The paper proposes a three‑stage training method that extracts fine‑grained stylistic representations from a single reference image and injects them into a pretrained generative model without altering its architecture. A style encoder and a style projection layer align these representations with textual cues, enabling controlled stylized image generation. The authors also introduce the Style30k‑captions dataset, which pairs images, style labels, and text descriptions to train the encoder and projection layer.
Image style is a highly abstract, human-constructed concept shaped by a range of visual factors and intrinsically entangled with content, yet a unified and explicit definition of image style remains l...
MegaStyle++ introduces a hierarchical definition of image style, ranging from overall style identity to fine‑grained visual attributes, to provide a more structured, transferable, and interpretable representation. Using this definition, the authors refined the MegaStyle annotation pipeline and released MegaStyle++‑8M, a dataset with 150K style identities, 1M fine‑grained prompts, and 8M stylized images. Analyses show that the hierarchical approach expands style diversity and semantic breadth while accurately capturing the intrinsic visual style of reference images.
This survey reviews 490 papers on text‑to‑image synthesis (T2I) over the past decade, covering key model families such as GANs, autoregressive models, and diffusion models. It organizes recent research by problem areas—controllability, personalization, editing, preference alignment, efficiency, safety, and spatial consistency—rather than strictly chronologically. The paper also examines unified multimodal models, datasets, evaluation metrics, and discusses applications, challenges, and future research directions in T2I.
The paper introduces a post‑training approach that enables a single inference process to transition from text reasoning to image synthesis, eliminating the need for explicit modality switching. Using the 14B BAGEL model, the authors demonstrate that targeted post‑training data and reward‑weighted training improve multimodal image generation across four independent T2I benchmarks. The study highlights the benefits of joint text‑image generation and strategic data selection for enhancing T2I performance.
RefDiT is a new framework for reference-guided image generation that addresses the shortcomings of previous methods when handling complex scenes with multiple objects. It introduces local region guidance by decomposing a single identifier token into attribute-level signals, allowing the model to learn correspondences between tokens and specific regions of a reference image. The approach incorporates a low-rank adapter (LoRA) within a diffusion transformer (DiT) to adjust the inference prompt based on user-provided guidance context, thereby enabling more precise local attribute control.
Fashion complementary image generation (CIG) aims to create garments that stylistically match a seed item based on user intent, making it a natural multimodal grounding problem where models must inter...
arXiv:2510.12041v3 Announce Type: replace Abstract: Recent advances in text-to-image (T2I) generation have achieved impressive results, yet existing models often struggle with simple or underspecifie...
arXiv:2608.23302v1 Announce Type: new Abstract: Fashion complementary image generation (CIG) aims to create garments that stylistically match a seed item based on user intent, making it a natural mul...
CompArt introduces a new approach to aesthetic alignment in text-to-image generation by using the Principles of Art (PoA) such as Balance, Rhythm, and Emphasis to define explicit compositional constraints. The authors create a large dataset of 80,032 WikiArt images, each annotated with PoA analyses generated by a multimodal LLM, and present ArtDapter, a lightweight adapter that steers a pretrained diffusion model along ten PoA dimensions while preserving semantic fidelity. Experiments demonstrate that CompArt outperforms strong baselines in adhering to PoA controls under a dual evaluation protocol.
arXiv:2610.33472v1 Announce Type: new Abstract: Learning visual design is essential for prompt enhancement (PE) to turn underspecified user requests into coherent visual specifications for text-to-im...
arXiv:2609.08032v1 Announce Type: cross Abstract: We introduce FlexMoGen, a novel framework for flexible human motion synthesis conditioned on both natural language descriptions and motion style refe...
arXiv:2412.00176v4 Announce Type: replace Abstract: We explore whether pre-training on datasets with paintings is necessary for a model to learn an artistic style with only a few examples. To investi...