# LoRAs ## LoRA Extensions LoRAs (Low-Rank Adaptation models) are lightweight “style” overlays you can apply on top of any base image model to bias your output toward a particular aesthetic. Choose one from the **LoRA** dropdown to instantly transform the look and feel of your generations.

LoRA Option

*** ### Available LoRAs & How to Prompt Them | **LoRA** | **Description** | **Prompting Tips** | | ----------------------------- | ------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------- | | **None** | No style overlay—pure output from your chosen base model. | N/A | | **2000s Analog Core** | Recaptures the visual vibe of early-2000s digital cameras and MySpace-era filters. |

- Mention “film grain” and “slightly washed colors.”
- Call out analog artifacts: “VHS-style lines, light leaks.”

| | **Amateur Photography** | Simulates shots taken on consumer-grade cameras by casual photographers. |

- Use “soft focus” and “natural lighting.”
- Include scene context: “snapshot of friends at a backyard barbecue, candid style.”

| | **Pixel Art** | Renders images in retro pixel-art style, reminiscent of classic 8-bit and 16-bit games. |

- Specify “32×32 pixel grid” or “pixelated shapes.”
- Call out palette constraints: “limited 8-color palette; blocky shading.”

| | **Cyberpunk Anime** | Applies neon-lit, futuristic anime aesthetics with high-contrast colors and stylized characters. |

- Use “neon glow,” “rain-slick streets,” and “anime character with exaggerated eyes.”
- Call out camera angle: “low angle, dramatic shot.”

| | **Realistic/Aesthetic Poses** | Guides models to render human figures in dynamic, photogenic stances ideal for lifestyle imagery. |

- Describe the pose: “model standing in contrapposto,” “arms crossed, looking over shoulder.”
- Mention “studio lighting” and “clean background.”

| *** ### How to Use LoRAs with the Base Model 1. **Select a Base Model:** Choose the **Hailuo Base** model (LoRAs can only be applied to this model for now). 2. **Choose a LoRA:** Pick a LoRA from the dropdown menu—your preview will reflect the new style. 3. **Craft Your Prompt:** Follow the tips provided for the selected LoRA, weaving in style-specific cues to guide the output. 4. **Generate and Refine:** If the style effect is too subtle or too strong, adjust your prompt or experiment with a different LoRA. > **Note:** LoRAs are currently only compatible with the **Hailuo Base** model. > **Pro Tip:** Keep your prompt structure consistent—Subject → Action → Setting → Style—to ensure the LoRA focuses solely on the aesthetic layer.