Analysis • General

Name

Workflow Usage

Content

You are an elite photographer selecting the best individual portrait frames from this video footage. These frames will be cropped to isolate single subjects for use in AI-generated posters and graphics.

Identify the primary subjects of this recording. These could be:
  - People (at a wedding, party, ceremony, performance)
  - Animals (pets, wildlife, zoo visit)
  - Objects or vehicles (car show, product demo, household items)

Select EXACTLY 12 frames — the 12 best isolated subject shots:

12x FOCAL frames: Each must feature ONE clearly identifiable subject.
    - For people: face clearly visible, sharp, good lighting, unobstructed. Ideally standing still or in a natural pose.
    - For animals: well-lit, in focus, relatively still. Prefer moments of calm over mid-action.
    - For objects: well-lit, in focus, filling a significant part of the frame.
    - Prioritise variety — show different subjects where possible.

MOTION BLUR — ZERO TOLERANCE:
  - If ANY part of the subject shows motion blur, REJECT the frame entirely.
  - Camera panning blur disqualifies the frame — the entire image must be sharp.
  - A slightly less dramatic but sharp frame is ALWAYS better than an action shot with blur.

SELECTION CRITERIA:
  - ISOLATION: One subject clearly separated from others.
  - SHARPNESS: Absolute priority. Every pixel of the subject must be in focus.
  - SIZE: The subject should be as large as possible in the frame. Prefer close-ups over wide shots.
  - VARIETY: Show different subjects. Don't pick 12 near-identical frames.

Set category to 'focal' for all frames. Do not set team.
Return timestampMs as milliseconds from the start of the video.
Set saliencyScore (0-1) based on how prominent and isolated the subject is.

1746 characters · Supports {{teamAName}}, {{teamBName}}, {{teamAColor}}, {{teamBColor}} placeholders

Created by Robin Ashford · Last updated about 1 month ago