Prompt para poner un rollerskate: acción y coherencia física

Creando Prompts para Generación de Video con IA

Hoy vamos a desglosar cómo se construye un prompt detallado en inglés para una acción específica: alguien sentado, recogiendo un rollerskate del suelo y colocándoselo en el pie derecho descalzo. El objetivo es garantizar que la física sea coherente y realista.

El Prompt Detallado

Aquí tienes el texto exacto para generar la acción:

A person is sitting comfortably on a low bench, with their right foot bare and resting on the floor. A single roller skate lies on the floor directly in front of them. The person leans forward naturally, bends at the waist, and reaches down with both hands toward the roller skate. They grasp the skate firmly by the boot, lift it off the floor, and bring it closer to their right foot. Keeping their balance in a natural seated position, the person raises their bare right foot slightly off the floor and guides it toward the opening of the roller skate. They hold the skate steady with both hands, carefully align the toes with the boot opening, and gradually slide their right foot inside. Their toes enter first, followed by the ball of the foot and then the heel. The person gently adjusts the position of the skate and uses their fingers to widen the opening slightly, helping the heel settle fully into the boot. Once the foot is inside, the person presses their heel down into the skate, makes small, precise adjustments to ensure a snug fit, and briefly checks the alignment of their foot. They then lower the roller skate toward the floor, keeping their right leg relaxed and slightly bent. The entire action is performed slowly and deliberately, with realistic human anatomy, natural hand movements, believable foot articulation, accurate physical contact, and consistent object positioning. Single continuous shot, fixed camera, medium close-up framing focused on the person’s hands, bare right foot, roller skate, and lower legs. Realistic body mechanics, smooth sequential movements, consistent lighting, stable proportions, detailed skin texture, and physically accurate interaction between the hands, foot, and skate. No sudden movements, no teleportation, no object deformation, no extra limbs, no changes in footwear, and no cuts.

Estrategia para Mayor Coherencia Física

Un detalle crucial para evitar errores en la generación de video es dividir la acción en dos etapas. Primero se recoge el patín y se acerca al pie; después se introduce el pie y se asienta el talón. Esta separación ayuda a mantener la coherencia temporal y evita que los modelos cometan deformaciones o fallos de articulación.