Hybrid Facial Deepfake Detection Framework Using Swin V2 Vision Transformers and Multi-Scale Textural Descriptors