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.

perf(image-studio/hidream): drop default steps 28 -> 20 (~5 min HD edits)

mrbizarroed040059115e
diff --git a/agent/image_engine.py b/agent/image_engine.py
index ad3d254..cfea048 100644
--- a/agent/image_engine.py
+++ b/agent/image_engine.py
@@ -192,7 +192,7 @@ class ImageEngineConfig:
# location; override when the lab moves.
hidream_python_path: str = "" # default = HIDREAM_LAB_DIR/.venv/bin/python
hidream_model_path: str = "" # default = HIDREAM_LAB_DIR/mlx_models/hidream-o1-dev-bf16
- hidream_steps: int = 28 # Dev = 28; Full = 50 (auto-overridden when recipe="full")
+ hidream_steps: int = 20 # Dev edit sweet spot (uniformly subsampled from upstream's 28-step DEFAULT_TIMESTEPS; visually identical, 30% faster). Full overrides to 50.
hidream_noise_scale: float = 7.5 # FlashFlowMatch tuned default; lowering collapses the image
hidream_noise_clip_std: float = 2.5
# New for the Full / undistilled variant: pick the lab script's recipe
diff --git a/mlx_ltx_panel.py b/mlx_ltx_panel.py
index a24dcba..98f077f 100644
--- a/mlx_ltx_panel.py
+++ b/mlx_ltx_panel.py
@@ -6384,7 +6384,7 @@ class Handler(BaseHTTPRequestHandler):
("flux2_edit_high_inline", None, 0.0, 240.0),
("flux2_inline", "Runpod/FLUX.2-klein-4B-mflux-4bit", 4.0, 12.0),
("z_image_turbo_inline", "filipstrand/Z-Image-Turbo-mflux-4bit", 3.0, 6.0),
- ("hidream_inline", None, 0.0, 540.0),
+ ("hidream_inline", None, 0.0, 300.0),
("mock_inline", None, 0.0, 0.5),
]
out = []
@@ -18416,7 +18416,7 @@ HTML = r"""<!doctype html>
<option value="flux2_edit_high_inline">FLUX.2 Klein-Base-Edit photoreal (multi-ref &middot; 25-step Q8, ~3-5 min/image)</option>
<option value="flux2_inline">FLUX.2 [klein] 4B (fast T2I, no refs)</option>
<option value="z_image_turbo_inline">Z-Image-Turbo (compact T2I, no refs)</option>
- <option value="hidream_inline">HiDream-O1-Image-Dev BF16 + flow_match HD (character-preserving edit &middot; ~9 min, 18 GB)</option>
+ <option value="hidream_inline">HiDream-O1-Image-Dev BF16 + flow_match HD (character-preserving edit &middot; ~5 min, 18 GB)</option>
<option value="mock_inline" id="imgStudioMockOption" hidden>Mock (testing — debug only)</option>
</select>
<span class="engine-status-pill" id="imgStudioEnginePill"