Whether you’re designing game characters, environments, props, or UI elements, the 2D Forge gives you a flexible, prompt-driven workflow that balances creative freedom with repeatable results. Use your trained art styles or explore built-in styles to generate high-quality 2D assets at scale - all from a single prompt. It’s a fast, focused way to bring your ideas to life using AI, with powerful tools that let you customize every generation and dial in exactly what you need.
What you can do in 2D Forging
Prompt-based asset generation
Describe what you want to create in natural language using the prompt box — Layer’s AI takes care of the rest. You can even upload an image and turn it into a prompt, or roll the dice for a randomized starting point.
Choose from custom or default styles
Use your own trained styles for style-consistent output, or choose from Layer’s default and exploration styles to get started quickly.
Enhance prompts with AI
Auto-enhance your prompt with richer, more descriptive language using the built-in AI enhancer. Toggle it on, off, or set it to run automatically for short prompts. You can also prompt in any language and translate automatically to English. Ideal for international teams or non-native speakers.
Use forge references
Guide your generation using references like layout guides, sketches, or other images. Different reference types give you more control over positioning, form, or context.
When to use the 2D Forge
2D Forging is ideal when you’ve got a clear creative direction and are ready to start producing production-ready assets or high-quality concept art. It’s best used after your initial exploration phase — when you want more control, consistency, and precision in your outputs. Perfect for generating batches of characters, items, environments, or UI elements that align with a specific art style.
UI Overview
2D Forging Prompt Box
 |
Style Selection Panel Access all of your created styles, defaultstyles and exploration styles. |
 | Style Weight If the style supports Weight Adjustment, you’ll be able to adjust the value. Style weight determines how closely the AI model will follow the art style. |
 |
Prompt Box Here is where you’ll describe the asset you’d like to create. You can also upload an image as a prompt. |
 | Enhance with AI AI will add more details and descriptive wording for detailed generations. Make sure to check for unnecessary details. |
 |
Random Prompt Create a random prompt, great for a starting point to get going. |
 |
Prompt Settings Auto Enhance: You can set the enhancing of the prompt to be. Automatic - Enhance the prompt automatically. On - Always enhance the prompt . Off (Default)
Translate: Prompt in any language with Auto-Translate or specific language. |
 |
Forge References Set what type of forge reference you’d like to use for your 2D generation. Read more about Forge References on this page.
Note: Depending on the style’s base model (flux vs SDXL vs BRIA) some forge references may not be supported by the base model. |
 | Aspect Ration Select the aspect ratio for your 2D generation. Options depends on the capabilities of the base model. |
 |
Transparency Generate your 2D image with transparency - only available on supported base models. |
 |
Batch Size Select how many assets you’d like to generate |
Advanced Settings
Performance | Select the type of performance of your generation.
Note: This feature is only available on supported base models. |
Seed | Seed acts as a unique identifier or a starting point for in an AI model. When you provide the same seed and prompt, the AI model will produce the same image each time. |
Guidance Scale | Adjusts how close you want the model to stick to prompts when looking for a related image. Lower guidance scales makes the model less constrained by your settings, allowing more creativity. |
Denoising Steps | Set the number of denoising steps to use for the image generation. Higher values can improve image quality but may take longer to generate. |
Base Shift | Sets a baseline adjustment for the model, ensuring consistent outputs even with weak inputs. Higher values make the image more accurate and consistent but can lead to repetitive textures. |
Max Shift | Sets the maximum adjustment the model can make, limiting how far it can deviate from its original behavior. Higher values risk straying too much from the intended form. |