MiniMax Hailuo 2.3

Overview

Hailuo 2.3 is an AI video generation model that turns text prompts and images into video clips. It focuses on keeping motion stable over time and making sure your characters and visual style don’t drift between frames.

Compared to earlier Hailuo models, version 2.3 behaves more predictably during motion and does a better job of keeping characters consistent. It’s built for day-to-day video generation rather than simple experimentation.

How it Works

Prompt Interpretation

Text prompts guide what appears in the video, how the scene looks, and how things move. Prompts that clearly describe actions or the scene tend to create more reliable clips. You can specify the visual style directly in the prompt, too.

Image-Based Video Generation

For I2V, your generated video will keep the structure and appearance of your input image while adding motion over time. You can use this to animate existing assets or renders without rebuilding the scene.

Motion and Continuity

Hailuo 2.3 produces more stable motion across frames compared to earlier versions. Generated clips show less flicker and do a better job of keeping subjects consistent as the video plays.

Stylization and Visual Content

The model supports both realistic and stylised output. When you define a style the prompt or input image, it generally holds up across the clip without drifting.

Key Features

• Motion and Continuity Improvements
  Produces smoother motion and more consistent results across frames compared to earlier models.

• Character Consistency
  Maintains character appearance more reliably throughout a clip.

• Stylisation Support
  Handles both realistic and stylized visuals with stable output.

• Text and Logo Handling
  Improved stability when rendering text and logos within generated videos.

Technical Specifications

• Model Name: MiniMax Hailuo 2.3
• Model Type: Text-to-video and image-to-video
• Input: Text prompt with optional input image
• Resolution: 768p or 1080p.
• Duration: 6 or 10 seconds (1366×768, default: 6), 6 seconds (1920×1080).

How to Use

1. Use a text prompt and/or upload a static image.
2. Pick your options, such as duration or resolution.
3. Run the generation and see your creation.
4. Adjust the prompt or input if you need to refine it.

Example prompt:
“A medium shot of a person walking through a quiet city street at night. The camera slowly tracks forward as the subject walks. Soft street lighting reflects on wet pavement. The scene has a cinematic, realistic style with natural motion.”

Documentation

You can find full usage details, parameters, and examples here:
https://runware.ai/docs/providers/minimax#hailuo-23