🎥 Cine Luma: Cinematic Luminescence
"Mastering the Physics of Dramatic Light."
Cine Luma is a specialized Style LoRA that goes beyond simple color grading. It is trained to emulate the physics of cinematic lighting—capturing the density of air (volumetrics), the bleed of bright light sources (bloom), and the rich separation between warm and cool tones that defines high-end cinema.
Whether you are rendering a gritty neo-noir street, a ethereal fantasy landscape, or a dramatic character portrait, Cine_Luma introduces a tangible atmosphere and deep, filmic contrast to your generations.
🗝️ Key Features
Volumetric Density: Adds a palpable sense of "air" to images, including dust motes, fog, and atmospheric haze that reacts to light sources.
Dynamic Range separation: Enforces a strong separation between light and shadow, preventing "grey/flat" images. It excels at the "Teal & Orange" separation often used in blockbusters.
Subsurface Scattering (SSS): Enhances portraits by simulating how light passes through skin, creating a translucent, lifelike glow rather than a plastic surface.
Film Bloom: Simulates the optical imperfection where bright lights (neon, fire, sun) "bleed" into surrounding darker areas, creating a soft, dreamlike halation.
🧠 Training Methodology
Platform: Trained entirely from scratch on ModelScope.
Base Models: Finetuned Qwen Image 2512 (for semantic lighting understanding).
Dataset: Curated from 100 high-resolution cinematic reference frames, ranging from deep sci-fi reds to clinical teals and golden hour naturals.
Technique: Utilized a "Negative Captioning" strategy where lighting and atmospheric terms were strictly excluded from training data, forcing the model to bake these complex lighting physics directly into the trigger word.
🎯 Use Cases
Cinematic Portraits: Adding dramatic rim lighting and skin texture to characters.
Concept Art: creating "mood pieces" where the atmosphere is as important as the subject.
Neo-Noir / Cyberpunk: Scenes requiring wet surfaces, neon reflections, and deep shadows.
Dark Fantasy: Environments that need heavy fog, god rays, and desaturated, gritty tones.
🚀 How to Use
Trigger Word: cine_luma
Recommended Strength: 0.8 – 1.0
Sampler: Euler a or DPM++ 2M SDE (for softer, filmic edges).
Sample Prompts
1. The "Skin & SSS" Test
cine_luma, extreme close-up side profile of a woman, strong backlighting from a sunset creating a rim light on her face, glowing translucent ears, soft focus background, wind blowing hair, highly detailed pores and skin texture
2. The "Atmospheric Volume" Test
cine_luma, interior of an abandoned cathedral, god rays beaming through dusty broken windows, floating dust particles, dark shadows, a single person standing in the beam of light, atmospheric haze
3. The "Neo-Noir" Reflection Test
cine_luma, wide shot of a futuristic cyberpunk street, raining heavily, wet asphalt reflecting pink and cyan neon signs, a detective in a trench coat walking away from camera, deep shadows, high contrast
Master Negative Prompt
cartoon, anime, 3d render, illustration, drawing, painting, flat lighting, bright ambient lighting, overexposed, washed out, low contrast, plastic skin, smooth skin, low quality, worst quality, jpeg artifacts, blurry, text, watermark, signature, logo
🤝 Credits & Acknowledgements
This model received technical support from ModelScope and was trained using their high-performance computing infrastructure. The training workflow utilized ModelScope's Qwen Image architectures to achieve high-fidelity style transfer.