Skip to main content

3D Model Generation

3D Model Generation is a Ludus AI feature that allows you to create custom 3D models directly from text descriptions or reference images. After a model is created, it will be placed in your scene as a placeholder, ready for you to position, rotate, and scale.

Generation Modes and Costs​

Ludus AI automatically selects the most appropriate generation mode based on your prompt description. Each mode is optimized for different asset types and comes with a different credit cost:

Mode: Sketch Good for stylized game assets. Less suited for photorealism or complex shapes. Cost: up to 1200 credits

Mode: Default A balanced option that works well for most models, producing clean geometry. Cost: up to 2000 credits

Mode: Detailed Best for high-quality, complex, or photorealistic models with HD textures. Cost: up to 2000 credits

In addition to the automatically selected mode, you can specify the final quality of the model's mesh and textures by requesting "low," "medium," "high," or "ultra" quality. This quality setting does not affect the credit cost.

Credit Consumption and Best Practices​

To manage costs and achieve the best results, follow these best practices:

• Break Down Complex Tasks: Start a new chat whenever possible and assign tasks / models one by one. Avoid asking the AI to create entire complex scenes at once. This will minimize the margin for error and allow you to follow the AI's thought process.

• Choose the mode that best suits your style. The 'Detailed mode' isn't always the best choice for your work. We've noticed that 'sketch mode', for example, handles textured surfaces, reflections, and simulated light much better, but it doesn't preserve a model with as much detail.

Workflow with 3D Generation​

To generate 3D models you need to switch the Ludus model to Ludus Agent directly in the plugin.

3D_model_options

In order to optimize work with 3D generation keep in mind the quality of the prompts, their length and directness. More tips on how to prompt to get satisfying 3D models, see the Prompting for 3D models tab.

Give Ludus a task.

3D_task

Ludus will place the placeholder. It is possible to place multiple placeholders in case of a multi-model generation order. As long as the placeholders are spinning - the model is generated.

3D_placeholder

Generation time may vary depending on the complexity of the object, the number of objects generated simultaneously, and the load on the servers. In the event of unsatisfactory generation, we recommend basing the prompt on a reference to an existing model and simply specifying the desired changes.

3D_model_ready

In order to optimize work with 3D generation keep in mind the quality of the prompts, their length and directness. Ludus is able to generate many objects at once. However, it is harder to create a good prompt for many objects at once so it may be less effective than generating individual elements. For each failed 3D object generation credits are returned to the user's account. The refund may not be immediately visible on the counter if the servers are overloaded.

Roadmap and Future Updates​

We are already working on expanding this feature. In the near future, we plan to add support for more model formats, including:

• Animated models • Models with skeletal rigs