music_local (ACE-Step) + motion_local (MoMask) operators with install scripts

install_momask.sh encodes the compat patches: numpy<2 + umath_tests shim,
matplotlib axes-clearing fix, np.float alias wrapper (mb_gen.py)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

scripts/install_acestep.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+# Install ACE-Step (text → music) into vendor/acestep + venvs/acestep.
+# Weights (~7GB) auto-download from HF on first generation.
+set -euo pipefail
+cd "$(dirname "$0")/.."
+[ -d vendor/acestep ] || git clone -q https://github.com/ace-step/ACE-Step.git vendor/acestep
+uv venv venvs/acestep --python 3.11 2>/dev/null || true
+cd vendor/acestep
+uv pip install --python ../../venvs/acestep/bin/python -e .
+echo "acestep installed"

scripts/install_momask.sh

@@ -0,0 +1,54 @@
+#!/usr/bin/env bash
+# Install MoMask (text → motion BVH) into vendor/momask + venvs/momask.
+# Research code: needs numpy<2 + matplotlib 3.6 + two small compat patches.
+set -euo pipefail
+cd "$(dirname "$0")/.."
+
+[ -d vendor/momask ] || git clone -q https://github.com/EricGuo5513/momask-codes.git vendor/momask
+uv venv venvs/momask --python 3.11 2>/dev/null || true
+uv pip install --python venvs/momask/bin/python \
+  torch "numpy<2" scipy "matplotlib==3.6.3" tqdm einops gdown ftfy regex \
+  "git+https://github.com/openai/CLIP.git"
+
+cd vendor/momask
+
+# HumanML3D checkpoints (188MB, Google Drive)
+if [ ! -d checkpoints/t2m/t2m_nlayer8_nhead6_ld384_ff1024_cdp0.1_rvq6ns ]; then
+  mkdir -p checkpoints/t2m && cd checkpoints/t2m
+  ../../../../venvs/momask/bin/gdown 1vXS7SHJBgWPt59wupQ5UUzhFObrnGkQ0
+  unzip -q humanml3d_models.zip && rm humanml3d_models.zip
+  cd ../..
+fi
+
+# Patch 1: numpy.core.umath_tests was removed — np.matmul is the same op
+python3 - <<'PYEOF'
+import pathlib
+SHIM = ("from types import SimpleNamespace as _SN; "
+        "import numpy as _np; ut = _SN(matrix_multiply=_np.matmul)")
+for f in ("visualization/Animation.py", "visualization/Quaternions.py"):
+    p = pathlib.Path(f); s = p.read_text()
+    if "umath_tests" in s:
+        p.write_text(s.replace("import numpy.core.umath_tests as ut", SHIM))
+PYEOF
+
+# Patch 2: matplotlib removed the ax.lines/ax.collections setters
+python3 - <<'PYEOF'
+import pathlib
+p = pathlib.Path("utils/plot_script.py"); s = p.read_text()
+s = s.replace("ax.lines = []", "[l.remove() for l in list(ax.lines)]")
+s = s.replace("ax.collections = []", "[c.remove() for c in list(ax.collections)]")
+p.write_text(s)
+PYEOF
+
+# Wrapper: restore np.float/np.int/... aliases the old code expects
+cat > mb_gen.py <<'PYEOF'
+"""MODELBEAST wrapper: numpy>=1.24 compat aliases, then run gen_t2m.py."""
+import numpy as np
+for _n, _v in {"float": float, "int": int, "object": object, "bool": bool, "str": str}.items():
+    if not hasattr(np, _n):
+        setattr(np, _n, _v)
+import runpy
+runpy.run_path("gen_t2m.py", run_name="__main__")
+PYEOF
+
+echo "momask installed"

server/operators/motion_local/manifest.json

@@ -0,0 +1,18 @@
+{
+  "id": "motion_local",
+  "name": "Motion (local, MoMask)",
+  "category": "motion",
+  "description": "Text → character motion as BVH (+ mp4 preview) via MoMask (HumanML3D). For the bespoke clips Mixamo doesn't have — describe the action in plain words. Output BVH retargets onto rigged characters in Blender (import BVH, retarget to the character_kit skeleton). CPU-only, ~1-2 min. Research-grade: expect roughly right, polish in Blender. Install: scripts/install_momask.sh",
+  "accepts": [],
+  "produces": ["motion"],
+  "resources": "cpu",
+  "entry": "run.py",
+  "python": "venvs/momask/bin/python",
+  "params_schema": {
+    "type": "object",
+    "properties": {
+      "text": {"type": "string", "default": "", "description": "The action: 'a person riffles through a crate of records, pulls one out and admires it'"},
+      "length": {"type": "integer", "default": -1, "minimum": -1, "maximum": 196, "description": "Frames at 20fps (196 max ≈ 10s); -1 = let the model estimate"}
+    }
+  }
+}

