AI Asset Generation
AI Asset Generation
V1 Documentation - Under Heavy Rework. Some sections may be incomplete. For the latest info, check the FAB listing or Discord.
Generate textures, PBR materials, 3D models, and sound effects from text descriptions, all imported directly into your project as usable Unreal assets.
Texture Generation
Generate textures from text prompts. The AI creates the image and imports it as a Texture2D asset.
Supported Providers
| Provider | Setup |
|---|---|
| Google Imagen | Gemini API key in any slot |
| OpenAI DALL-E | OpenAI API key in any slot |
| Together AI | Together API key in Custom provider |
| OpenRouter | OpenRouter API key, pick an image model |
How to Use
In the Architect chat, simply ask:
- “Generate a texture of a mossy stone wall”
- “Create a stylized wooden plank texture”
- “Generate a sci-fi metal panel texture for the floor”
The AI calls the texture generation tool, creates the image, and imports it to your project’s current Content Browser folder.
Configuration
Go to Settings > Image & 3D to configure:
- Image generation provider
- API key
- Model selection
- Default resolution
PBR Material Generation
Generate a complete PBR material from a single text prompt. The AI:
- Generates one base color texture from your description
- Algorithmically derives Normal, Roughness, Metallic, and AO maps from the base color
- Creates a Material with all maps connected
- Creates a Material Instance with tiling support
How to Use
- “Generate a PBR material that looks like rusty corroded metal”
- “Create a PBR brick wall material with mortar”
- “Make a stylized grass PBR material”
PBR generation is the default when you ask for any material with realistic textures.
3D Model Generation
Generate 3D models from text descriptions or reference images, imported as StaticMesh assets.
Supported Providers
| Provider | Capabilities |
|---|---|
| Meshy AI | Text-to-3D, Image-to-3D, mesh retexturing |
| Tripo3D | Text-to-3D, Image-to-3D |
| Hunyuan3D | Text-to-3D |
How to Use
Text-to-3D:
- “Generate a 3D model of a medieval sword”
- “Create a low-poly tree model”
- “Make a 3D treasure chest”
Image-to-3D:
- Attach a reference image and say “Create a 3D model from this image”
Retexture:
- “Retexture this mesh with a wooden appearance” (with a mesh referenced via @)
Generation Flow
- Preview mesh is generated first (faster, lower quality)
- Auto-refinement produces the final high-quality mesh
- The model is downloaded as FBX/GLB and imported as a StaticMesh
Configuration
Go to Settings > Image & 3D to configure:
- 3D model provider (Meshy, Tripo, Hunyuan)
- API key for the selected provider
- Output quality preferences
Sound Generation
Generate sound effects, music, and voice from text descriptions, imported as USoundWave assets.
How to Use
- “Generate a gunshot sound effect in a warehouse”
- “Create ambient forest sounds with birds chirping”
- “Generate a voice saying ‘Welcome to the arena‘“
Configuration
Go to Settings > Voice & Sound > Sound Generation to configure:
- Provider (OpenAI, ElevenLabs, StabilityAI, HuggingFace, Custom)
- API key
- Model and voice selection