cs.CVNov 2, 2026

PixelBench: Can LLMs Understand and Generate Images Pixel by Pixel?

Authors: Haoquan Zhang, Feiyang Liu, Juanxi Tian, Wanhua Li, Haodong Duan, Weiyang Liu

Organizations: CUHK · SCUT · NTU Singapore

Abstract

Large language models (LLMs) can generate visual assets through text-based formats such as SVG and CAD, but these symbolic representations describe images using high-level primitives rather than explicit raster values. We ask whether LLMs can instead generate pixel images directly through their existing vocabulary, without a learned image tokenizer or decoder. We introduce PixelBench, a benchmark for text-to-pixel (T2P) and image-to-pixel (I2P) generation across multiple resolutions, together with a complementary pixel-understanding evaluation. Models generate an RGB color palette followed by a grid of indices specifying the palette color of each output pixel. In T2P, the strongest models produce recognizable images directly from text. In I2P, the strongest models outperform palette-matched downsampling and specialized pixelization algorithms at low resolutions. Pixel-understanding performance correlates strongly with generation performance. A resolution-scaling study shows that pixel-generation performance often improves initially before declining at high resolutions. Finally, we introduce Pixel Reconstruction Tuning (PRT), which augments visual instruction tuning with caption-palette-grid trajectories. We post-train Molmo2-4B from its pretraining checkpoint by mixing general instruction data with our curated PRT data, improving performance across 16 visual-centric benchmarks, indicating that explicit pixel reconstruction can provide useful auxiliary supervision.

Figures & tables

Appendix figures & tables17 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Pixel is All You Need

    Nov 17, 2026Jie Chu, Haoting Shi, Fucheng Zhang +4Generative ModelingUnified Multimodal Models

  2. PixelBench: Evaluating Language Models for Pixel-Art Creation

    Nov 16, 2026Mutian Hong, Jianxuan Yao, Mingkun Yue +3Image GenerationT2I Generation Evaluation

  3. L2P: Unlocking Latent Potential for Pixel Generation

    May 12, 2026Zhennan Chen, Junwei Zhu, Xu Chen +7Latent Diffusion ModelsPixel-Space Diffusion Models