Skip to content

Free, open-source AI video generator for Mac. Run LTX-2.5 and MiniMax H3 (Hailuo) locally on Apple Silicon: text-to-video and image-to-video with sound, lip-sync, character LoRA training, songs from lyrics (YuE2) and image editing (Qwen-Image-Edit). One-click install via Pinokio. Offline: no cloud, no API key, no NVIDIA GPU.

feat(helper): wire TeaCache through Extend's denoise loop

mrbizarroea98aad8bb77
diff --git a/mlx_warm_helper.py b/mlx_warm_helper.py
index cf4e437..84661eb 100644
--- a/mlx_warm_helper.py
+++ b/mlx_warm_helper.py
@@ -179,6 +179,53 @@ _hq_model_dir = None # remember which model the HQ pipe was built against
_pipe_lock = threading.Lock()
+# ---- TeaCache for Extend (2026-05-18) -----------------------------------------
+#
+# Lightricks' RetakePipeline.extend() shares its denoise loop
+# (ltx_pipelines_mlx.utils.samplers.guided_denoise_loop) with the standard
+# two-stage T2V pipeline's Stage 1 — same dev transformer, same Euler sampler,
+# same ltx2_schedule sigma layout. That stage is exactly what TeaCache is
+# calibrated against in ti2vid_two_stages.LTX2_TEACACHE_COEFFICIENTS. The
+# coefficients are reusable here; what's missing is the wiring — extend()
+# doesn't pass `teacache=` into guided_denoise_loop today.
+#
+# Rather than fork retake.py we monkey-patch the symbol the retake module
+# imports, so only calls FROM retake (i.e. extend() and retake()) go through
+# the patched version. Two-stages, HQ, and any other caller are untouched.
+#
+# Activation is per-call: the extend action below sets _EXTEND_TC_CONFIG before
+# pipe.extend_from_video() and clears it after. The helper inference loop is
+# already serialized (single _is_busy gate) so the module-level state is safe.
+#
+# Default threshold matches Stage 1's `LTX2_TEACACHE_THRESH=0.5` — moderate
+# 1.2× speedup with ~22% skip rate. Aggressive users can pass higher values
+# in the job spec (1.0 → ~2×, 1.5 → ~3× per the upstream comment).
+_EXTEND_TC_CONFIG: dict | None = None
+
+try:
+ import ltx_pipelines_mlx.retake as _ltx_retake_mod
+ from ltx_pipelines_mlx.ti2vid_two_stages import _build_teacache_controller as _build_stage1_teacache
+ _orig_guided_denoise_loop_for_retake = _ltx_retake_mod.guided_denoise_loop
+
+ def _guided_denoise_loop_with_extend_teacache(*args, teacache=None, **kwargs):
+ cfg = _EXTEND_TC_CONFIG
+ if teacache is None and cfg and cfg.get("enable"):
+ sigmas = kwargs.get("sigmas")
+ n_steps = len(sigmas) - 1 if sigmas else cfg.get("num_steps", 12)
+ try:
+ teacache = _build_stage1_teacache(n_steps, cfg.get("thresh"))
+ except Exception:
+ # Don't fail the gen if calibration is somehow unusable.
+ teacache = None
+ return _orig_guided_denoise_loop_for_retake(*args, teacache=teacache, **kwargs)
+
+ _ltx_retake_mod.guided_denoise_loop = _guided_denoise_loop_with_extend_teacache
+ _EXTEND_TC_PATCH_OK = True
+except Exception as _tc_patch_exc:
+ print(f"[warm-helper] Extend TeaCache patch skipped: {_tc_patch_exc}", flush=True)
+ _EXTEND_TC_PATCH_OK = False
+
+
def release_pipelines(keep_kind=None):
"""Free every loaded pipeline except the one matching keep_kind.
@@ -1532,6 +1579,28 @@ for line in sys.__stdin__:
# extends into swap (240s/step instead of ~25s/step). The panel
# exposes a "Fast" / "Quality" toggle that flips this to 3.0.
cfg_scale = float(p.get("cfg_scale", 1.0))
+ num_steps = int(p.get("steps", 12))
+
+ # TeaCache for Extend (2026-05-18). Active only if the boot-time
+ # patch landed AND the job spec doesn't explicitly opt out. The
+ # monkey-patched guided_denoise_loop reads _EXTEND_TC_CONFIG to
+ # decide whether to construct a controller. Threshold defaults
+ # match the Stage-1 calibration's `LTX2_TEACACHE_THRESH=0.5`.
+ enable_tc = bool(p.get("enable_teacache", True)) and _EXTEND_TC_PATCH_OK
+ tc_thresh = p.get("teacache_thresh")
+ if tc_thresh is not None:
+ try:
+ tc_thresh = float(tc_thresh)
+ except (TypeError, ValueError):
+ tc_thresh = None
+ _EXTEND_TC_CONFIG = {
+ "enable": enable_tc,
+ "thresh": tc_thresh,
+ "num_steps": num_steps,
+ }
+ if enable_tc:
+ emit({"event": "log",
+ "line": f"TeaCache active on extend (thresh={tc_thresh if tc_thresh is not None else 'default 0.5'})."})
with _override_default_negative_prompt(p.get("negative_prompt")) as neg_active:
if neg_active:
emit({"event": "log", "line": "Avoid terms active via native CFG negative prompt."})
@@ -1541,7 +1610,7 @@ for line in sys.__stdin__:
extend_frames=int(p.get("extend_frames", 5)),
direction=p.get("direction", "after"),
seed=seed,
- num_steps=int(p.get("steps", 12)),
+ num_steps=num_steps,
cfg_scale=cfg_scale,
)
# Decode + save (mirrors the CLI _decode_and_save)
@@ -1566,6 +1635,10 @@ for line in sys.__stdin__:
_last_activity = time.time()
emit({"event": "error", "id": job_id, "error": str(exc), "trace": traceback.format_exc()})
finally:
+ # Always clear the per-call TeaCache config so a future
+ # non-extend call (or extend with enable=False) doesn't
+ # inherit a stale state.
+ _EXTEND_TC_CONFIG = None
_is_busy = False
continue