LucidNFT:
LR-Anchored Multi-Reward Preference Optimization for Flow-Based Real-World Super-Resolution

🌐 Website | 📘 Arxiv | 🤗 Models

---

📰 News & Updates

[2026.03.26] — Thanks to smthemex for developing ComfyUI_LucidNFT.

---

👥 Authors

Song Fei^{1, †}, Tian Ye^{1, †}, Sixiang Chen¹, Zhaohu Xing¹, Jianyu Lai¹, Lei Zhu^1,2,*

¹The Hong Kong University of Science and Technology (Guangzhou)
²The Hong Kong University of Science and Technology

† Equal Contribution, * Corresponding Author

---

💡 We also have other projects on 4K text-to-image generation and photo-realistic image estoration that may interest you. ✨

[CVPR 2026 Highlight] UltraFlux: Data-Model Co-Design for High-quality Native 4K Text-to-Image Generation across Diverse Aspect Ratios
Tian Ye¹*‡, Song Fei¹*, Lei Zhu^1,2†

[ICLR 2026] LucidFlux: Caption-Free Photo-Realistic Image Restoration via a Large-Scale Diffusion Transformer
Song Fei¹*, Tian Ye¹*‡, Lujia Wang¹ , Lei Zhu^1,2†

---

🌟 What is LucidNFT?

Generative real-world image super-resolution can synthesize visually convincing details from severely degraded low-resolution inputs, yet stochastic sampling makes a critical failure mode hard to avoid: outputs may look sharp but be unfaithful to the LR evidence, exhibiting semantic or structural hallucinations.

LucidNFT is a multi-reward RL framework for flow-matching Real-ISR. It introduces LucidConsistency, a degradation-invariant and hallucination-sensitive LR-referenced evaluator trained with content-consistent degradation pools and original-inpainted hard negatives; a decoupled reward normalization strategy that preserves objective-wise contrasts within each LR-conditioned rollout group before fusion; and LucidLR, a large-scale collection of real-world degraded images for robust RL fine-tuning.

Why LucidNFT?

• LR faithfulness is missing. Without HR references, no-reference perceptual metrics can reward sharp but unsupported details. Real-ISR needs an LR-referenced signal that is robust to degradations and sensitive to hallucination.
• Rollout groups need contrast. Preference learning compares multiple stochastic restorations conditioned on the same LR image. Scalarizing heterogeneous rewards before normalization can collapse perceptual-faithfulness distinctions.
• Real degradations need scale. RL alignment benefits from diverse LR inputs that induce informative rollout variation. Small benchmark datasets and synthetic pipelines limit degradation coverage.

🏗️ Method Overview

LucidNFT consists of three key ingredients:

1. LucidConsistency. A Qwen3-VL-Embedding-8B backbone with trainable LoRA adapters learns global and native-token representations through pool-based contrastive losses. At inference, it combines global and local LR-SR consistency into an LR-referenced score.
2. Decoupled reward normalization. Each reward dimension is normalized within the same LR-conditioned rollout group before fusion. The fused advantage is stabilized at batch level and mapped to the bounded DiffusionNFT reward weight.
3. LucidNFT fine-tuning. LucidNFT uses UniPercept IQA as the perceptual reward and LucidConsistency as the LR-faithfulness reward. Fine-tuning uses LoRA rank 32, 12 rollouts per LR input, and LucidLR as the real-world LR source.

Overview of LucidConsistency. Same-content views under different degradation levels form positive pools, while original-inpainted pools provide spatially aligned hard negatives for localized AI-generated hallucinations.

📦 LucidLR Dataset

LucidLR is a 20K-image real-world low-quality dataset collected from Wikimedia Commons through its official API. Images are gathered from public low-quality and blurred-image categories, filtered from an approximately 22K-image raw pool with NSFW classification, corrupted-file removal, and manual review.

Representative LucidLR samples. LucidLR provides diverse real-world degradations as LR inputs for RL fine-tuning.

Dataset	Pairing	Primary Usage	Type	# Images
RealSR	Paired	Testing / Benchmark	Real-captured	100
DRealSR	Paired	Testing / Benchmark	Real-captured	93
RealLQ250	Unpaired	Testing / Benchmark	Real-world	250
LucidLR	Unpaired	RL / Unsupervised Training	Real-world	20K

📊 Performance Benchmarks

Advantage separability analysis on LucidFlux using RealLQ250. Decoupled normalization produces larger advantage gaps and more distinct advantage levels than scalar-first reward aggregation under the same DiffusionNFT objective.

