Lead with view and subject
Identify the intended view and visible subject: for example, a strict 2D side-view knight facing right with a long spear. Include identity-defining props when they matter to the animation.
The most useful animation prompt explains what moves, in what order, and how the subject interacts with the ground, a weapon, or another object. Style constraints should support the action instead of replacing it.
Identify the intended view and visible subject: for example, a strict 2D side-view knight facing right with a long spear. Include identity-defining props when they matter to the animation.
For a multi-step move, describe anticipation, execution, follow-through, and recovery. Timed segments can help a longer animation, but each segment should advance the action.
Specify planted feet, steps, jumps, body lean, hand placement, weapon path, and momentum where relevant. This gives the model actionable spatial information.
Request trails, sparks, smoke, magic, or impact effects only when you want them. Simple solid-color effects read more clearly in small sprites than translucent gradients.
No. A longer prompt helps only when it adds useful motion, timing, or identity information. Repetition and conflicting instructions can make the result worse.
No. Include effects only when they are part of the intended action, and describe their color, shape, and timing concisely.
Describe transitions as continuous movement, avoid asking the character to hold multiple poses, and allow enough duration for the action stages.