Generate a character pose from a production-ready brief

Start with identity details that can be checked visually. Adult courier is too broad. Adult courier in her early thirties, angular face, short black bob with one silver forelock, athletic medium build, mustard cropped rain jacket, dark teal shirt, charcoal cargo trousers, black ankle boots, and a muted red cross-body satchel gives the generation a bounded design vocabulary. The description does not need fictional biography unless that biography changes the visible body, clothing, prop, or posture.

Order the anchor from the character outward. Name age range and presentation, then face and hair, then build, then garments from outer layer to footwear, then a signature object, and finally the palette. Concrete nouns are more diagnostic than stacks of style adjectives. If the jacket disappears, you know which clause failed. If the prompt only says cinematic, dynamic, premium, expressive character, there is little visual evidence to compare.

Keep pose instructions out of the identity anchor. Writing crouching courier with windblown jacket in the identity field makes it harder to reuse the character for a standing or guarding pose. Stable details belong in the anchor; changeable performance belongs in the action and balance controls. This separation also reduces contradictions such as relaxed posture plus airborne leap or both hands visible plus both hands behind the back.

Text continuity has limits. Separate image generations can change facial proportions, hair part, garment closures, pocket placement, shoe details, and prop shape even when the words stay identical. Treat the anchor as a review checklist, not a mathematical lock. If exact identity is required across a sequence, approve a character design in a dedicated character workflow and use manual drawing, controlled reference tools, or post-production methods that genuinely support identity preservation.

A believable static pose has a support area: the space under and between the body parts touching the ground. In contrapposto, most weight travels through one standing leg, the pelvis shifts, the free leg relaxes, and the shoulders counterbalance. In a low crouch, bent legs widen or lower the support. If the projected center of mass falls far outside that area without a step, brace, or external force, the figure looks accidentally unstable.

Forward momentum is different from a stable lean. During acceleration, the center of mass can lead the current support because the rear leg is pushing and the next footfall is coming. The body should show that chain: torso direction, arm swing, hip drive, knee lift, push-off foot, and clothing or prop movement. If the torso leans forward but both feet are planted symmetrically, the pose often feels like a tilted cutout.

Controlled imbalance is useful for dodging, slipping, recovering, reaching, or preparing a directional change. It needs an explanation. A foot may pivot, a hand may brace, the torso may counter-rotate, or an arm may extend opposite the falling direction. Airborne action removes the support area entirely, so the figure's limbs and loose clothing should share one momentum rather than pointing in unrelated directions.

Review ground contact closely. Feet should meet a coherent floor plane, not float at different heights. The ankle should align with the direction of pressure. A planted boot should not bend like fabric, and a pointed toe should not carry the full body unless the action supports it. The generated background is intentionally plain so these mechanical relationships are easier to see.

The square, portrait, and landscape options guide composition inside the generated image; they do not change the model's configured export dimensions. Portrait framing suits upright standing, guarding, and vertical jumps. Landscape framing provides lead room for a sprint, reach, fall, or horizontal leap. Square framing is a neutral review canvas. In every case, the prompt asks for one complete figure with required hands and feet inside the frame.

Lead room matters when the action has direction. A runner needs space in front of the face and driving knee. A reaching character needs visible space toward the target, even if the target itself remains off-frame. Centering every action can make motion feel trapped. Conversely, leaving too much empty space can shrink the figure and make hands, face, and costume details difficult to inspect.

Cropping is not automatically wrong in finished comic art, but it is risky during pose development. A cropped boot hides whether the support is believable. A cropped hand hides finger and grip errors. A close-up may be useful for an expression tool, but this page targets full-body action planning. First approve the complete pose; crop it later in the panel composition where the story reason is clear.

Use the AI character creator when the main question is who the character is: face, hair, clothing, palette, role, and baseline visual direction. Use a character sheet workflow when you need an organized reference with approved views, proportion notes, and recurring design evidence. Use an expression sheet when the body can remain stable and the task is facial performance across emotions.

Use AI camera angle control when the main problem is where the viewer sits in relation to an existing scene concept. Camera angle can influence pose, but it does not replace pose direction. This page includes a camera-view choice only to make the selected body action readable. It does not manage a multi-character scene, background staging, lens package, shot sequence, or panel continuity.

After approving a pose concept, move to the comic panel generator when you need environment, other characters, props, lighting, and story context in a finished frame. Use the speech-bubble generator after the visual composition has protected readable negative space. Keeping these tasks on separate canonical pages prevents vague all-in-one prompts and makes each revision easier to diagnose.