Scenario

create

Compose Models

Merge and balance LoRAs to create hybrid AI models and fresh art styles with precise influence control and consistency.

Key benefits

  • Merge up to 5 LoRAs

    Merge up to 5 LoRA models for maximum creative flexibility

  • Multi-character scenes

    Generate multi-character scenes with precision

  • Style + character blending

    Combine character models with style models for unique variations

  • New artistic directions

    Create entirely new artistic styles through strategic blending

How it works

  1. 1

    Create Composed Model

    Go to Models page, click 'New Model', select 'Compose Models'. Choose compatible models of the same type (SDXL or Flux) — mix your own models with platform models as needed.

  2. 2

    Add & Configure Models

    Select your base model (e.g., character model), then add 2–4 additional LoRA models for style influences. You can combine up to 5 models total, though 2–3 typically provide the best balance.

  3. 3

    Test with Base Settings

    Save your composed model and test with a simple prompt. Start with only your main model active (other sliders at 0) to establish the baseline appearance and style.

  4. 4

    Adjust Model Influences

    Use the influence sliders to blend styles and effects. Experiment with different values to find your ideal style balance and artistic direction.

  5. 5

    Refine & Experiment

    Test different value combinations and model pairings. Blend multiple styles at varying influences to create entirely new artistic directions. Save successful combinations for future use.

Use cases

  • Character Style Variations

    Merge character models with artistic style models to explore new visual directions while maintaining character consistency across scenes

  • Multi-Character Compositions

    Combine multiple character models to create complex scenes with consistent art styles and seamless character interactions

  • Brand Style Adaptation

    Apply different stylistic treatments to brand characters for various marketing contexts while preserving core identity