Color encoding in Latent Space of Stable Diffusion Models

• URL: http://arxiv.org/abs/2512.09477v1
• Date: Wed, 10 Dec 2025 09:54:03 GMT
• Title: Color encoding in Latent Space of Stable Diffusion Models
• Abstract summary: This work explores how color is encoded in a generative model through a systematic analysis of the latent representations in Stable Diffusion.<n>Our findings indicate that the latent space of Stable Diffusion exhibits an interpretable structure aligned with a efficient coding representation.
• Score: 0.05599792629509228
• License: http://creativecommons.org/publicdomain/zero/1.0/
• Abstract: Recent advances in diffusion-based generative models have achieved remarkable visual fidelity, yet a detailed understanding of how specific perceptual attributes - such as color and shape - are internally represented remains limited. This work explores how color is encoded in a generative model through a systematic analysis of the latent representations in Stable Diffusion. Through controlled synthetic datasets, principal component analysis (PCA) and similarity metrics, we reveal that color information is encoded along circular, opponent axes predominantly captured in latent channels c_3 and c_4, whereas intensity and shape are primarily represented in channels c_1 and c_2. Our findings indicate that the latent space of Stable Diffusion exhibits an interpretable structure aligned with a efficient coding representation. These insights provide a foundation for future work in model understanding, editing applications, and the design of more disentangled generative frameworks.

