FireRed-Image-Edit-1.0 Techinical Report

2026-02-12

AI Summary

This work addresses the challenges of weak semantic alignment, limited controllability, and insufficient generalization in instruction-driven image editing by proposing a systematic optimization framework. Leveraging a high-quality training corpus of 1.6 billion samples and a diffusion Transformer architecture, the approach integrates multi-stage training—encompassing pretraining, supervised fine-tuning, and reinforcement learning—with several novel techniques: a multi-condition-aware bucket sampler, stochastic instruction alignment, asymmetric gradient DPO, DiffusionNFT, differentiable consistency loss, and a layout-aware OCR reward. The authors also introduce REDEdit-Bench, a comprehensive benchmark covering 15 editing tasks. The resulting model consistently outperforms existing open- and closed-source systems across REDEdit-Bench, ImgEdit, and GEdit. All code, models, and evaluation suites are publicly released.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

We present FireRed-Image-Edit, a diffusion transformer for instruction-based image editing that achieves state-of-the-art performance through systematic optimization of data curation, training methodology, and evaluation design. We construct a 1.6B-sample training corpus, comprising 900M text-to-image and 700M image editing pairs from diverse sources. After rigorous cleaning, stratification, auto-labeling, and two-stage filtering, we retain over 100M high-quality samples balanced between generation and editing, ensuring strong semantic coverage and instruction alignment. Our multi-stage training pipeline progressively builds editing capability via pre-training, supervised fine-tuning, and reinforcement learning. To improve data efficiency, we introduce a Multi-Condition Aware Bucket Sampler for variable-resolution batching and Stochastic Instruction Alignment with dynamic prompt re-indexing. To stabilize optimization and enhance controllability, we propose Asymmetric Gradient Optimization for DPO, DiffusionNFT with layout-aware OCR rewards for text editing, and a differentiable Consistency Loss for identity preservation. We further establish REDEdit-Bench, a comprehensive benchmark spanning 15 editing categories, including newly introduced beautification and low-level enhancement tasks. Extensive experiments on REDEdit-Bench and public benchmarks (ImgEdit and GEdit) demonstrate competitive or superior performance against both open-source and proprietary systems. We release code, models, and the benchmark suite to support future research.

Problem

• instruction-based image editing
• diffusion transformer
• image editing benchmark
• semantic alignment
• controllable image generation

Innovation

• diffusion transformer
• instruction-based image editing
• multi-stage training pipeline
• asymmetric gradient optimization
• REDEdit-Bench