# MOSAIK: Multi-Patch Content-Aware Spatial Allocation of Image Tokens for Efficient Generation

arXiv: https://arxiv.org/abs/2608.05450 · HTML: https://mlsanity.com/paper/2608.05450/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models.
- Builds on 24 of its 36 references that are in the map.
- First posted on arXiv 5 Aug 2026.

## Placed in

- Diffusion models / [Diffusion foundations and sampling](https://mlsanity.com/diffusion/diffusion-foundations-and-sampling/)

## Builds on (most cited first, 24 of 24)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [ELLA: Equip Diffusion Models with LLM for Enhanced Semantic Alignment](https://mlsanity.com/paper/2403.05135/) (2024)
- [Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model](https://mlsanity.com/paper/2411.19108/) (2024)
- [Deep Compression Autoencoder for Efficient High-Resolution Diffusion Models](https://mlsanity.com/paper/2410.10733/) (2024)
- [From Reusing to Forecasting: Accelerating Diffusion Models with TaylorSeers](https://mlsanity.com/paper/2503.06923/) (2025)
- [PixelDiT: Pixel Diffusion Transformers for Image Generation](https://mlsanity.com/paper/2511.20645/) (2025)
- [DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation](https://mlsanity.com/paper/2511.19365/) (2025)
- [DiP: Taming Diffusion Models in Pixel Space](https://mlsanity.com/paper/2511.18822/) (2025)
- [Attention-Driven Training-Free Efficiency Enhancement of Diffusion Models](https://mlsanity.com/paper/2405.05252/) (2024)
- [Layer- and Timestep-Adaptive Differentiable Token Compression Ratios for Efficient Diffusion Transformers](https://mlsanity.com/paper/2412.16822/) (2024)
- [MSViT: Dynamic Mixed-Scale Tokenization for Vision Transformers](https://mlsanity.com/paper/2307.02321/) (2023)
- [Learning Adaptive and Temporally Causal Video Tokenization in a 1D Latent Space](https://mlsanity.com/paper/2505.17011/) (2025)
- [Accelerating Vision Transformers with Adaptive Patch Sizes](https://mlsanity.com/paper/2510.18091/) (2025)
- [HyperDiT: Hyper-Connected Transformers for High-Fidelity Pixel-Space Diffusion](https://mlsanity.com/paper/2605.15741/) (2026)
- [Pyramidal Patchification Flow for Visual Generation](https://mlsanity.com/paper/2506.23543/) (2025)
- [DDiT: Dynamic Patch Scheduling for Efficient Diffusion Transformers](https://mlsanity.com/paper/2602.16968/) (2026)
- [Foveated Diffusion: Efficient Spatially Adaptive Image and Video Generation](https://mlsanity.com/paper/2603.23491/) (2026)
- [MPDiT: Multi-Patch Global-to-Local Transformer Architecture For Efficient Flow Matching and Diffusion Model](https://mlsanity.com/paper/2603.26357/) (2026)

## Abstract (arXiv)

Pixel-space diffusion models avoid the reconstruction ceiling of latent diffusion models by generating directly in image space. However, their substantially higher token count makes generation expensive due to the quadratic complexity of self-attention. Several existing efficiency methods reduce this cost by using larger patches at selected denoising steps, thereby representing the image with fewer tokens. Yet, each step still uses a single patch size uniformly across the entire image, overlooking that different regions suffer different fidelity losses when coarsened. We introduce MOSAIK, a damage-guided framework that varies patch size across regions and denoising steps. MOSAIK adapts the PixelDiT backbone to generate arbitrary heterogeneous patch layouts, and a lightweight predictor uses intermediate denoising features to estimate the fidelity loss caused by coarsening each region. Given a token budget, our damage-guided layout predictor assigns fine patches to sensitive regions and coarse patches elsewhere. Remarkably, while reducing FLOPs by 70% and token count by 83%, MOSAIK matches the full-compute PixelDiT on GenEval and its DPG-Bench score drops by only 1.0 point. Compared to diverse efficiency paradigms, including temporal patch scheduling and feature caching, our approach delivers highly competitive performance at moderate budgets and consistently outperforms these baselines in highly constrained compute regimes.