Boba Video
  • Getting Started
    • What is Boba Video?
    • Quickstart
  • Video Generation
    • Models
    • Text to Video
    • Image to Video
    • Subject Reference
    • Extend Video
    • First-Last Frame
    • Boba Mode
  • Image Generation
    • Models
    • LoRAs
    • Upscaling
  • Speech Generation
    • Lip Sync
  • Other
    • Prompt Optimization
    • How Credits Work
Powered by GitBook
On this page
  • LoRA Extensions
  • Available LoRAs & How to Prompt Them
  • How to Use LoRAs with the Base Model
  1. Image Generation

LoRAs

PreviousModelsNextUpscaling

Last updated 10 days ago

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.

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.

LoRA Option