SCAIL-2 GGUF (Optimized Weights)
This repository contains GGUF quantizations of SCAIL-2, the cutting-edge end-to-end controlled character animation and video motion-transfer model built on the Wan 2.1 14B architecture by zai-org.
These weights are specifically optimized to execute the heavy SCAIL-2 Diffusion Transformer (DiT) inside ComfyUI at a fraction of the native FP16/FP8 VRAM overhead.
⚡ Quick Start (ComfyUI)
- Required Loader: Install city96's ComfyUI-GGUF extension.
- Node Placement: Use the
Unet Loader (GGUF)node. - File Path: Download your preferred quant from this repository and drop it into: ComfyUI/models/unet/
📦 Quantization Tiers
| File Name | File Size | Recommended VRAM / Profile |
|---|---|---|
| SCAIL-2-Q3_K_M.gguf | 9.11 GB | Budget setup; extreme VRAM savings with minor structural degradation. |
| SCAIL-2-Q4_K_S.gguf | 10.7 GB | Light daily driver; balanced compression. |
| SCAIL-2-Q4_K_M.gguf | 11.5 GB | Recommended baseline; excellent compromise between detail retention and performance. |
| SCAIL-2-Q5_K_S.gguf | 12.1 GB | Higher tier baseline; minimizes artifacts in fine textures. |
| SCAIL-2-Q5_K_M.gguf | 13.0 GB | Optimal performance sweet-spot for mid-to-high tier GPUs. |
| SCAIL-2-Q6_K.gguf | 14.2 GB | Premium quantization; near-identical output to high precision. |
| SCAIL-2-Q8_0.gguf | 18.1 GB | Maximum fidelity; mathematically closest to native fp16 behaviors. |
🛠️ Required Pipeline Dependencies (Not Included)
SCAIL-2 requires an ensemble of specialized modules to establish its end-to-end processing pipeline. Download these separately and place them in their designated ComfyUI structural paths:
1. 📝 Text Encoder
- Path:
ComfyUI/models/text_encoders/ - Download: umt5-xxl-enc-fp8_e4m3fn.safetensors
2. 🎛️ Distillation LoRA (LightX2V step/cfg distill)
- Path:
ComfyUI/models/loras/ - Download: Wan21_I2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors
3. 🎯 SAM 3.1 Multiplex
- Path:
ComfyUI/models/sam/ - Download: sam3.1_multiplex_fp16.safetensors
4. 👁️ CLIP Vision
- Path:
ComfyUI/models/clip_vision/ - Download: clip_vision_h.safetensors
5. 🎨 VAE Checkpoint
- Path:
ComfyUI/models/vae/ - Download: wan_2.1_vae.safetensors