cs.CVNov 9, 2026

Continuous Angular Control in Text-to-Image Generation via Geometric Token Encoding

Authors: Furkan Güzelant, Basavaraj Sunagad, Aysegul Dundar, Adam Kortylewski

Organizations: Bilkent University, Turkey · CISPA Helmholtz Center for Information Security, Germany · University of Freiburg, Germany

Abstract

Text-to-image diffusion models can render a photorealistic analog clock but cannot control the time it displays. Across four pretrained models, fewer than 1% of generated clocks place both hands within 6 degrees of the requested time; the models instead collapse onto visually dominant configurations such as the canonical 10:10 pose. The same failure appears when a specific fraction of a circular object is requested, where models return complete or conventionally sliced objects. We observe that these two failures share a geometric structure: both target quantities can be parameterized by angular variables. We introduce a geometric token encoding framework that conditions generation directly on these angles. Each angular component is encoded as a periodic sine-cosine feature and projected into a role-specific token embedding that is injected into the prompt before a frozen text encoder. To spatially ground these tokens without requiring localization input at inference, we introduce Stochastic Localization Relaxation (SLR), which stochastically relaxes attention supervision from component-level to object-level to unconstrained localization during training. For clocks, we additionally introduce a hand-swap preference objective that discourages assigning the correct angles to the wrong hand roles. On analog clocks our method reaches 63-78% Acc@6 degrees across evaluation methods, compared with below 1% for the baselines. On circular-fraction generation, it achieves 89.5% [email protected], while maintaining 88.8% on unseen object categories and 84.9% on unseen target values, compared with roughly 20% for pretrained baselines.

Figures & tables

Appendix figures & tables11 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Camera Control for Text-to-Image Generation via Learning Viewpoint Tokens

    Apr 21, 2026Xinxuan Lu, Charless Fowlkes, Alexander C. BergT2I GenerationControllable Image Generation

  2. Improving Controllable Generation: Faster Training and Better Performance via x0x_0-Supervision

    Apr 7, 2026Amadou S. Sangare, Adrien Maglo, Mohamed Chaouch +1Controllable Image GenerationConditional Diffusion Models

  3. Concept Guidance: Precise, Training-Free Latent Control for Text-to-Image Generation

    Aug 14, 2026Nikolai Röhrich, Isabell Hans, Felix Krause +1Diffusion Model GuidanceT2I Generation