server/operators/motion_local/run.py

@@ -0,0 +1,61 @@
+import argparse
+import json
+import os
+import shutil
+import subprocess
+import sys
+import time
+import uuid
+from pathlib import Path
+
+os.environ["PATH"] = "/opt/homebrew/bin:/usr/local/bin:" + os.environ.get("PATH", "")
+
+ap = argparse.ArgumentParser()
+ap.add_argument("--input", action="append", default=[])
+ap.add_argument("--outdir", required=True)
+ap.add_argument("--params", default="{}")
+a = ap.parse_args()
+p = json.loads(a.params)
+
+text = (p.get("text") or "").strip()
+if not text:
+    print("ERROR: text is required")
+    sys.exit(1)
+
+ROOT = Path(__file__).resolve().parents[3]
+MOMASK = ROOT / "vendor" / "momask"
+if not (MOMASK / "mb_gen.py").exists():
+    print("ERROR: momask not installed — run scripts/install_momask.sh")
+    sys.exit(1)
+
+ext = f"mb_{uuid.uuid4().hex[:10]}"  # unique run dir inside the shared vendor tree
+cmd = [sys.executable, "mb_gen.py", "--gpu_id", "-1", "--ext", ext, "--text_prompt", text]
+length = int(p.get("length", -1))
+if length > 0:
+    cmd += ["--motion_length", str(length)]
+
+t0 = time.time()
+r = subprocess.run(cmd, cwd=str(MOMASK), capture_output=True, text=True)
+gen_dir = MOMASK / "generation" / ext
+files = sorted(gen_dir.rglob("*")) if gen_dir.exists() else []
+bvh = [f for f in files if f.suffix == ".bvh"]
+if r.returncode != 0 or not bvh:
+    print(r.stdout[-1000:])
+    print(r.stderr[-1500:], file=sys.stderr)
+    print("ERROR: momask generation failed")
+    sys.exit(1)
+
+outdir = Path(a.outdir)
+slug = "".join(c if c.isalnum() else "_" for c in text[:40]).strip("_")
+n = 0
+for f in files:
+    if f.suffix == ".bvh":
+        # _ik = foot-contact cleaned; that's the one to retarget
+        kind = "ik" if "_ik" in f.stem else "raw"
+        shutil.copy(f, outdir / f"motion_{slug}_{kind}.bvh")
+        n += 1
+    elif f.suffix == ".mp4" and "_ik" in f.stem:
+        shutil.copy(f, outdir / f"motion_{slug}_preview.mp4")
+        n += 1
+shutil.rmtree(gen_dir, ignore_errors=True)
+print(f"done in {time.time() - t0:.0f}s: {n} files for '{text[:60]}'", flush=True)

server/operators/music_local/manifest.json

@@ -0,0 +1,21 @@
+{
+  "id": "music_local",
+  "name": "Music (local, ACE-Step)",
+  "category": "audio",
+  "description": "Style tags → music WAV via ACE-Step 3.5B on MPS (Apache 2.0, ~7GB weights). Jingles, menu loops, shop muzak, boss themes; add lyrics for actual songs ([verse]/[chorus] markers work, '[instrumental]' for none). Runs on the primary only. Install: scripts/install_acestep.sh",
+  "accepts": [],
+  "produces": ["audio"],
+  "resources": "gpu",
+  "entry": "run.py",
+  "python": "venvs/acestep/bin/python",
+  "params_schema": {
+    "type": "object",
+    "properties": {
+      "prompt": {"type": "string", "default": "", "description": "Style tags, comma-separated: 'lo-fi synth funk, playful, toy robot bleeps, 90 bpm'"},
+      "lyrics": {"type": "string", "default": "[instrumental]", "description": "Lyrics with [verse]/[chorus] markers, or [instrumental]"},
+      "duration": {"type": "number", "default": 30, "minimum": 5, "maximum": 240, "description": "Seconds"},
+      "steps": {"type": "integer", "default": 27, "minimum": 10, "maximum": 100, "description": "27 = fast/good, 60 = best"},
+      "seed": {"type": "integer", "default": 42}
+    }
+  }
+}

server/operators/music_local/run.py

@@ -0,0 +1,49 @@
+import argparse
+import json
+import os
+import sys
+import time
+from pathlib import Path
+
+os.environ.setdefault("TQDM_DISABLE", "1")
+os.environ["PATH"] = "/opt/homebrew/bin:/usr/local/bin:" + os.environ.get("PATH", "")
+
+ap = argparse.ArgumentParser()
+ap.add_argument("--input", action="append", default=[])
+ap.add_argument("--outdir", required=True)
+ap.add_argument("--params", default="{}")
+a = ap.parse_args()
+p = json.loads(a.params)
+
+prompt = (p.get("prompt") or "").strip()
+if not prompt:
+    print("ERROR: prompt (style tags) is required")
+    sys.exit(1)
+
+from acestep.pipeline_ace_step import ACEStepPipeline  # noqa: E402
+
+seed = int(p.get("seed", 42))
+out = Path(a.outdir) / f"music_{seed}.wav"
+t0 = time.time()
+pipe = ACEStepPipeline(dtype="float32")  # bfloat16 unsupported on MPS
+print(f"pipeline ready in {time.time() - t0:.1f}s", flush=True)
+
+t1 = time.time()
+pipe(
+    prompt=prompt,
+    lyrics=(p.get("lyrics") or "[instrumental]").strip(),
+    audio_duration=float(p.get("duration", 30)),
+    infer_step=int(p.get("steps", 27)),
+    guidance_scale=15.0,
+    omega_scale=10.0,
+    manual_seeds=[seed],
+    save_path=str(out),
+)
+if not out.exists():
+    # some versions append suffixes; register whatever wav landed in outdir
+    wavs = list(Path(a.outdir).glob("*.wav"))
+    if not wavs:
+        print("ERROR: no wav produced")
+        sys.exit(1)
+    out = wavs[0]
+print(f"done in {time.time() - t1:.1f}s: {out.name}", flush=True)