MUSAR trains multi-subject text-to-image customization from a single-subject dataset by synthesizing diptych pairs and routing each image region's attention to the correct reference subject.
Scaling rectified flow transform- ers for high-resolution image synthesis
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
-
MUSAR: Exploring Multi-Subject Customization from Single-Subject Dataset via Attention Routing
MUSAR trains multi-subject text-to-image customization from a single-subject dataset by synthesizing diptych pairs and routing each image region's attention to the correct reference subject.