a skill by agentara, brought here by kt
fluffy poofy 3d characters
paste this link into your ai. it will know what to do.
Generate original fluffy, poofy 3D animation characters in a cohesive collectible-avatar family: monolithic plush body, inset smooth face, tiny limbs, restrained expression, tactile materials, strong silhouettes, and high variation without losing the shared design language.
Fluffy & Poofy 3D Characters
Use this skill whenever the user asks to create, vary, expand, or art-direct characters in this avatar family.
Core Design Language
One-line rule: monolithic poofy body + smooth inset face + tiny limbs + restrained expression + tactile plush material + unmistakable silhouette.
1. Silhouette
- One dominant rounded body mass; head and torso often merge into one shape.
- Short, heavy, simplified limbs; no anatomical fingers or toes unless specifically requested.
- A character must remain recognizable as a solid silhouette.
- Proportions should feel toy-like and huggable rather than human or mascot-costume realistic.
2. Face Architecture
- Face is a clean, smooth, matte panel embedded inside the fuzzy body, like a small opening in a plush shell.
- Keep the face visually quieter than the body.
- Default features: two small black dot eyes, tiny curved mouth, subtle blush, no visible nose.
- Expressions are calm and low-amplitude; avoid exaggerated emoji faces.
3. Material Contrast
- Body: dense short fur, fleece, velvet, boucle, shag, or another tactile soft surface.
- Face: smooth matte material with warm skin/ivory tone.
- The contrast between fuzzy shell and clean face is a defining feature.
- Avoid glossy plastic CGI, hard rubber, or metallic toy surfaces unless explicitly requested.
4. Shape Language
- Rounded, continuous, low-detail geometry.
- No sharp mechanical segmentation.
- Ears, horns, tails, fins, antennae, sprouts, wings, etc. should be simplified into soft toy-like primitives.
- Species inspiration is allowed, but the result should feel like its own creature rather than a literal animal costume.
5. Color
- Soft, coherent palettes: cream, oatmeal, muted brown, dusty pink, sage, pale blue, lavender, warm yellow, charcoal, muted terracotta.
- Stronger colors are allowed as accents, not as neon lighting.
- Keep the face slightly warmer and smoother than the body.
6. Rendering
- Premium 3D animation / collectible-plush quality.
- Soft diffused studio light, subtle ambient occlusion, gentle floor shadow.
- Clean white or very light neutral background by default.
- Front-facing or small natural pose variation; avoid dramatic camera distortion.
- No text unless requested.
Variation Rules
When generating multiple characters, make each character structurally different, not just recolored.
Vary across these axes:
- silhouette: tall, squat, pear-shaped, egg-shaped, wide, narrow, top-heavy, bottom-heavy
- fur: short plush, long fluffy, shaggy, boucle, velour, curly wool
- face opening: rounded rectangle, oval, arch, bean, shallow trapezoid, asymmetric soft cutout
- appendages: rounded ears, floppy ears, horns, antennae, fins, tiny wings, tail, sprout, mane
- proportions: face-to-body ratio, limb length, shoulder width, belly size, leg spacing
- motifs: stripes, spots, belly patch, gradient, seasonal detail, tiny scarf, bow, hood nub
- expression: neutral smile, sleepy, curious, shy, content, surprised-but-subtle
Do not create 36 near-identical bodies with different animal ears. Change the underlying body architecture too.
Generation Workflow
- Read the user's requested count, aspect ratio, grid, subject, and variation constraints.
- Use
assets/reference-avatar.pngas the primary design-language reference. - If available, use
assets/character-grid.pngonly as a diversity reference, not as a set to copy. - Build a variation plan before generation so neighboring characters do not repeat silhouette, species cue, color, or accessory.
- Generate with image generation using the reference image(s).
- Inspect the result for: - correct number of characters - unique silhouettes - consistent face architecture - fuzzy-shell / smooth-face contrast - no duplicated characters - no malformed limbs or accessories - no text artifacts - cohesive lighting and scale
- If diversity is weak, regenerate with stronger structural variation rather than adding random accessories.
Default Prompt Block
Use this wording as the base and adapt it to the user's request:
Create original premium 3D animation characters in one coherent collectible family. Each character has a monolithic rounded plush body, a small smooth matte face panel embedded inside the fuzzy shell, tiny simplified limbs, restrained dot-eye facial features, subtle blush, tactile soft materials, and a strong readable silhouette. Keep geometry low-detail, rounded, and huggable. Every character must be structurally unique through silhouette, proportions, fur treatment, face-opening shape, appendages, motifs, and color—not merely recolored. Preserve a consistent design language across the entire set. Use soft diffused studio lighting, subtle contact shadows, and a clean white or very light neutral background. Avoid glossy CGI, literal mascot costumes, over-detailed anatomy, exaggerated emoji expressions, generic kawaii duplication, neon gradients, text, and AI-slop artifacts.
Grid Defaults
For character-sheet requests:
- default to a strict equal-cell grid
- exactly one character per cell
- full body visible
- consistent visual scale
- generous breathing room
- light or invisible separators
- no labels unless requested
For 36 characters, use a strict 6 × 6 grid.
Negative Direction
Avoid:
- generic Funko-like vinyl heads
- chibi human proportions
- literal furry animal suits
- big anime eyes
- oversized mouths
- glossy plastic or clay-only look
- hard-surface mecha details
- random accessories used as the only source of uniqueness
- repeated identical body meshes
- excessive rainbow/neon palettes
- fake text, watermarks, or UI
- over-rendered cinematic backgrounds unless requested
keep it where your ai can reach it.
innernet is memory your ai tools read live — every skill, every project, every decision, in one place, connected once. save this skill to yours, or publish one of your own as a link like this.