# AffectDelta: Beyond Emotion Labels for Image Editing

arXiv: https://arxiv.org/abs/2609.02616 · HTML: https://mlsanity.com/paper/2609.02616/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 35 references that are in the map.
- First posted on arXiv 2 Sep 2026.

## Builds on (most cited first, 18 of 18)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Prompt-to-Prompt Image Editing with Cross Attention Control](https://mlsanity.com/paper/2208.01626/) (2022)
- [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)
- [UltraEdit: Instruction-based Fine-Grained Image Editing at Scale](https://mlsanity.com/paper/2407.05282/) (2024)
- [EmoSet: A Large-scale Visual Emotion Dataset with Rich Attributes](https://mlsanity.com/paper/2307.07961/) (2023)
- [Building a Large Scale Dataset for Image Emotion Recognition: The Fine Print and The Benchmark](https://mlsanity.com/paper/1605.02677/) (2016)
- [GPT-IMAGE-EDIT-1.5M: A Million-Scale, GPT-Generated Image Dataset](https://mlsanity.com/paper/2507.21033/) (2025)
- [ImageBrush: Learning Visual In-Context Instructions for Exemplar-Based Image Manipulation](https://mlsanity.com/paper/2308.00906/) (2023)
- [EmoEdit: Evoking Emotions through Image Manipulation](https://mlsanity.com/paper/2405.12661/) (2024)
- [A Circular-Structured Representation for Visual Emotion Distribution Learning](https://mlsanity.com/paper/2106.12450/) (2021)
- [Affective Image Editing: Shaping Emotional Factors via Text Descriptions](https://mlsanity.com/paper/2505.18699/) (2025)
- [Towards Deeper Emotional Reflection: Crafting Affective Image Filters with Generative Priors](https://mlsanity.com/paper/2512.17376/) (2025)
- [Automatic Image Transformation for Inducing Affect](https://mlsanity.com/paper/1707.08148/) (2017)
- [EmoKGEdit: Training-free Affective Injection via Visual Cue Transformation](https://mlsanity.com/paper/2601.12326/) (2026)
- [EmoAgent: A Multi-Agent Framework for Diverse Affective Image Manipulation](https://mlsanity.com/paper/2503.11290/) (2025)
- [Moodifier: MLLM-Enhanced Emotion-Driven Image Editing](https://mlsanity.com/paper/2507.14024/) (2025)
- [MooD: Perception-Enhanced Efficient Affective Image Editing via Continuous Valence-Arousal Modeling](https://mlsanity.com/paper/2605.02521/) (2026)

## Built on by (most cited first, 1 of 1)

- [What We Talk About When We Talk About Emotion Editing](https://mlsanity.com/paper/2610.13558/) (2026)

## Abstract (arXiv)

Emotion-driven image editing aims to evoke a specified target emotion by modifying emotion-relevant visual cues in a source image, while preserving the overall composition and semantic-structural coherence of the original scene. Existing scene-level editors typically specify the target with a single emotion category and often learn visual transformations from operation-level text instructions. A category collapses a mixed affective endpoint into one dominant label, while language cannot precisely quantify how coexisting emotions should increase, decrease, or remain stable. We introduce AffectDelta, a source-aware editor that treats editing as a transition between eight-dimensional emotion distributions. A frozen Emotion Distribution Predictor estimates the source state, and the signed source-to-target difference encodes the direction and magnitude of the requested transition. Within AffectDelta, an internal transition encoder and a source-aware diffusion backbone jointly translate this signal into context-dependent semantic and appearance changes. To train this formulation, we construct AffectPair-249K, comprising 248,841 source-target pairs with predicted eight-dimensional distributions and spanning both cross-category and within-category transitions. Experiments against six baselines, combining quantitative evaluation with qualitative comparisons, demonstrate improved affective alignment and content preservation, while ablations validate our design choices. Code and dataset will be made publicly available upon acceptance.