The Uncanny Valley of AI-Generated Faces: What Makes Them Feel Off

Temporal consistency remains one of the biggest challenges in AI video. While single-frame quality has improved, maintaining coherent motion across frames still leads to flicker, morphing, and inconsistent lighting. We analyze recent techniques—optical flow conditioning, temporal attention, and frame interpolation—and assess how well they address these issues.