Match author's validated inference defaults: 50 steps, CFG 3.5 (anima_minimal_inference_control_net_vace.py)

Match input aspect ratio (sliders = pixel budget); Depth-Anything-V2-Large on GPU for crisper control

Fix VACE context: feed LLMAdapter-adapted embeddings to control blocks (matches ComfyUI Anima.extra_conds); default flow shift 5.0 per model card

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload folder using huggingface_hub

initial commit