Training curves of LucidNFT and DPO fine-tuning. LucidNFT steadily improves perceptual rewards while keeping LucidConsistency stable.

Experiments evaluate LucidNFT on two flow-based Real-ISR models, LucidFlux and DiT4SR. All methods are evaluated at 1024 x 1024 output resolution with 4x upscaling. The paper reports eight no-reference quality metrics and LucidConsistency as an LR-referenced consistency score without HR ground truth.

📈 Quantitative Results

Quantitative comparison on RealLQ250, DRealSR, and RealSR. Higher is better except NIQE.

Benchmark	Metric	DiffBIRv2	SeeSR	DreamClear	SUPIR	DiT4SR	DiT4SR(+LucidNFT)	LucidFlux	LucidFlux(+DPO)	LucidFlux(+LucidNFT)
RealLQ250	CLIP-IQA+ ↑	0.6919	0.7034	0.6813	0.6532	0.7098	0.7124	0.7208	0.7228	0.7465
	Q-Align ↑	3.9755	4.1423	4.0647	4.1347	4.2270	4.2358	4.4052	4.4430	4.4855
	MUSIQ ↑	67.5313	70.3757	67.0899	65.8133	71.6682	72.1732	72.3351	72.4504	73.4475
	MANIQA ↑	0.4900	0.4895	0.4405	0.3826	0.4607	0.4719	0.5227	0.5258	0.5443
	CLIP-IQA ↑	0.7137	0.7063	0.6957	0.5767	0.7141	0.7355	0.6855	0.6917	0.7233
	NIQE ↓	5.1193	4.4383	3.8709	3.6591	3.5556	3.5007	3.7410	3.7785	3.2532
	UniPercept IQA ↑	65.4760	69.2015	68.8465	68.6430	73.0740	73.3430	70.9300	71.1330	73.4790
	VisualQuality-R1 ↑	4.3428	4.5118	4.4430	4.4265	4.6146	4.6304	4.5474	4.5644	4.6510
	LucidConsistency ↑	0.9609	0.9466	0.9578	0.9522	0.9052	0.9172	0.9237	0.9296	0.9345
DRealSR	CLIP-IQA+ ↑	0.6476	0.6258	0.4462	0.5494	0.6537	0.6757	0.6516	0.6530	0.6867
	Q-Align ↑	3.0487	3.2746	2.4214	3.4722	3.6008	3.6641	3.7141	3.7408	3.8423
	MUSIQ ↑	60.0759	61.3222	35.1912	54.9280	63.8051	65.1915	64.6025	64.5607	68.1545
	MANIQA ↑	0.4900	0.4505	0.2676	0.3483	0.4419	0.4572	0.4678	0.4669	0.5004
	CLIP-IQA ↑	0.6782	0.6760	0.4361	0.5310	0.6732	0.7111	0.6673	0.6713	0.7073
	NIQE ↓	6.4853	6.4503	7.0164	5.9092	5.7001	5.6329	5.0742	5.0143	4.1788
	UniPercept IQA ↑	46.2298	50.3414	34.2473	55.1371	58.1290	59.9328	59.9032	59.7782	63.7782
	VisualQuality-R1 ↑	3.4796	3.6116	2.5655	3.7349	3.9603	4.0239	3.9955	3.9828	4.1455
	LucidConsistency ↑	0.9332	0.9275	0.9607	0.8911	0.8438	0.8544	0.8813	0.8890	0.8879
RealSR	CLIP-IQA+ ↑	0.6543	0.6731	0.5331	0.5640	0.6753	0.6881	0.6669	0.6695	0.7151
	Q-Align ↑	3.3156	3.6073	3.0040	3.4682	3.7106	3.7959	3.8728	3.9147	3.9918
	MUSIQ ↑	61.7751	67.5660	49.4766	55.6807	67.9828	69.1092	67.8962	67.9362	70.5625
	MANIQA ↑	0.4745	0.5087	0.3092	0.3426	0.4533	0.4654	0.4889	0.4907	0.5284
	CLIP-IQA ↑	0.6806	0.6993	0.5390	0.4857	0.6631	0.6963	0.6359	0.6427	0.6936
	NIQE ↓	6.0700	5.4594	5.2873	5.2819	5.0912	4.8332	4.8134	4.6804	3.9526
	UniPercept IQA ↑	53.6550	58.0538	46.7850	56.6063	63.2025	64.8425	60.0925	60.4775	64.7588
	VisualQuality-R1 ↑	3.8928	4.0635	3.5028	3.7821	4.1953	4.2429	4.1376	4.1503	4.3389
	LucidConsistency ↑	0.9544	0.9138	0.9475	0.9141	0.8318	0.8498	0.8853	0.8932	0.8923

