# 🧜♀️ Serene Dive LoRA
> Transform portraits into ethereal mermaids diving into cosmic depths .
## 🎬 Overview
Model Name: `Serene Dive LoRA`
This LoRA generates cinematic backward dives where a serene human portrait transforms into a glowing mermaid, descending into an emotionally resonant, surreal underwater world — blending dreamlike motion with character-driven visuals.
Upload.............. Generate......... Admire the dive.
## 🛠️ How to Use —
1. Upload a portrait or upper-body image
→ Best results with a calm, centered face (eyes open, looking forward).
→ Half-body shots also work beautifully for full transformation.
2. Use the default prompt (you can keep it as-is or customize):
```text
masterpiece, best quality, cinematic, backward dive into the deep, starting from portrait of serene woman, legs transforming into glowing mermaid tail, camera descending with her, character's inspired underwater world unfolding below — surreal, emotional, vast, sunbeams, caustic light, floating particles, 16fps, slow motion, Unreal Engine 5, volumetric lighting, teal and gold palette
```
✨ Feel free to modify:
- The tail style: “iridescent,” “bioluminescent,” “crystalline”
- The underwater world: “abyssal canyon,” “coral dreamscape,” “starlit ocean”
- The mood: “mysterious,” “peaceful,” “melancholic”
3. Generate & admire the video
→ Watch as your character dives backward, transforms, and reveals a world shaped by her essence.
---
## 💡 Pro Tip: Try Image Captions Too!
> A short caption describing your input image can lead to surprisingly beautiful results!
> Example:
> “Young woman with silver hair, eyes reflecting ocean light, calm expression”
> → The model may subtly adapt lighting, color, or motion to match your description.
You don’t need complex prompts — sometimes, less is more.
---
## 🌌 Key Visual Features
- ✅ Smooth backward dive motion (16 frames @ 16fps)
- ✅ Seamless human-to-mermaid transformation
- ✅ Character-inspired underwater world — emotionally rich, surreal, expansive
- ✅ Cinematic quality: volumetric sunbeams, caustics, floating particles
- ✅ Signature teal & gold palette (customizable)
---
## 🎬 Overview
Model Name: `Serene Dive LoRA`
Base Model: `WAN_2_2_A14B_HIGH_NOISE Img2Video`
Trained By: OleaSTeR (via Tensor.Art — trained on Tensor.Art. )
Training Steps: 80 | Epochs: 10
## 📊 Technical Details
| Parameter | Value |
|------------------|---------------------------|
| Base Model | WAN_2_2_A14B_HIGH_NOISE Img2Video |
| Training Steps | 80 |
| Epochs | 10 |
| Target Frames | 16 |
| FPS | 16 |
| First Frame | Required (portrait/upper body) |
---