Re-rendered the Pixar-style portrait with naturally proportioned human eyes so VASA-1 can track them. Same face shape, hair, glasses, smile, shirt and blue studio backdrop — only the eye size and inter-pupil distance changed.
Source
otto_avatar.jpg
Model
gpt-image-2
Resolution
1024 × 1024
Variants
—
Constraint applied: each eye ≈ 1/5 face width · normal inter-pupil distance (≈ one eye-width apart) · natural almond shape · iris/eye-white ratio kept lively. Glasses, hair, smile, shirt, background untouched.