# Why Do DiT Editors Drift? Plug-and-Play Low Frequency Alignment in VAE Latent Space

arXiv: https://arxiv.org/abs/2605.08250 · HTML: https://mlsanity.com/paper/2605.08250/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Image generation, text-to-image and editing, one of 14 directions in Diffusion models.
- Builds on 30 of its 40 references that are in the map.
- First posted on arXiv 7 May 2026.

## Placed in

- Diffusion models / [Image generation, text-to-image and editing](https://mlsanity.com/diffusion/image-generation-text-to-image-and-editing/)

## Builds on (most cited first, 30 of 30)

- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2403.03206/) (2024)
- [InstructPix2Pix: Learning to Follow Image Editing Instructions](https://mlsanity.com/paper/2211.09800/) (2022)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [Qwen-Image Technical Report](https://mlsanity.com/paper/2508.02324/) (2025)
- [FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space](https://mlsanity.com/paper/2506.15742/) (2025)
- [MagicBrush: A Manually Annotated Dataset for Instruction-Guided Image Editing](https://mlsanity.com/paper/2306.10012/) (2023)
- [Emu Edit: Precise Image Editing via Recognition and Generation Tasks](https://mlsanity.com/paper/2311.10089/) (2023)
- [Seedream 4.0: Toward Next-generation Multimodal Image Generation](https://mlsanity.com/paper/2509.20427/) (2025)
- [SmartEdit: Exploring Complex Instruction-based Image Editing with Multimodal Large Language Models](https://mlsanity.com/paper/2312.06739/) (2023)
- [InstructDiffusion: A Generalist Modeling Interface for Vision Tasks](https://mlsanity.com/paper/2309.03895/) (2023)
- [DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation](https://mlsanity.com/paper/2511.19365/) (2025)
- [FlexiEdit: Frequency-Aware Latent Refinement for Enhanced Non-Rigid Editing](https://mlsanity.com/paper/2407.17850/) (2024)
- [Multi-turn Consistent Image Editing](https://mlsanity.com/paper/2505.04320/) (2025)
- [VINCIE: Unlocking In-context Image Editing from Video](https://mlsanity.com/paper/2506.10941/) (2025)
- [LUVE : Latent-Cascaded Ultra-High-Resolution Video Generation with Dual Frequency Experts](https://mlsanity.com/paper/2602.11564/) (2026)
- [VISION-XL: High Definition Video Inverse Problem Solver using Latent Image Diffusion Models](https://mlsanity.com/paper/2412.00156/) (2024)
- [Decomposition-based Unsupervised Domain Adaptation for Remote Sensing Image Semantic Segmentation](https://mlsanity.com/paper/2404.04531/) (2024)
- [EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing](https://mlsanity.com/paper/2509.13399/) (2025)
- [Decoupling High and Low Frequencies for Faithful Image Generation with Fine Details](https://mlsanity.com/paper/2509.05441/) (2025)
- [Spectrum Matching: a Unified Perspective for Superior Diffusability in Latent Diffusion](https://mlsanity.com/paper/2603.14645/) (2026)
- [Structured Spectral Reasoning for Frequency-Adaptive Multimodal Recommendation](https://mlsanity.com/paper/2512.01372/) (2025)
- [FreqEdit: Preserving High-Frequency Features for Robust Multi-Turn Image Editing](https://mlsanity.com/paper/2512.01755/) (2025)
- [Improving Editability in Image Generation with Layer-wise Memory](https://mlsanity.com/paper/2505.01079/) (2025)
- [I2E: From Image Pixels to Actionable Interactive Environments for Text-Guided Image Editing](https://mlsanity.com/paper/2601.03741/) (2026)
- [Towards Architecture-Agnostic Untrained Network Priors for Image Reconstruction with Frequency Regularization](https://mlsanity.com/paper/2312.09988/) (2023)
- [Latent Space Disentanglement in Diffusion Transformers Enables Precise Zero-shot Semantic Editing](https://mlsanity.com/paper/2411.08196/) (2024)
- [REED-VAE: RE-Encode Decode Training for Iterative Image Editing with Diffusion Models](https://mlsanity.com/paper/2504.18989/) (2025)
- [SMSP: A Plug-and-Play Strategy of Multi-Scale Perception for MLLMs to Perceive Visual Illusions](https://mlsanity.com/paper/2603.23118/) (2026)

## Built on by (most cited first, 3 of 3)

- [On-Policy Self-Distillation for Multi-Turn Image Editing](https://mlsanity.com/paper/2609.35611/) (2026)
- [Latent-Frequency Validity: Fast Spectral Editing with Screened Video-VAE Transfer Operators](https://mlsanity.com/paper/2608.07569/) (2026)
- [Learning to Retrieve: Internalizing Memory Retrieval for Video World Models](https://mlsanity.com/paper/2610.11444/) (2026)

## Abstract (arXiv)

Recent advances in diffusion transformers (DiTs) have enabled promising single-turn image editing capabilities. However, multi-turn editing often leads to progressive semantic drift and quality degradation.In this work, we study this problem from a latent-space frequency perspective by decomposing the editing process into two functional components: VAE and DiT. Through systematic analysis in the VAE latent space, we uncover that the DiT introduces dominant low-frequency drift that accumulates as semantic misalignment across editing rounds, while the VAE contributes comparatively stable reconstruction bias.Based on this insight, we propose VAE-LFA (Low Frequency Alignment), a training-free, plug-and-play method that performs alignment in VAE latent space. VAE-LFA decomposes latent discrepancies across editing rounds via low-pass filtering, and aligns low-frequency statistics to an exponential moving average of previous rounds, effectively suppressing accumulated semantic drift while preserving high-frequency details.Our method requires no retraining, ground-truth priors, or access to diffusion parameters, making it applicable to both white-box and black-box DiT editors. For white-box models, VAE-LFA is seamlessly integrated into the editing pipeline by eliminating redundant VAE round trips; for black-box models, it operates via an off-the-shelf VAE to perform inter-round latent alignment.Extensive experiments demonstrate that VAE-LFA improves semantic consistency and visual fidelity across diverse multi-turn editing scenarios, including both controlled and in-the-wild images.