mira-3dlisted

# Mira 3D One subject becomes one mesh. The model reads the prompt or the photo as a description of a single object seen from all sides, so everything that helps it understand the shape belongs in the request, and everything that describes a scene, a mood or a camera does not. ## When to use A character, a prop, a product or a piece of environment that has to exist as geometry: to drop into Blender, Unity, Unreal or a browser game, to print, to show in AR, to animate as a rigged humanoid. Anything where the deliverable is a file with polygons, not a picture. ## Doctrine This is a single-asset 3D generation: one subject, complete on all sides, standing alone on a neutral background with nothing else in frame. Describe the object as a physical thing: its overall silhouette, the parts it is made of, their proportions to each other, the materials and colours of each part, and the surface finish. Name the category first (a humanoid character, a creature, a vehicle, a prop, a piece of furniture, a building fragment) so the model knows what kind of shape to expect. For characters and creatures, ask for a symmetrical A-pose or T-pose with arms away from the body and legs apart, limbs fully visible, no held props overlapping the body, so the mesh can be rigged. For products and props, give the real-world size or the scale relative to a hand, and keep the label or logo geometry simple. Say nothing about lighting, camera angle, depth of field, motion or atmosphere: none of it exist