Google Flow 50th Reunion Character Prompts - keeping the face exactly the same in the video

I’m producing a short, upbeat video for my 50-year high-school reunion and want every classmate to recognise themselves immediately. From each 1976 yearbook head-shot I need to generate a full-body Google Flow character whose face stays virtually identical to the photo, then place that character in ten-second clips—some simple outdoor strolls and conversations, others light sports moments like passing a basketball or tossing a baseball.

Your job is to give me fool-proof Google Flow text-to-video prompts (and any complementary settings, seeds, or negative prompts) that lock the facial likeness in place through motion, while dressing the characters in unmistakable 1976-era clothes and hair styles. I also need clear, repeatable steps so I can apply the same workflow to dozens of portraits without losing consistency from clip to clip.

Please deliver:
• A reusable prompt template for portrait-to-character conversion that preserves facial features with minimal drift.
• At least three scene prompts: an outdoor sidewalk walk-and-chat, a friendly basketball exchange, and a casual baseball throw.
• Any critical Google Flow parameters (image weight, frame length, resolution, etc.) and troubleshooting tips for refining likeness if the model starts to wander.
• A brief set of batch-processing instructions so I can scale this to the whole graduating class.

Acceptance criteria: when I run your prompts on a clean yearbook scan, the resulting 1080p, 24-fps clip must show a face that classmates instantly recognise, era-appropriate styling, and smooth motion throughout the ten-second scene.