YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/model-cards#model-card-metadata)

πŸ“– Wan2.1 VACE + Phantom Causvid (Finetune)

Author / Creator: Inner_Reflections_AI
Packaged for WanGP by: hdtrnk


πŸ”Ή About This Finetune

This is a custom merge of Wan 2.1 VACE and Phantom, created by Inner_Reflections_AI and purified to FP16 for compatibility with WanGP.

  • Architecture: vace_14B
  • Mode: Image/Video conditioning with multi-reference support
  • Multi-Image: Optimized for 2–4 image references in WanGP custom UI
  • Variants: Both FP16 and INT8 versions provided

πŸ”Ή Files


πŸ”Ή Usage in WanGP

Place the JSON definition file into:

app/finetunes/vace_phantom_causvid.json

Sample finetune entry:

{
  "model": {
    "name": "VACE Phantom Causvid 14B",
    "architecture": "vace_14B",
    "description": "VACE-first UI with Phantom multi-image references enabled. Causvid merge, purified FP16 for clean loading & optional INT8 quantization.",
    "URLs": [
      "https://huggingface.co/hdtrnk/VACEPhantomCausvid/blob/main/Wan2.1_VACE_Phantom_Causvid_fp16_pure.safetensors",
      "https://huggingface.co/hdtrnk/VACEPhantomCausvid/blob/main/Wan2.1_VACE_Phantom_Causvid_quanto_bf16_int8_pure.safetensors"
    ],
    "modules": [],
    "auto_quantize": false
  }
}

πŸ”Ή Notes

  • This is an experimental finetune. You may need to tweak steps, guidance scale, and reference image setup for best results.
  • Works best with multi-image prompts where identities are clearly separated (e.g., β€œCharacter A on left, Character B on right”).
  • If you encounter dropdown errors (Value: on is not in the list...), refresh the UI and reselect options.

πŸ”Ή Credits

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support