Swift-Image is a compact unified model that performs text-to-image generation, single-image editing, and multi-image editing using a 6B parameter DiT architecture. It employs a progressive training pipeline, parallel expert reinforcement learning, and multi-teacher distillation to balance diverse objectives, while a Prompt Enhancer decouples high-level reasoning from pixel-level rendering. After training, structural pruning and few-step distillation produce efficient 3B and accelerated variants that maintain near‑lossless performance and improve editing efficiency.
By Taihang Hu, Zhao Wang, Zuan Gao, Tao Liu, Hao Yan, Zhengze Xu, Yuhang Yu, Yongchao Du, Xingjian Wang, Jun Zheng, Qinye Zhou, Yaqi Cai, Zhengrui Chen, Chao Lin, Yefeng Shen, Yuan Wang, Zhengtao Wu, Ge Wu, Xiaoli Xu, Denghui Yang, Huayu Zhang, Mingzhou Zhang, Mengting Chen
PixelBench is a benchmark that evaluates large language models on their ability to generate and understand images pixel by pixel, without relying on learned image tokenizers or decoders. The benchmark includes text-to-pixel (T2P) and image-to-pixel (I2P) generation tasks across multiple resolutions, as well as a pixel-understanding evaluation. Models produce an RGB palette followed by a grid of indices that map each pixel to a palette color, and the study shows that strong models can generate recognizable images from text and outperform specialized pixelization algorithms at low resolutions, with performance correlating with pixel-understanding ability and improving up to a point before declining at high resolutions. Additionally, Pixel Reconstruction Tuning (PRT) augments visual instruction tuning with caption‑palette‑grid trajectories, and post‑training Molmo2‑4B with PRT data improves performance on 16 visual‑centric benchmarks, demonstrating the utility of explicit pixel reconstruction as auxiliary supervision.
By Haoquan Zhang, Feiyang Liu, Juanxi Tian, Wanhua Li, Haodong Duan, Weiyang Liu
arXiv:2608. 20334v1 Announce Type: new Abstract: We present Swift-Image, a compact unified model for text-to-image generation, single-image editing, and multi-image editing.
By Taihang Hu, Zhao Wang, Zuan Gao, Tao Liu, Hao Yan, Zhengze Xu, Yuhang Yu, Yongchao Du, Xingjian Wang, Jun Zheng, Qinye Zhou, Zhengrui Chen, Chao Lin, Yefeng Shen, Zhengtao Wu, Ge Wu, Xiaoli Xu, Denghui Yang, Huayu Zhang, Mingzhou Zhang, Mengting Chen
arXiv:2607. 00371v1 Announce Type: cross Abstract: Visual AutoRegressive modeling (VAR) has pioneered a coarse-to-fine multi-scale autoregressive generative paradigm, demonstrating strong capabilities in image generation.
By Nuoyan Zhou, Zhijun Tu, Lei Yu, Kun Cheng, Jie Hu, Nannan Wang, Xinghao Chen
UVU is a vision-language unified autoregressive framework that integrates visual supervision directly into the pre-training stage of multimodal large language models. By using continuous visual encoding and a large-scale iterative hierarchical clustering algorithm to build a pixel-level visual codebook, UVU enables lossless representation of visual inputs and autoregressive generation of pixel-level image tokens alongside textual tokens. This approach synergizes pixel-level visual perception with semantic-level visual understanding, allowing models to internalize visual reconstruction capabilities and improve multimodal understanding performance.
By Zhehan Kan, Xinghua Jiang, Yubo Zhu, Yanlin Liu, Xiaochen Yang, Zhixiang Wei, Shifeng Liu, Qingmin Liao, Wenming Yang, Xin Li, Yinsong Liu, Deqiang Jiang, Xing Sun
arXiv:2609.01147v1 Announce Type: cross
Abstract: Text-rich visual inputs require models that can read, retrieve, and compress language directly in pixel space, yet existing pixel-text encoders strug...
By Chaohao Yuan, Ruifeng Yuan, Zhuoxu Huang, Yu Rong, Hong Cheng, Hou Pong Chan, Chenghao Xiao
arXiv:2608. 08676v1 Announce Type: cross Abstract: Semantic vision encoders have become a central visual interface for multimodal understanding and semantic conditioning in image generation.
By Jinbo Yan, Limeng Qiao, Jie Qin, Junyan He, Feize Wu, Guanglu Wan
We present Swift-Image, a compact unified model for text-to-image generation, single-image editing, and multi-image editing. Our goal is to explore how far a relatively small visual generator can be pushed through systematic training engineering under a constrained computational budget.
The paper introduces LILIE, a Language‑Informed and Lossless Image Encoder that transforms images via a normalizing flow transformer, enabling a vision‑language model to recover captions from noised features. By training JiT models with a language‑informed loss derived from LILIE, the authors achieve a 14‑point reduction in gFD on the GPIC dataset compared to a pixel‑only baseline. The code, encoders, and generative models are publicly released at https://suchiragarwal.com/lilie.
By Suchir Agarwal, Keshigeyan Chandrasegaran, Ziyu Chen, Kyle Sargent, Juan Carlos Niebles, Li Fei-Fei, Jiajun Wu
MoETok introduces a sparse expert routing mechanism for 1D visual tokenizers, enabling distinct Content and Detail queries to specialize in high‑level semantics and fine‑grained visual details, respectively. By jointly optimizing reconstruction and semantic objectives, the router learns role‑specialized processing paths that improve reconstruction, semantic representation, and autoregressive generation across various model scales and token budgets. On ImageNet, MoETok achieves state‑of‑the‑art class‑conditional generation within the 1D tokenizer family and, when scaled to a unified autoregressive model, matches or surpasses diffusion‑based models on benchmarks such as DPG‑Bench, GenEval2, and WISE.
By Jiahao Guo, Bao Tang, Haoxiang Cao, Jingfeng Yao, Bo Li, Wenyu Liu, Kun Gai, Changqian Yu, Xinggang Wang
VIVAS is a new Vision‑Language Model pre‑training framework that addresses the lack of fine‑grained visual perception in existing VLMs. It introduces a unified token space and a dense‑structural‑semantic vision tokenizer that expands the textual vocabulary with visual tokens, enabling vision‑language unified autoregressive supervision over both visual details and linguistic content. Trained on 12.4 T tokens, VIVAS achieves state‑of‑the‑art results on 7 tasks and 39 multimodal benchmarks.
By Zhehan Kan, Yubo Zhu, Xinghua Jiang, Zhixiang Wei, Shifeng Liu, Wei Tong, Sheng Zhong, Qingmin Liao, Wenming Yang, Xin Li, Yinsong Liu, Deqiang Jiang, Xing Sun
WithinVAE is a compact variational visual autoencoder that learns semantic structure internally using a sequential bottleneck and self‑distillation from an exponential‑moving‑average copy of itself. It achieves a four‑fold compression ratio on ImageNet‑1K, improving reconstruction metrics such as PSNR by 1.36 dB over RPiAE while using half the latent capacity. The model also boosts linear probing performance, demonstrating that semantic gains arise from self‑distillation rather than external supervision.
By Yuzhou Ji, Jiepeng Wang, Tianyu Jin, Lizhuang Ma, Haibin Huang, Chi Zhang, Xuelong Li