CHARACTER: Canvas 1024x1024 transparent PNG.
Style: Care Bears-style plush teddy turned horror, thick dark outline, flat cel shading, stitched seams, slightly matted fur clumps, menacing brow.
Pose: the BODY is turned about 40 degrees so the chest points to the viewer's left. The HEAD faces the viewer straight on (creator's request), both eyes visible and level. The head is 1.5x the template size: crown y=150, eye line y=380 with eyes at x=474 and x=634, chin/jaw bottom y=610, head 435px wide centred on x=554 (x=337 to x=771).
Head shape is round and plush, not a human skull. Two round ears sit on the top corners of the head, around x=380 and x=728, reaching up to about y=110.
Muzzle: a lighter oval patch centred x=554, y=470, with a stitched nose at y=445. Leave the mouth area blank: no mouth and no teeth, because the mouth is a separate layer.
Neck: short and thick, narrowest at y=630 and meeting the shoulders at y=660 between x=470 and x=640.
Torso: chunky plush belly with a lighter belly patch (no symbol). It runs straight off the bottom edge at y=1024 with no bottom outline and no empty space below it.
Arms: the bear's arm on the viewer's left hangs relaxed at the side. The arm on the viewer's right is raised out to the side: upper arm angled away from the torso, forearm upright. The paw is centred at x=880, y=600 in a loose, empty curled grip.
Paw and forearm stay fully clear of the chest and belly. Do not draw any held item, hat, glasses, clothing or jewellery. Use one neutral mid-tone fur colour.
Output: character/base.png.
Weights: character/weights.json {"base.png":1}.
BACKGROUND: Canvas 1024x1024 opaque PNG.
Style: same flat cel-shaded cartoon horror, thick outlines, muted desaturated palette with one sickly accent colour. Draw full scenes, since the creator asked for locations. Keep the centre column x=300 to 810 and the area from y=100 down low in detail and contrast, because the bear covers it.
Put the scene detail at the edges and the horizon around y=700. No characters, no text.
Outputs: background/bg_graveyard.png (crooked tombstones, dead trees, fog, full moon), background/bg_school.png (abandoned classroom, chalkboard, toppled desks, flickering lights), background/bg_fairground.png (rusted ferris wheel and carousel at night, broken bulbs), background/bg_hell.png (red-orange cavern, lava glow, jagged rock), background/bg_haunted_mansion.png (gothic mansion on a hill, lit windows, storm sky), background/bg_toy_factory.png (abandoned toy factory, conveyor of teddy parts, dim green light).
Weights: background/weights.json, each file weight 1. Also write artifacts/base_b.png and artifacts/base_c.png: two more takes of the SAME character on the SAME template and pose, each a SEPARATE, FRESHLY GENERATED image (run the image tool again for each one; never copy, edit, recolour or re-save base.png), so that if one comes out too big, too small or off-centre another may fit.
Write artifacts/base.png, artifacts/background_01.png .. background_06.png, and artifacts/traits.json: {"backgrounds":[{"file","name","weight"}...]} with weights summing to 100.
