Add model card

README.md

@@ -1,3 +1,30 @@
 ---
 license: apache-2.0
+library_name: diffusers
+pipeline_tag: image-to-image
 ---
+
+# DNF-SR
+
+DNF-SR is a real-world image super-resolution model built on [FLUX.1-Kontext-dev](https://huggingface.co/black-forest-labs/FLUX.1-Kontext-dev). It uses a dual-input and negative-aware feature fine-tuning method, with LoRA adapters trained on both the FLUX transformer and the VAE encoder.
+
+- **Paper:** [DNF-SR: Dual-Input and Negative-Aware Feature Fine-Tuning for Real-World Image Super-Resolution](https://huggingface.co/papers/2609.15120)
+- **Code:** [https://github.com/SHH-Han/DNF-SR](https://github.com/SHH-Han/DNF-SR)
+
+The repository contains LoRA adapters for the FLUX transformer and VAE encoder, enabling one-step real-world image super-resolution. The base FLUX weights remain frozen.
+
+## Sample Usage
+
+The official GitHub README provides inference scripts. Example:
+
+```bash
+INPUT=./data/test_LR OUTPUT=./outputs/infer UPSCALE=4 SEED=42 bash scripts/infer.sh
+```
+
+Or directly on a single image:
+
+```bash
+python3 infer/infer_dual.py --input demo/Canon_050_LR4.png --output_dir ./outputs/infer \
+    --flux_path black-forest-labs/FLUX.1-Kontext-dev --lora_path ./ckpt/dual_sft/ \
+    --prompt "A close-up, shallow-focus photograph highlights a bottle of pink Martini rosé wine against a blurred background of other liquor bottles on a shelf. Make this image more realistic and detailed." --upscale 4
+```