Search papers, labs, and topics across Lattice.
The paper introduces Generative Ground Truth (GGT), a method leveraging generative multimodal foundation models (MFMs) to create high-quality (HQ) target images from real-world low-quality (LQ) images for image restoration (IR) training. They systematically evaluated nine MFMs, identifying Nano-Banana-2 with VLM-based adaptive prompting as the most effective for synthesizing realistic HQ targets. Using this, they built a GGT synthesis pipeline and constructed GGT-100K, a large-scale LQ-HQ paired dataset, demonstrating its effectiveness in improving the real-world generalization of various IR models.
Forget synthetic data: generative multimodal models can now hallucinate high-quality ground truth images from real-world degraded inputs, finally bridging the sim2real gap for image restoration.
Real-world image restoration (IR) is bottlenecked by the scarcity of high-quality paired training data. Synthetic datasets are abundant but often fail to model real-world degradations, while real-world paired datasets are expensive and difficult to capture. As a result, IR models trained on these datasets show limited generalization in real-world scenarios. In this work, we propose Generative Ground Truth (GGT) by using generative multimodal foundation models (MFMs) to produce high-quality (HQ) targets from real-world low-quality (LQ) images. We first conduct a systematic evaluation of nine state-of-the-art MFMs, including Nano-Banana-2 and GPT-Image-2, on images of various scenes and degradation types. The results demonstrate that Nano-Banana-2 with VLM-based adaptive prompting shows the highest capability to synthesize perceptually realistic and content-faithful HQ targets, which can serve as the GGT for the LQ input. We then employ Nano-Banana-2 to build a GGT synthesis pipeline, which involves multi-stage quality control to ensure data reliability, and construct GGT-100K, an LQ-HQ paired dataset comprising 103,707 training pairs and covering diverse scenes and complex real-world degradations. A test set of 500 image pairs is also established. Extensive experiments show that GGT-100K consistently improves the real-world generalization of a wide range of IR models, with particularly strong benefits for finetuning generative models for IR tasks. Our results suggest that MFMs can serve as practical tools for restoration-oriented data generation, and GGT-100K is a useful resource to expand the generalization boundaries of real-world IR models.
