Compare output against the approved take
Same-Shot Score is a review record for how closely a generated result follows the approved prompt, selected references, route, and continuity expectations. It is meant to support decisions, not replace human approval.
Look at the right differences
A score can consider prompt alignment, reference preservation, chronology, visual continuity, model settings, provider request differences, and whether proxy and master outputs stayed on the same shot.
Turn misses into better next takes
When a take fails, the score and feedback should help identify whether the problem was the script, prompt rewrite, reference lane, model capability, provider drift, or result review. That evidence is what makes future fixes reusable.