Qwen-"Light and Shadow Personality: Star Poster Generation Model"

Model Introduction

"Light and Shadow Personality" is a multimodal celebrity poster generation model based on deep learning. By integrating visual aesthetic analysis , personalized feature extraction , and artistic style transfer technology, it creates highly realistic and compelling celebrity posters. The model not only reproduces the celebrity's appearance and temperament, but also integrates their public image, work style, and even fan culture to generate visual works with both commercial value and artistic depth.

Core technology highlights

1. Persona Embedding uses NLP to analyze celebrity interviews and social media posts, extracting keywords (such as "rebellious" and "elegant") and converting them into visual elements (color, composition, and lighting). For example, when generating a poster for an actor known for his "broken look," the model automatically emphasizes dark tones and cracked textures.
2. Dynamic Style Fusion uses improved CLIP-Adapter technology to dynamically integrate different identity characteristics of celebrities, such as film and television roles and stage appearances, to avoid style conflicts.
3. The fan-aware loss training data includes analysis of fan secondary creations and sentiment in the comment section, making the generated posters more in line with the group's aesthetic preferences.

The meaning of model naming

“Luminous Persona” literally means “luminous personality”, metaphorically:

• Luminous: A star's public image is a projection of the media and the spotlight;
• Persona: Originated from Jungian psychology, it refers to the "mask" people wear in order to adapt to society, implying that posters not only show the truth but also shape the character.

1. Images can be generated by ComfyUI workflow or sd

2. Choose the KONTEXT official large model for the large model

3. Loader: t5xxl fp8 e4m3fn clip.1

4. Pruning type: fp8 e4m3fn

5. The maximum output size is recommended not to exceed 1080*1920

6. The recommended intensity is around 0.8  

7. Amplification algorithm 4×-UltraShrt high score step 30 sampler euler

8. Reverse thruster: joy2

1. You may not host this model and its derivatives (such as fusion model versions) on websites/applications that plan to earn income or donations. If you need to use this model and its derivatives for commercial purposes (such as generative services, selling images, using images for articles or publications published on short video platforms, etc.)

2. You may not directly sell images generated by this model, you may not mix images from other models and publish them on other platforms, and you may not use images created by this model for traffic generation and teaching on other platforms.

3. You may not use this model to intentionally create or share illegal or harmful content. Please abide by public order and morality and use this model for positive purposes.