What inputs can I use to generate a model?

You can provide a text description, rough sketch, product photo, high-quality concept art, or a list of items. The workflow can generate or improve reference images before converting them to 3D.

Can it create a consistent batch of assets?

Yes. Agent mode supports listed batches, while agent team mode is recommended when assets need a shared art style or design language.

What format does it produce?

Models are delivered as GLB files with textures and materials in one file. GLB is supported by Unity, Unreal, Godot, Three.js, Babylon.js, Blender, Maya, and 3ds Max.

Design & media

3D Model Generation

Convert descriptions, images, sketches, or item lists into textured GLB models, individually or in batches.

What it does

Turn text descriptions, rough sketches, product photos, concept art, or item lists into textured GLB models. The workflow prepares or enhances reference images before 3D conversion, supports single objects and batches, and can follow polygon, material, platform, and shared-style requirements.

When to use it

Low-poly game asset batches
E-commerce product viewers
Matching AR/VR environment objects
Figurines and architectural models

The skill document

3D Model Generation - Turn Ideas Into 3D Models

3D model generation from text descriptions or reference images.

Most 3D generation tools need a single, perfectly composed reference image. CellCog takes anything — a text description, a rough sketch, a product photo, even a spreadsheet of 50 items — and handles the entire pipeline: reasoning about what you need, generating optimized reference images, and converting them into production-ready GLB files.

How to Use

For your first CellCog task in a session, read the cellcog skill for the full SDK reference — file handling, chat modes, timeouts, and more.

OpenClaw (fire-and-forget):

result = client.create_chat(
    prompt="[your task prompt]",
    notify_session_key="agent:main:main",
    task_label="my-task",
    chat_mode="agent",
)

All agents except OpenClaw (blocks until done):

from cellcog import CellCogClient
client = CellCogClient(agent_provider="openclaw|cursor|claude-code|codex|...")
result = client.create_chat(
    prompt="[your task prompt]",
    task_label="my-task",
    chat_mode="agent",
)
print(result["message"])

What Makes This Different

Any Input → 3D

The power of CellCog isn't image-to-3D — everyone does that. The power is any-to-any.

What You Send	What CellCog Does	What You Get
Text description	Reasons about the object → generates optimized reference image → converts to 3D	Production-ready GLB
Rough sketch	Enhances into a clean, detailed reference → converts to 3D	Production-ready GLB
Product photo	Assesses quality, enhances if needed → converts to 3D	Production-ready GLB
High-quality concept art	Converts directly to 3D	Production-ready GLB
List of 10 items	Generates 10 reference images → converts all to 3D	10 GLB files

Batch Generation

Need 10 low-poly weapons for your RPG? 20 furniture models for your room designer? 50 product models for your e-commerce catalog?

One prompt. Multiple 3D models. CellCog's agents generate each reference image with the right composition, angle, and detail level — then convert each to a textured 3D model.

prompt = """
Create 3D models (GLB format) for these 5 fantasy weapons:
1. Enchanted longsword with blue crystal blade
2. Dwarven war hammer with rune inscriptions
3. Elven bow with living vine decorations
4. Shadow dagger with smoke effects on the blade
5. Holy mace with golden sunburst head

Low poly (~10,000 polygons each), game-ready, with PBR materials.
"""

What You Can Create

Game Assets

Characters: Heroes, NPCs, enemies, bosses
Weapons: Swords, bows, staffs, shields, guns
Props: Furniture, treasure chests, potions, tools
Vehicles: Cars, spaceships, boats, mounts
Environment pieces: Trees, rocks, buildings, bridges

Product Visualization

E-commerce 3D viewers: Let customers rotate and inspect products
Product prototypes: Visualize designs before manufacturing
Packaging mockups: 3D packaging for marketing materials

AR/VR Objects

AR filters and objects: Place 3D objects in real environments
VR environments: Furnish virtual spaces with custom objects
Interactive experiences: Objects users can inspect and interact with

