Sampling that mixes concept and superclass guidance improves prompt adherence in personalized text-to-image models, with simple mixed guidance often outperforming more complex switching and masking strategies.
High-resolution image synthesis with latent diffusion models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Beyond Fine-Tuning: A Systematic Study of Sampling Techniques in Personalized Image Generation
Sampling that mixes concept and superclass guidance improves prompt adherence in personalized text-to-image models, with simple mixed guidance often outperforming more complex switching and masking strategies.