Faces detected. Anything else, on your call.
Plekto marks the faces. You mark the rest.
Drop the still in. Plekto detects every face and pre-marks each one as a region. From there, add your own regions for hands, props, backgrounds, anything in the image you want to direct. Each region is independent, and the rest of the still stays exactly where you left it. Plekto routes each region through Gemini, Replicate, or Fal, then composites the result back at full resolution with the seams gone.