Skip to main content

Usage Examples of 3D Generations

Ludus AI provides a flexible text-to-3D and image-to-3D generation system integrated directly into Unreal Engine. This guide covers common workflows, from basic object creation to advanced scene-aware generation.

Core Generation Modes​

Standard Generation (Preview First)​

By default, when you ask Ludus to generate an object, it first generates a 2D preview of the model, and then automatically proceeds to generate the full 3D model based on this preview.

How to use: Simply type your prompt (e.g., "Generate a medieval wooden chair").
Workflow: Prompt -> Generating 2D preview and automatically generating 3D model.

Preview Iterations​

You can ask Ludus to generate just the 2D preview of the model you wish to create and adjust it via prompts before proceeding to the actual 3D generation.

How to use: Simply ask Ludus to generate a first preview without generating a 3D model yet. Then ask Ludus to proceed with changes you want to see on a proposed 2D model and ask for a changed 2D model. Then proceed with 3D generation.
Workflow: Prompt -> Generating 2D preview -> Ask for changes and new 2D model -> Generate final 3D model.

Fast Generation (Skip Preview)​

If you need results quickly and trust the AI's interpretation, you can bypass the preview stage explicitly in your prompt. If you write in your prompt to skip Preview (e.g., "skip preview", "no preview"), Ludus will generate the 3D model straight away without waiting for image confirmation.

How to use: Add a specific instruction to your prompt, such as "Generate a sci-fi crate without preview" or "Fast generation of a rock."

Generate Variations​

To ensure you get the perfect asset, you can explore multiple options before finalizing.

Request Variations: You can ask for multiple options upfront to compare them and pick the best one.
Example Prompt: "Generate 3 different 2D concepts for a fantasy sword."
Action: Select the best preview from the proposed set and proceed with 3D generation for that specific version.

Batch Generation (Multiple Objects)​

Ludus can generate multiple distinct objects in a single request, streamlining the population of your scene.

How to use: Specify the quantity and types of objects in your prompt.
Example Prompt: "Generate 5 forest rocks and 3 pine trees."
Result: Ludus will process these requests and generate the assets into your content browser/scene.

Optimizing Results: Modes & Quality Settings​

You can tailor your generation for specific needs—whether you need a quick, low-cost prototype or a high-fidelity asset—by combining Generation Modes (Sketch, Default, Detailed) with Quality Settings (Low, Medium, High, Ultra). You can specify these using natural language in your prompt.

Quality Settings Explained: The quality setting is not just about "looks" but also about geometry complexity and performance.

  • Low/Medium Quality: produces cleaner, simpler geometry (lower poly count), which is better for game performance, especially for background objects or mobile games.
  • High/Ultra Quality: produces complex, high-poly meshes with intricate details. This looks better but is heavier on performance and best suited for hero assets or cinematic renders.

Rapid Prototyping (Low Cost)​

Best for: Quick iterations, background props, or simple shapes where budget is a concern.

  • Combination: Sketch Mode + Low Quality
  • Prompt Example: "Generate a simple wooden crate in sketch mode, low quality."
  • Benefits: Lowest credit cost, lightweight mesh (best for performance), fastest generation time.

High-Fidelity & Animation (High Detail)​

Best for: Hero assets, complex props, and complex objects.

  • Combination: Detailed Mode + Ultra Quality
  • Prompt Example: "Generate a detailed antique throne, ultra quality."
  • Benefits: Highest fidelity geometry, detailed textures.

Image-Based Generation​

Using Reference Images you can influence the generation by providing your own reference material.

1. Upload: Click the Upload Image button in the Ludus interface to attach a file.
2. Prompt: Describe what you want to generate based on the image.

Extracting Specific Objects from Photos​

You don't need a clean, isolated image of an object to generate it. Ludus can identify and extract specific elements from a complex photo.

Workflow:
1. Upload a photo containing the object (e.g., a photo of a fully furnished room).
2. Prompt: "Cut out the vintage lamp from this photo." (Using "cut out" yields the best results).
3. Result: Ludus analyzes the image, isolates the requested object (the lamp), and generates a 3D model of it.
---

Scene & Style Context Workflows​

Scene Rearrangement & Stylization​

Ludus can help you visualize changes to your entire scene before you generate individual assets. This workflow allows for high-level creative direction.

1. Capture & Restyle: Ludus captures a screenshot of your current viewport.
2. Prompt for Style: Ask Ludus to re-imagine the scene (e.g., "Show me this scene in a cyberpunk style" or "Rearrange this room to look like a messy office").
3. Visual Feedback: Ludus generates a 2D "paint-over" or concept image of your scene with the requested changes.
4. Generate from Concept: You can then ask Ludus to generate actual 3D models of the new objects visible in that stylized preview.
Example: "Cut out the neon vending machine from the stylized preview."

Context-Aware Style Matching​

Ensure consistency in your level by generating new assets that match existing ones.

How to use: Indicate a target object in your scene (by name) and ask for a matching asset.
Example Prompt: "Generate a coffee table that matches the style of the sofa selected in the scene."
Mechanism: Ludus analyzes the visual features of the reference object and synthesizes a new model that shares similar aesthetic characteristics (texture, shape language, era).