Related papers

• BC-IHV: Conditioning the Color Space for Stable Rectified-Flow Low-Light Enhancement [51.79298431282401]
  Low-light image enhancement must correct ambiguous exposure without overwriting structure already supported by the input.<n>We propose Structure-Anchored Rectified Flow (SA-RF), which maintains correspondence through separate chromaticity/intensity stems.<n>We also introduce BC-IHV, a learnable Box--Cox polar color space whose analytically invertible intensity mapping controls the inverse-gradient dynamic range through a single exponent.
  arXiv  Detail & Related papers  (2026-08-22T08:43:00Z)
• Deep Dreams Are Made of This: Visualizing Monosemantic Features in Diffusion Models [0.12277343096128711]
  This paper proposes latent visualization by optimization (LVO), a mechanistic interpretability technique that extends feature visualization by optimization to latent diffusion models.<n>LVO employs sparse autoencoders (SAEs) to disentangle polysemantic layer representations into monosemantic features.
  arXiv  Detail & Related papers  (2026-05-06T13:21:24Z)
• Color Matters: Demosaicing-Guided Color Correlation Training for Generalizable AI-Generated Image Detection [10.845173807400533]
  We propose a Demosaicing-guided Color Correlation Training framework for AI-generated image detection.<n>A self-supervised U-Net is trained to model the conditional distribution of the missing channels from the given one.<n>Our theoretical analysis reveals that DCCT targets a provable distributional difference in color-correlation features between photographic and AI-generated images.
  arXiv  Detail & Related papers  (2026-01-30T10:01:49Z)
• Latent Diffusion Model without Variational Autoencoder [78.34722551463223]
  SVG is a novel latent diffusion model without variational autoencoders for visual generation.<n>It constructs a feature space with clear semantic discriminability by leveraging frozen DINO features.<n>It enables accelerated diffusion training, supports few-step sampling, and improves generative quality.
  arXiv  Detail & Related papers  (2025-10-17T04:17:44Z)
• Latent Diffusion U-Net Representations Contain Positional Embeddings and Anomalies [2.1261727383260043]
  We analyze popular Stable Diffusion models using representational similarity and norms.<n>Our findings reveal three phenomena: (1) the presence of a learned positional embedding in intermediate representations, (2) high-similarity corner artifacts, and (3) anomalous high-norm artifacts.
  arXiv  Detail & Related papers  (2025-04-09T16:26:26Z)
• Leveraging Semantic Attribute Binding for Free-Lunch Color Control in Diffusion Models [53.73253164099701]
  We introduce ColorWave, a training-free approach that achieves exact RGB-level color control in diffusion models without fine-tuning.<n>We demonstrate that ColorWave establishes a new paradigm for structured, color-consistent diffusion-based image synthesis.
  arXiv  Detail & Related papers  (2025-03-12T21:49:52Z)
• Color Alignment in Diffusion [29.15171578869268]
  Diffusion models have shown great promise in synthesizing visually appealing images.<n>We introduce a novel color alignment algorithm that confines the generative process in diffusion models within a given color pattern.<n>Results demonstrate our state-of-the-art performance in conditioning and controlling of color pixels, while maintaining on-par generation quality and diversity.
  arXiv  Detail & Related papers  (2025-03-09T20:02:52Z)
• Derivative-Free Diffusion Manifold-Constrained Gradient for Unified XAI [59.96044730204345]
  We introduce Derivative-Free Diffusion Manifold-Constrainted Gradients (FreeMCG)<n>FreeMCG serves as an improved basis for explainability of a given neural network.<n>We show that our method yields state-of-the-art results while preserving the essential properties expected of XAI tools.
  arXiv  Detail & Related papers  (2024-11-22T11:15:14Z)
• Diffusion Model with Cross Attention as an Inductive Bias for Disentanglement [58.9768112704998]
  Disentangled representation learning strives to extract the intrinsic factors within observed data. We introduce a new perspective and framework, demonstrating that diffusion models with cross-attention can serve as a powerful inductive bias. This is the first work to reveal the potent disentanglement capability of diffusion models with cross-attention, requiring no complex designs.
  arXiv  Detail & Related papers  (2024-02-15T05:07:54Z)
• Navigating the Structured What-If Spaces: Counterfactual Generation via Structured Diffusion [20.20945739504847]
  We introduce Structured Counterfactual diffuser or SCD, the first plug-and-play framework leveraging diffusion for generating counterfactual explanations in structured data. Our experiments show that our counterfactuals not only exhibit high plausibility compared to the existing state-of-the-art but also show significantly better proximity and diversity.
  arXiv  Detail & Related papers  (2023-12-21T07:05:21Z)
• ToddlerDiffusion: Interactive Structured Image Generation with Cascaded Schrödinger Bridge [63.00793292863]
  ToddlerDiffusion is a novel approach to decomposing the complex task of RGB image generation into simpler, interpretable stages. Our method, termed ToddlerDiffusion, cascades modality-specific models, each responsible for generating an intermediate representation. ToddlerDiffusion consistently outperforms state-of-the-art methods.
  arXiv  Detail & Related papers  (2023-11-24T15:20:01Z)
• Diffusion Models as Masked Autoencoders [52.442717717898056]
  We revisit generatively pre-training visual representations in light of recent interest in denoising diffusion models. While directly pre-training with diffusion models does not produce strong representations, we condition diffusion models on masked input and formulate diffusion models as masked autoencoders (DiffMAE) We perform a comprehensive study on the pros and cons of design choices and build connections between diffusion models and masked autoencoders.
  arXiv  Detail & Related papers  (2023-04-06T17:59:56Z)
• Enhanced Controllability of Diffusion Models via Feature Disentanglement and Realism-Enhanced Sampling Methods [27.014858633903867]
  We present a training framework for feature disentanglement of Diffusion Models (FDiff)<n>We propose two sampling methods that can boost the realism of our Diffusion Models and also enhance the controllability.
  arXiv  Detail & Related papers  (2023-02-28T07:43:00Z)
• Unifying Diffusion Models' Latent Space, with Applications to CycleDiffusion and Guidance [95.12230117950232]
  We show that a common latent space emerges from two diffusion models trained independently on related domains. Applying CycleDiffusion to text-to-image diffusion models, we show that large-scale text-to-image diffusion models can be used as zero-shot image-to-image editors.
  arXiv  Detail & Related papers  (2022-10-11T15:53:52Z)
• Joint Intensity-Gradient Guided Generative Modeling for Colorization [16.89777347891486]
  This paper proposes an iterative generative model for solving the automatic colorization problem. Joint intensity-gradient constraint in data-fidelity term is proposed to limit the degree of freedom within generative model. Experiments demonstrated that the system outperformed state-of-the-art methods whether in quantitative comparisons or user study.
  arXiv  Detail & Related papers  (2020-12-28T07:52:55Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.