Ewald
a4154ccf05
- Fix batch bypass tuple handling, memory budget device-awareness, grain RNG device, roto payload limits, preview temp-file cleanup, and route double-registration. - Add native VIDEO output (fps/audio) to effect nodes that previously only returned IMAGE/MASK when fed a VIDEO input. - Split the oversized src/preview/ops/geometry.ts and add a starting golden-fixture test harness (spherize, transform) comparing backend and frontend math. - Remove ImageOpsAppendV2 (Autogrow) and ImageOpsLensArtifacts; inline Append's helper back into its single caller. - Collapse the Color Correct curve widget behind a toggle so the node no longer grows too tall. - Fix Crop/generic-node video preview playback running at the RAF rate instead of the source video's real fps. - Fix Crop losing its saved output size after a workflow reload: syncCropWidgets recomputed height from width and a preset aspect ratio on every resync (including onConfigure on load), discarding the restored value. It now only refreshes aspect-ratio bookkeeping on a resync.
133 lines
2.5 KiB
Python
133 lines
2.5 KiB
Python
from __future__ import annotations
|
|
|
|
import importlib.util
|
|
from pathlib import Path
|
|
|
|
import pytest
|
|
import torch
|
|
|
|
|
|
ROOT = Path(__file__).resolve().parents[2]
|
|
BATCH_FILE = ROOT / "nodes" / "core" / "batch.py"
|
|
|
|
spec = importlib.util.spec_from_file_location(
|
|
"imageops_batch_test_module",
|
|
BATCH_FILE,
|
|
)
|
|
|
|
assert spec is not None
|
|
assert spec.loader is not None
|
|
|
|
batch_module = importlib.util.module_from_spec(spec)
|
|
spec.loader.exec_module(batch_module)
|
|
|
|
|
|
match_batch = batch_module.match_batch
|
|
match_mask_batch = batch_module.match_mask_batch
|
|
BatchMismatchError = batch_module.BatchMismatchError
|
|
cycle_to_length = batch_module.cycle_to_length
|
|
|
|
|
|
def test_equal_batches_unchanged():
|
|
a = torch.zeros(4, 8, 8, 3)
|
|
b = torch.ones(4, 8, 8, 3)
|
|
|
|
aa, bb = match_batch(a, b)
|
|
|
|
assert aa.shape == a.shape
|
|
assert bb.shape == b.shape
|
|
|
|
|
|
def test_singleton_broadcast():
|
|
a = torch.zeros(1, 8, 8, 3)
|
|
b = torch.ones(4, 8, 8, 3)
|
|
|
|
aa, bb = match_batch(a, b)
|
|
|
|
assert aa.shape[0] == 4
|
|
assert bb.shape[0] == 4
|
|
|
|
|
|
def test_strict_mismatch_raises():
|
|
a = torch.zeros(3, 8, 8, 3)
|
|
b = torch.ones(7, 8, 8, 3)
|
|
|
|
with pytest.raises(BatchMismatchError):
|
|
match_batch(a, b)
|
|
|
|
|
|
def test_hold_last():
|
|
a = torch.arange(3).view(3, 1)
|
|
b = torch.arange(5).view(5, 1)
|
|
|
|
aa, bb = match_batch(
|
|
a,
|
|
b,
|
|
policy="hold_last",
|
|
)
|
|
|
|
assert aa[:, 0].tolist() == [0, 1, 2, 2, 2]
|
|
assert bb[:, 0].tolist() == [0, 1, 2, 3, 4]
|
|
|
|
|
|
def test_loop_is_explicit():
|
|
a = torch.arange(3).view(3, 1)
|
|
b = torch.arange(7).view(7, 1)
|
|
|
|
aa, _ = match_batch(
|
|
a,
|
|
b,
|
|
policy="loop",
|
|
)
|
|
|
|
assert aa[:, 0].tolist() == [
|
|
0, 1, 2,
|
|
0, 1, 2,
|
|
0,
|
|
]
|
|
|
|
|
|
def test_mask_singleton_broadcast():
|
|
mask = torch.ones(1, 16, 16)
|
|
|
|
result = match_mask_batch(
|
|
mask,
|
|
target_batch=4,
|
|
)
|
|
|
|
assert result.shape == (4, 16, 16)
|
|
|
|
|
|
def test_mask_mismatch_raises():
|
|
mask = torch.ones(3, 16, 16)
|
|
|
|
with pytest.raises(BatchMismatchError):
|
|
match_mask_batch(
|
|
mask,
|
|
target_batch=7,
|
|
)
|
|
|
|
|
|
def test_cycle_to_length_exact_multiple():
|
|
t = torch.arange(3).view(3, 1)
|
|
|
|
out = cycle_to_length(t, 9)
|
|
|
|
assert out[:, 0].tolist() == [0, 1, 2] * 3
|
|
|
|
|
|
def test_cycle_to_length_truncates_final_repeat():
|
|
t = torch.arange(3).view(3, 1)
|
|
|
|
out = cycle_to_length(t, 7)
|
|
|
|
assert out[:, 0].tolist() == [0, 1, 2, 0, 1, 2, 0]
|
|
|
|
|
|
def test_cycle_to_length_noop_when_already_target():
|
|
t = torch.arange(4).view(4, 1)
|
|
|
|
out = cycle_to_length(t, 4)
|
|
|
|
assert out is t
|