Hey folks!
This is a complete, single-file solution (HTML+CSS+JS) that acts as a dedicated Prompt Builder & Educational Guide for HunyuanVideo 1.5. It enforces the official "5-Step Structure" recommended by the developers and helps you format natural language prompts correctly.
How to use:
Download the HTML file and simply open it in your browser.
HunyuanVideo 1.5 Capabilities & Prompting Guide
1. The 5-Step Structure:
Unlike older models that prefer "tag soup" (comma, separated, lists), Hunyuan 1.5 performs best with Natural Language Sentences. This tool assembles your inputs into this specific flow:
Subject: Who is it? (Appearance + Texture)
Action: What are they doing? (Active verbs are crucial)
Surroundings: Where are they?
Camera: How is it filmed? (Angles + Movement)
Style: The aesthetic quality (8k, Cinematic, etc.)
2. Mode Logic:
Text-to-Video: Requires a full descriptive paragraph.
Image-to-Video: Critical Difference: Unlike other models, for Hunyuan you should briefly describe the Reference Image content first to ground the model, followed by the specific Motion you want to see.
3. Technical Settings Reference:
The tool includes a built-in settings calculator for the different model versions:
Speed (Default): 20 Steps | CFG 6.0 | Shift 9 (720p)
High Quality (Official): 50 Steps | CFG 6.0 | Shift 9 (720p)
Distilled Models: 50 Steps | CFG 1.0 (Crucial!)
Lightx2v LoRA: 4 Steps (CFG 1.0) or 8 Steps (CFG 0.5)
4. Camera Language:
The tool visualizes complex camera terms for you:
Tracking/Dolly: Moving with the subject.
Static/Rack Focus: Good for subtle portraits.
FPV/Drone: High velocity, immersive angles.
Info source and prompt examples: https://github.com/Tencent-Hunyuan/HunyuanVideo-1.5/blob/main/assets/HunyuanVideo_1_5_Prompt_Handbook_EN.md
Good luck!