Human-Aligned LR-Faithfulness Evaluation

Evaluator	Criterion	Agreement	Recall@1	Filter@1
CLIP-IQA	Perceptual Quality	0.391	0.186	0.093
MUSIQ	Perceptual Quality	0.349	0.116	0.093
Q-Align	Perceptual Quality	0.322	0.093	0.047
UniPercept-IQA	Perceptual Quality	0.322	0.070	0.093
Qwen3-VL-Embedding-8B	Generic Semantics	0.643	0.465	0.302
LucidConsistency	LR Faithfulness	0.690	0.558	0.558

Visual comparison on RealLQ250. LucidNFT variants recover more accurate text structures and finer local textures while preserving LR-supported semantics.

Visual comparison on RealLQ250. LucidFlux(+LucidNFT) improves local details while avoiding LR-inconsistent artifacts.

🎭 Gallery & Examples

LucidFlux vs LucidFlux(+LucidNFT)

LR	LucidFlux	LucidFlux + LucidNFT

---

🚀 Quick Start

🔧 Installation

git clone https://github.com/W2GenAI-Lab/LucidNFT.git
cd LucidNFT

python -m venv .venv
source .venv/bin/activate
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128
pip install -r requirements.txt

Prepare Weights

Run the downloader to populate weights/ with the required assets, including the FLUX base model, SwinIR, LucidFlux checkpoint, prompt embeddings, LucidNFT LoRA, UltraFlux VAE, and SigLIP:

python -m tools.hf_login --token "$HF_TOKEN"
python -m tools.download_weights --dest weights

This script also generates weights/env.sh. Source it before inference so the FLUX base paths are exported correctly:

source weights/env.sh

Inference

Run the LucidFlux baseline:

python inference.py \
  --checkpoint weights/lucidflux/lucidflux.pth \
  --control_image /path/to/lr_image_or_dir \
  --output_dir outputs \
  --width 1024 \
  --height 1024 \
  --num_steps 24 \
  --swinir_pretrained weights/swinir.pth \
  --siglip_ckpt weights/siglip \
  --offload

Run LucidFlux + LucidNFT LoRA:

python inference.py \
  --checkpoint weights/lucidflux/lucidflux.pth \
  --control_image /path/to/lr_image_or_dir \
  --output_dir outputs-lora \
  --width 1024 \
  --height 1024 \
  --num_steps 24 \
  --swinir_pretrained weights/swinir.pth \
  --siglip_ckpt weights/siglip \
  --lora_path weights/lucidflux/LucidFlux+LucidNFT_lora \
  --offload

LucidConsistency Scoring

The repository also includes a lightweight LucidConsistency scoring entrypoint for comparing an LR image against an HR/SR image, or two benchmark folders with matched file counts. The learned projection-head score is reported as LucidConsistency.

Score a single image pair:

python test_LucidConsistency.py \
  --model_name_or_path weights/LucidConsistency/Qwen3-VL-Embedding-8B \
  --proj_head weights/LucidConsistency/proj_head.pt \
  --lr /path/to/lr.png \
  --hr /path/to/hr_or_sr.png

Score two benchmark folders:

python test_LucidConsistency.py \
  --model_name_or_path weights/LucidConsistency/Qwen3-VL-Embedding-8B \
  --proj_head weights/LucidConsistency/proj_head.pt \
  --lr /path/to/lr_benchmark \
  --hr /path/to/hr_benchmark

Citation

@article{fei2026lucidnft,
  title={LucidNFT: LR-Anchored Multi-Reward Preference Optimization for Flow-Based Real-World Super-Resolution},
  author={Fei, Song and Ye, Tian and Chen, Sixiang and Xing, Zhaohu and Lai, Jianyu and Zhu, Lei},
  journal={arXiv},
  year={2026}
}

License

This repository is released under the license specified in LICENSE.

Acknowledgments

• LucidFlux
• FLUX
• DiffusionNFT
• Qwen3-VL-Embedding