# StyleVAR: Controllable Image Style Transfer via Visual Autoregressive Modeling

arXiv: https://arxiv.org/abs/2604.21052 · HTML: https://mlsanity.com/paper/2604.21052/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 6 of its 9 references that are in the map.
- First posted on arXiv 22 Apr 2026.

## Builds on (most cited first, 6 of 6)

- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction](https://mlsanity.com/paper/2404.02905/) (2024)
- [DreamSim: Learning New Dimensions of Human Visual Similarity using Synthetic Data](https://mlsanity.com/paper/2306.09344/) (2023)
- [Inversion-Based Style Transfer with Diffusion Models](https://mlsanity.com/paper/2211.13203/) (2022)
- [OmniStyle: Filtering High Quality Style Transfer Data at Scale](https://mlsanity.com/paper/2505.14028/) (2025)
- [VAR RL Done Right: Tackling Asynchronous Policy Conflicts in Visual Autoregressive Generation](https://mlsanity.com/paper/2601.02256/) (2026)

## Abstract (arXiv)

We build on the Visual Autoregressive Modeling (VAR) framework and formulate style transfer as conditional discrete sequence modeling in a learned latent space. Images are decomposed into multi-scale representations and tokenized into discrete codes by a VQ-VAE; a transformer then autoregressively models the distribution of target tokens conditioned on style and content tokens. To inject style and content information, we introduce a blended cross-attention mechanism in which the evolving target representation attends to its own history, while style and content features act as queries that decide which aspects of this history to emphasize. A scale-dependent blending coefficient controls the relative influence of style and content at each stage, encouraging the synthesized representation to align with both the content structure and the style texture without breaking the autoregressive continuity of VAR. We train StyleVAR in two stages from a pretrained VAR checkpoint: supervised fine-tuning on a large triplet dataset of content--style--target images, followed by reinforcement fine-tuning with Group Relative Policy Optimization (GRPO) against a DreamSim-based perceptual reward, with per-action normalization weighting to rebalance credit across VAR's multi-scale hierarchy. Across three benchmarks spanning in-, near-, and out-of-distribution regimes, StyleVAR consistently outperforms an AdaIN baseline on Style Loss, Content Loss, LPIPS, SSIM, DreamSim, and CLIP similarity, and the GRPO stage yields further gains over the SFT checkpoint, most notably on the reward-aligned perceptual metrics. Qualitatively, the method transfers texture while maintaining semantic structure, especially for landscapes and architectural scenes, while a generalization gap on internet images and difficulty with human faces highlight the need for better content diversity and stronger structural priors.