From First Image to Final Short: How FourFeetz Plans a Character Scene

The published HARU and LUNA rainy-window short began with a quiet shared pause, not with a request for impressive movement. The scene needed to introduce both characters, keep the rain outside the window and leave enough visual calm for the original music. That purpose determined the reference frame, camera distance and vertical composition before video generation began.

We use an approved character image or an actual frame from the sequence as visible evidence. For this short, HARU and LUNA had to remain readable together at phone size. We checked their relative scale, face direction, window placement and the open space around their ears and paws. This is the public decision process; the finished character prompts and internal reference system remain private.

When a short shot is asked to introduce the room, move two animals, change their expressions and move the camera, every layer can drift at once. Different reaction timing can become synchronized, a paw can lose contact with the floor, or the rainy window can be reconstructed as a different background. A visually polished result is still rejected when those changes weaken the scene connection.

We also avoid choosing a reference only because it is the prettiest image. A close portrait may protect the face but provide no room for the planned action; a wide frame may show the environment but make a name tag or expression unreadable. The correct First Shot is the one that supports the intended edit.

Our stable order is reference approval, scene purpose, one observable action, motion review and final edit. In the rainy-window short, restrained glances and small posture changes gave each character separate timing without turning the moment into a performance. The generated material was judged beside the preceding and following shots, not as an isolated clip.

CapCut was used for selecting usable ranges, trimming starts and endings, arranging the scene order and balancing the original music. Editing can improve rhythm and make a valid transition feel intentional. It cannot restore a changed face, missing paw or broken window geometry, so clips with structural failures do not advance to finishing.

A First Shot is not complete because it looks cinematic. It is complete when it protects the character, makes the planned action physically possible and can connect to the neighboring material. Reviewing at phone size catches different problems from reviewing a full-resolution frame, so FourFeetz uses both views before accepting a scene.

Creators can apply the same method without a private prompt system: define the scene purpose, mark identity details that must remain visible, limit the action and test the result inside the final aspect ratio. The finished short, character pages and production notes provide honest reference points for what survived the process.