# Elastic Token Compression for Pixel-Space Diffusion Transformers

arXiv: https://arxiv.org/abs/2608.29281 · HTML: https://mlsanity.com/paper/2608.29281/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models.
- Builds on 17 of its 38 references that are in the map.
- First posted on arXiv 29 Aug 2026.

## Placed in

- Diffusion models / [Diffusion foundations and sampling](https://mlsanity.com/diffusion/diffusion-foundations-and-sampling/)
- Efficient LLM inference and serving / [Visual token pruning in multimodal models](https://mlsanity.com/inference/visual-token-pruning-in-multimodal-models/)

## Builds on (most cited first, 17 of 17)

- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2403.03206/) (2024)
- [CLIPScore: A Reference-free Evaluation Metric for Image Captioning](https://mlsanity.com/paper/2104.08718/) (2021)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference](https://mlsanity.com/paper/2310.04378/) (2023)
- [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://mlsanity.com/paper/2107.14795/) (2021)
- [Point Transformer V3: Simpler, Faster, Stronger](https://mlsanity.com/paper/2312.10035/) (2023)
- [Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach](https://mlsanity.com/paper/2502.05171/) (2025)
- [From Sparse to Soft Mixtures of Experts](https://mlsanity.com/paper/2308.00951/) (2023)
- [Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolution](https://mlsanity.com/paper/2307.06304/) (2023)
- [Token Merging for Fast Stable Diffusion](https://mlsanity.com/paper/2303.17604/) (2023)
- [Scalable Adaptive Computation for Iterative Generation](https://mlsanity.com/paper/2212.11972/) (2022)
- [Mixture of Nested Experts: Adaptive Processing of Visual Tokens](https://mlsanity.com/paper/2407.19985/) (2024)
- [Importance-Based Token Merging for Efficient Image and Video Generation](https://mlsanity.com/paper/2411.16720/) (2024)
- [Layer- and Timestep-Adaptive Differentiable Token Compression Ratios for Efficient Diffusion Transformers](https://mlsanity.com/paper/2412.16822/) (2024)
- [One Model, Many Budgets: Elastic Latent Interfaces for Diffusion Transformers](https://mlsanity.com/paper/2603.12245/) (2026)
- [DC-DiT: Adaptive Compute and Elastic Inference for Visual Generation via Dynamic Chunking](https://mlsanity.com/paper/2603.06351/) (2026)

## Built on by (most cited first, 2 of 2)

- [Level-of-Token Diffusion](https://mlsanity.com/paper/2610.05816/) (2026)
- [BudgetPix: Compute-Adaptive Tokenization for Pixel-Space Image Diffusion](https://mlsanity.com/paper/2610.12307/) (2026)

## Abstract (arXiv)

Natural images concentrate their detail in a small fraction of the frame, yet diffusion models spend a full token on every patch, in every layer and at every timestep. The waste is largest in pixel-space models, with no autoencoder to absorb low-level redundancy first. Probing a pretrained pixel text-to-image transformer, we find its middle-block tokens redundant wherever the image is flat. The redundancy occupies connected, content-shaped regions, and exploiting it requires tokens with the same geometry. Cutting a Hilbert ordering of the patches provides them. Consecutive positions are always image neighbours, so any contiguous run is a connected region whose size and shape follow the content, and grouping in two dimensions becomes a cut in one. Existing reductions each lose part of this. Similarity merging scatters its groups, latent bottlenecks discard position, and skipping deletes what it should summarize. We cut where the model's features change most and pool each run into one region token. Our Region Token Interface (method) adapts a diffusion model to these tokens, with the region count drawn at random during fine-tuning so one checkpoint serves every budget. method leads prior reduction methods at matched budgets, matches dense quality at 2.0× the speed, and stays close at 2.6×. The code and models are open-sourced at https://eduardzamfir.github.io/rti