core - inputs
|
|
destinations = {}
|
|
for i in reversed(order):
|
|
destinations[i] = frozenset().union(
|
|
*(frozenset([j]) if j in core else destinations[j] for j in children[i])
|
|
)
|
|
groups = defaultdict(set)
|
|
for i in fields:
|
|
groups[destinations[i]].add(i)
|
|
blocks = []
|
|
for targets, members in groups.items():
|
|
pending = set(members)
|
|
while pending:
|
|
component = set()
|
|
queue = [min(pending)]
|
|
while queue:
|
|
i = queue.pop()
|
|
if i not in pending:
|
|
continue
|
|
pending.remove(i)
|
|
component.add(i)
|
|
queue.extend((parents[i] | children[i]) & pending)
|
|
target = min(targets) if targets else None
|
|
ports = list(
|
|
dict.fromkeys(
|
|
e.to_socket.name
|
|
for e in edges
|
|
if ids[e.from_node] in component and ids[e.to_node] == target
|
|
)
|
|
)
|
|
if ports:
|
|
title = nodes[target].name + " · " + ", ".join(ports)
|
|
else:
|
|
outlets = [i for i in component if children[i] - component]
|
|
source = (
|
|
max(outlets, key=lambda i: (len(children[i] - component), i))
|
|
if outlets
|
|
else max(component)
|
|
)
|
|
title = "共享 · " + nodes[source].name + " → " + str(len(targets)) + " 处计算"
|
|
blocks.append(dict(title=title, members=sorted(component), targets=sorted(targets)))
|
|
block_of = {i: k for k, b in enumerate(blocks) for i in b["members"]}
|
|
# Keep one local original read; additional consumers receive local copies later.
|
|
for i in sorted(inputs):
|
|
owners = [block_of[j] for j in sorted(children[i]) if j in block_of]
|
|
if any(j in core for j in children[i]):
|
|
core.add(i)
|
|
elif owners:
|
|
k = owners[0]
|
|
blocks[k]["members"].append(i)
|
|
block_of[i] = k
|
|
else:
|
|
core.add(i)
|
|
|
|
return blocks, core
|