NumColor: Precise Numeric Color Control in Text-to-Image Generation

• URL: http://arxiv.org/abs/2603.13547v1
• Date: Fri, 13 Mar 2026 19:37:25 GMT
• Title: NumColor: Precise Numeric Color Control in Text-to-Image Generation
• Abstract summary: Text-to-image models excel at generating images from natural language descriptions, yet fail to interpret numerical colors.<n>We present NumColor, that enables precise numerical color control across multiple diffusion architectures.<n> NumColor improves numerical color accuracy by 4-9x across five models, while simultaneously improving color harmony scores by 10-30x on GenColorBench benchmark.
• Score: 57.72106507860579
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Text-to-image diffusion models excel at generating images from natural language descriptions, yet fail to interpret numerical colors such as hex codes (#FF5733) and RGB values (rgb(255,87,51)). This limitation stems from subword tokenization, which fragments color codes into semantically meaningless tokens that text encoders cannot map to coherent color representations. We present NumColor, that enables precise numerical color control across multiple diffusion architectures. NumColor comprises two components: a Color Token Aggregator that detects color specifications regardless of tokenization, and a ColorBook containing 6,707 learnable embeddings that map colors to embedding space of text encoder in perceptually uniform CIE Lab space. We introduce two auxiliary losses, directional alignment and interpolation consistency, to enforce geometric correspondence between Lab and embedding spaces, enabling smooth color interpolation. To train the ColorBook, we construct NumColor-Data, a synthetic dataset of 500K rendered images with unambiguous color-to-pixel correspondence, eliminating the annotation ambiguity inherent in photographic datasets. Although trained solely on FLUX, NumColor transfers zero-shot to SD3, SD3.5, PixArt-α, and PixArt-Σ without model-specific adaptation. NumColor improves numerical color accuracy by 4-9x across five models, while simultaneously improving color harmony scores by 10-30x on GenColorBench benchmark.

Related papers

• Recolour What Matters: Region-Aware Colour Editing via Token-Level Diffusion [43.72039047537043]
  Colour is one of the most perceptually salient yet least controllable attributes in image generation.<n>ColourCrafter transforms colour editing from global tone transfer into a structured, region-aware generation process.<n>We build ColourfulSet, a largescale dataset of high-quality image pairs with continuous and diverse colour variations.
  arXiv  Detail & Related papers  (2026-03-19T03:53:38Z)
• Controllable-Continuous Color Editing in Diffusion Model via Color Mapping [73.62340517056619]
  We introduce a color mapping module that explicitly models the correspondence between the text embedding space and image RGB values.<n>Users can specify a target RGB range to generate images with continuous color variations within the desired range.<n> Experimental results demonstrate that our method performs well in terms of color continuity and controllability.
  arXiv  Detail & Related papers  (2025-09-17T07:12:51Z)
• Paint Bucket Colorization Using Anime Character Color Design Sheets [72.66788521378864]
  We introduce inclusion matching, which allows the network to understand the relationships between segments. Our network's training pipeline significantly improves performance in both colorization and consecutive frame colorization. To support our network's training, we have developed a unique dataset named PaintBucket-Character.
  arXiv  Detail & Related papers  (2024-10-25T09:33:27Z)
• Color Space Learning for Cross-Color Person Re-Identification [31.110110147991033]
  We propose Color Space Learning (CSL) for those Cross-Color Person ReID problems. CSL guides the model to be less color-sensitive with two modules: Image-level Color-Augmentation and Pixel-level Color-Transformation. Our method surpasses the state-of-the-art methods consistently.
  arXiv  Detail & Related papers  (2024-05-15T16:26:06Z)
• Control Color: Multimodal Diffusion-based Interactive Image Colorization [81.68817300796644]
  Control Color (Ctrl Color) is a multi-modal colorization method that leverages the pre-trained Stable Diffusion (SD) model. We present an effective way to encode user strokes to enable precise local color manipulation. We also introduce a novel module based on self-attention and a content-guided deformable autoencoder to address the long-standing issues of color overflow and inaccurate coloring.
  arXiv  Detail & Related papers  (2024-02-16T17:51:13Z)
• Learning to Structure an Image with Few Colors and Beyond [59.34619548026885]
  We propose a color quantization network, ColorCNN, which learns to structure an image in limited color spaces by minimizing the classification loss. We introduce ColorCNN+, which supports multiple color space size configurations, and addresses the previous issues of poor recognition accuracy and undesirable visual fidelity under large color spaces. For potential applications, we show that ColorCNNs can be used as image compression methods for network recognition.
  arXiv  Detail & Related papers  (2022-08-17T17:59:15Z)
• The Utility of Decorrelating Colour Spaces in Vector Quantised Variational Autoencoders [1.7792264784100689]
  We propose colour space conversion to enforce a network learning structured representations. We trained several instances of VQ-VAE whose input is an image in one colour space, and its output in another.
  arXiv  Detail & Related papers  (2020-09-30T07:44:01Z)
• Learning to Structure an Image with Few Colors [59.34619548026885]
  We propose a color quantization network, ColorCNN, which learns to structure the images from the classification loss in an end-to-end manner. With only a 1-bit color space (i.e., two colors), the proposed network achieves 82.1% top-1 accuracy on the CIFAR10 dataset. For applications, when encoded with PNG, the proposed color quantization shows superiority over other image compression methods in the extremely low bit-rate regime.
  arXiv  Detail & Related papers  (2020-03-17T17:56:15Z)

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.