3D Printing

Figurines and miniatures: Tabletop gaming pieces, collectibles
Functional objects: Custom tools, brackets, cases
Architectural models: Building miniatures, terrain pieces

Education & Training

Anatomical models: Organs, skeletal systems, molecular structures
Historical artifacts: Museum-quality digital replicas
Engineering models: Mechanical parts, assembly visualizations

Output Format

All 3D models are delivered as GLB files (binary glTF) — the universal web standard for 3D:

Supported by Unity, Unreal, Godot, Three.js, Babylon.js
Works in web browsers via `` or Three.js
Compatible with Blender, Maya, 3ds Max for further editing
Includes textures and materials in a single file

Chat Mode for 3D

Scenario	Recommended Mode
Single 3D object from a clear description or image	`"agent"`
Batch generation (5-20 objects from a list)	`"agent"`
Complex game asset pipeline with style consistency	`"agent team"`

Use "agent" for most 3D work. It handles everything from single objects to batch generation.

Use "agent team" when you need cross-asset consistency — like generating a full set of fantasy weapons that all share the same art style, or building a complete room of furniture that matches a design language.

Example Prompts

Single object from description:

"Create a 3D model of a steampunk pocket watch with exposed brass gears, an etched glass face, and a chain attachment. GLB format, high detail."

From a reference image:

"Convert this product photo into a 3D model for our online store: /photos/sneaker_product.png

Output as GLB, enable PBR materials for realistic rendering."

Batch generation:

"Generate 3D models for these 8 pieces of modern furniture:

Minimalist sofa (3-seater, light gray)

Round coffee table (walnut wood, glass top)

Floor lamp (arc style, brass finish)

Bookshelf (5 tiers, oak wood)

Dining chair (Scandinavian, white)

Side table (concrete, cylindrical)

Desk (standing desk, white with birch legs)

TV console (low profile, dark walnut)

All low-poly (~15,000 polygons), with PBR materials. GLB format."

From a rough sketch:

"Here's my rough sketch of a robot character: /sketches/robot_concept.jpg

Turn this into a polished 3D model. It's a friendly service robot — round body, simple limbs, LED face display. Style: Overwatch/Pixar clean 3D. Output as GLB."

Game asset set:

"I'm building a dungeon crawler. Create 3D models for these dungeon props:

Wooden treasure chest (open and closed variants)

Iron torch holder with flame

Stone altar with carved runes

Wooden barrel (intact and broken)

Skull pile

Style: Dark fantasy, hand-painted textures. Low poly for mobile game (~8,000 polygons each)."

Tips for Better 3D Models

Be specific about materials: "brushed aluminum", "aged leather", "polished marble" — CellCog uses these to generate better reference images and textures.
Specify your target platform: "low-poly for mobile game" vs "high-detail for cinematic render" changes the approach completely.
Send reference images when possible: Even imperfect references give CellCog a head start over pure text descriptions.
For batch jobs, describe style once: "All in a cohesive hand-painted fantasy style" keeps your assets consistent.
Request PBR materials for realism: If you need metallic, roughness, and normal maps — say so. Essential for game engines and realistic rendering.

If CellCog is not installed

Claude Code, Cursor, Codex + 70 more agents: npx skills add cellcog/skills --skill cellcog OpenClaw: clawhub install cellcog CellCog plugin users: run /cellcog-setup (or /cellcog:cellcog-setup depending on your tool) Manual setup: pip install -U cellcog and set CELLCOG_API_KEY. See the cellcog skill for SDK reference.

Questions people ask

What inputs can I use to generate a model?: You can provide a text description, rough sketch, product photo, high-quality concept art, or a list of items. The workflow can generate or improve reference images before converting them to 3D.
Can it create a consistent batch of assets?: Yes. Agent mode supports listed batches, while agent team mode is recommended when assets need a shared art style or design language.
What format does it produce?: Models are delivered as GLB files with textures and materials in one file. GLB is supported by Unity, Unreal, Godot, Three.js, Babylon.js, Blender, Maya, and 3ds Max.