It is not bad, I suppose.

Here is prompt.

subject_definitions:
<Subject 1> is the girl depicted in <Picture 1>. <Subject 1> is wearing white panties. <Picture 1> is the sole and only visual reference for this character. Preserve <Subject 1>'s appearance throughout the entire target video, including facial structure, head, skin or body coloration, eye appearance, body proportions, silhouette, clothing or visible accessories, textures, and every other visible characteristic.
<Subject 2> is the trap environment generated for the target action: a concealed pitfall opening in the ground containing a dense field of rigid pointed stakes at the bottom. The trap is a newly generated environmental element and is not a visual reference for <Subject 1>.
summary:
[reference generation] The target video is a 10-second I2V sequence using only <Picture 1> as the single visual reference for <Subject 1>. <Subject 1>'s appearance must remain fully consistent with <Picture 1> throughout the entire sequence while the character falls into a concealed pitfall trap, collides with the dense stakes inside the pit, becomes immobilized with the stakes visibly surrounding and pinning the body while leaving the head unobstructed, and is having died at the end.
retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 3], [Shot 4]): identity_preserved - Only pose, orientation, position, and physical state change as required by the action.
<Subject 2> (appears in [Shot 2], [Shot 3], [Shot 4]): fully_preserved - the generated pitfall, ground opening, and dense pointed stakes remain spatially consistent throughout the sequence.
detailed_description:
The target video uses a cinematic dark-fantasy action style with physically coherent motion, dramatic but readable lighting, and strong temporal continuity. <Picture 1> is the sole character reference and must be treated as an immutable appearance anchor.
[Shot 1] The video opens with <Subject 1> standing or moving across a seemingly ordinary patch of ground. The camera shows the character clearly enough to establish the exact appearance from <Picture 1>, including the head, face, body silhouette, coloration, and all visible details. The surrounding ground appears stable and natural, with no obvious indication of the trap. The camera follows <Subject 1> from a slightly elevated three-quarter view as the character approaches the concealed trap. At approximately 00:01.500, the ground suddenly gives way directly beneath <Subject 1>. The character looks surprised. The clothing beneath the character's outfit rides up, revealing white panties. The character loses support and drops vertically into the hidden pit, with gravity-driven body motion and no change to the character's appearance.
[Shot 2] At 00:02.300, the camera rapidly follows the falling <Subject 1> downward into the pit. <Subject 2> is revealed clearly: a deep rectangular or irregularly shaped pit containing a dense arrangement of rigid pointed stakes extending upward from the bottom. The stakes are positioned at different locations and angles, leaving the head's eventual position unobstructed. Around the 00:03.000 mark, <Subject 1> reaches the bottom, and a stake pierces their back. Stakes pierce <Subject 1>'s chest, both thighs, the area between the breasts, both arms, and abdomen, causing bleeding; <Subject 1> screams and grimaces in agony. The impact of the penetration brings the fall to an abrupt halt, and the resulting force causes the body and limbs to spread out naturally.
[Shot 3] At 00:04.000, the camera settles into a medium-wide view from inside or near the edge of the pit. <Subject 1> is visibly immobilized by numerous stakes penetrating and pinning the non-head portions of the body. Blood flows from the spot where the stake has pierced the body, and the character wears a look of agony while continuing to groan. The head remains clearly visible and unobstructed. The character's original face, colors, proportions, silhouette, and other visual traits remain identical to <Picture 1>; only the pose and physical positioning have changed. The stakes remain rigid and stationary while <Subject 1> makes a brief unsuccessful movement attempting to escape. The camera slowly pushes closer, emphasizing the defeated state without changing the character design.
[Shot 4] At 00:07.000, the camera holds on the trapped <Subject 1>. The character gradually ceases moving, the highlights in her eyes slowly fade, and she moans in agony, signaling that she has been completely subdued by the trap. The stakes remain fixed in place and the environment stays physically consistent. At approximately 00:09.000, the camera makes a slow final push toward the unobstructed head and upper body. <Subject 1> remains completely recognizable as the exact character from <Picture 1>. In the final frame at the 00:10.000 mark, the image shows a girl who has become trapped in a hole, having died with her eyes closed, with no additional characters, transformations, costume changes, or visual redesigns.
overall_soundscape:
A dark outdoor ambience continues throughout the video, with subtle wind and environmental room tone. Diegetic sounds include the sudden cracking and collapsing of the ground, the rapid rush of air during the fall, a heavy impact as the character reaches the trap, rigid stake-impact sounds, brief physical movement against the trap, and a final quiet environmental ambience as the girl becomes motionless.
non_diegetic_music:
N/A. No background music is audible to the audience.