ONE-STEP
1.4B DiT
VISION-ONLY
VOSR 2.0
Recover fine structures from dense visual cues.
01
Tiny Face
Facial details preserved within a crowded, large-scale scene.
02
Tiny Text Chinese
Cleaner character strokes and improved readability.
03
Tiny Text English
More legible letterforms across dense handwritten text.
04
Building
Sharper geometric structures, windows, and distant lights.
05
Landscape
Clearer foliage and natural high-frequency details across the scene.