🎥 next-scene-qwen-image-lora-2509

🎉 ✨ UPDATE - Version 2 Now Available! (21 Oct 2025) ✨ 🎉

🚀 New Model: next-scene_lora-v2-3000.safetensors

What's New in V2:

🎯 Trained on higher quality data for significantly improved results
💪 Better command responsiveness - the model follows your prompts more accurately
🖼️ Fixed black bar artifacts - no more unwanted black borders on generated images
⚡ Overall enhanced performance - smoother transitions and better cinematic flow

Recommended: Use V2 for all new projects.

📥 ComfyUI Workflow: workflow-comfyui-basic-next-scene-v2.json

V2 Demo Examples:

next-scene-qwen-image-lora-2509 is a LoRA adapter fine-tuned on Qwen-Image-Edit (build 2509), purpose-built to generate cinematic image sequences with natural visual progression from frame to frame.

This model enables Qwen Image Edit to think like a film director—understanding camera dynamics, visual composition, and narrative continuity to create shots that flow seamlessly into one another.

📦 Version 1 (Legacy)

Model File: next-scene_lora_v1-3000.safetensors
ComfyUI Workflow: workflow-comfyui-basic-next-scene.json

V1 Demo Examples:

🧠 What This Model Does

This LoRA brings cinematic storytelling continuity into AI image generation workflows.

Each output frame functions as the "Next Scene" in an evolving visual narrative, maintaining compositional coherence while introducing organic transitions such as:

Camera movement: Dolly shots, push-ins, pull-backs, and tracking moves
Framing evolution: Wide to close-up transitions, angle shifts, reframing
Environmental reveals: New characters entering frame, expanded scenery, spatial progression
Atmospheric shifts: Lighting changes, weather evolution, time-of-day transitions

Examples of Cinematic Logic:

"Next Scene: The camera pulls back from a tight close-up on the airship to a sweeping aerial view, revealing an entire fleet of vessels soaring through a fantasy landscape."
"Next Scene: The camera tracks forward and tilts down, bringing the sun and helicopters closer into frame as a strong lens flare intensifies."
"Next Scene: The camera pans right, removing the dragon and rider from view while revealing more of the floating mountain range in the distance."

⚙️ Usage Instructions

Basic Setup:

Load Qwen-Image-Edit 2509 as your base model
Add a LoRA Loader node and select:
- V2 (Recommended): next-scene_lora-v2-3000.safetensors
- V1 (Legacy): next-scene_lora_v1-3000.safetensors
Set LoRA strength: 0.7 – 0.8 (recommended)
Structure your prompts with "Next Scene:" prefix for optimal results

Example Prompt:

Next Scene: The camera moves slightly forward as sunlight breaks through the clouds, casting a soft glow around the character's silhouette in the mist. Realistic cinematic style, atmospheric depth.

Pro Tips:

Begin prompts with camera direction for stronger continuity
Specify lighting and atmospheric changes for mood consistency
Chain multiple generations to create sequential storyboards
Works particularly well with landscape and establishing shots

🎬 Design Philosophy

Trained on an extensive, curated cinematic dataset (proprietary), this model has learned to think directionally rather than just visually.

It doesn't simply modify an image—it advances the story, preserving spatial relationships, lighting consistency, and emotional resonance across sequential frames.

Ideal Applications:

Storyboard generation for film and animation pre-production
Cinematic AI video pipelines requiring frame-to-frame coherence
Sequential narrative workflows in ComfyUI and similar tools
Concept art evolution showing scene progression
Visual storytelling for creative projects and presentations

⚠️ Important Limitations

Not optimized for: Static portraits, single-image illustration tasks, or non-sequential edits
Best suited for: Multi-frame workflows with narrative progression
Design priority: Storytelling flow and continuity over isolated image perfection
Recommended use case: Scene-to-scene transitions rather than detailed object manipulation

🧱 Technical Specifications

Base Model: Qwen-Image-Edit (build 2509)
Architecture: Low-Rank Adaptation (LoRA)
Training Objective: Scene continuity and cinematic shot coherence
Dataset: Large-scale proprietary cinematic imagery
Recommended Strength: 0.7–0.8
Compatible Platforms: ComfyUI, Automatic1111 (with Qwen support), custom pipelines

📄 License

MIT License — Free for research, educational, and creative use.

Commercial applications require independent testing and proper attribution. See LICENSE file for full terms.

🌐 Creator

Developed by @lovis93

Pushing the boundaries of AI-directed visual storytelling and cinematic image generation.

🐦 Share This Model

🎥 Introducing next-scene-qwen-image-lora-2509

A LoRA fine-tuned for Qwen-Image-Edit 2509 that thinks like a film director.

It evolves each frame naturally—new angles, new lighting, same coherent world.

Perfect for cinematic storyboards, sequential edits, and "Next Scene" workflows.

👉 https://huggingface.co/lovis93/next-scene-qwen-image-lora-2509

#AIart #ComfyUI #Qwen #LoRA #GenerativeAI #AIcinema #ImageEditing