ScoreDial: Precise Control of Visual Attributes in Image Generation via Regressor Guidance from Limited Human Ratings
arXiv:2611.05004
Abstract
Requested-value control is stricter than continuous steering: given on a human-rating scale, the generated image should attain that value from only limited ratings. We present ScoreDial, which fits a differentiable regressor to 100-479 human-rated images; once is fixed, a requested value is defined on its output scale, and learned target conditioning is paired with error-adaptive guidance that repeatedly reduces during sampling. Across tire slip resistance, interior visual complexity, food desire-to-eat, and soap translucency, using one set of sampler settings, ScoreDial reaches requested values and achieves -MAE at most 0.013 over a range of width 0.8 and Spearman , the maximum attainable under our protocol. Because also drives sampling, we evaluate agreement with separately from matched sweeps, an independent auditor fitted to the same ratings on a backbone no deployed regressor uses, and blinded human judgments. MPGD-flow also reaches the requested values under , but ScoreDial has larger auditor displacement on every attribute, the largest among all compared methods, and a stronger MPGD-flow setting gives up target agreement without closing this gap. On every attribute, removing target conditioning leaves low -MAE but more than halves auditor displacement, separating precise score correction from attribute movement. Blinded rating studies with 15 raters per attribute show mean human ratings track on tire and soap, with linear and , and in forced-choice comparisons of matched sweeps, raters chose ScoreDial's as the clearer progression in at least 84% of judgments against every baseline. Code: https://github.com/Kamijo-Gen/ScoreDial.
arXiv commentAccepted to WACV 2027. 37 pages, 27 figures, 4 tables, including the supplementary material. v2: code link moved from the conclusion to the abstract; results unchanged